PDA

View Full Version : Abysmal OGL



Andyjaggy
06-14-2007, 10:31 AM
Am I the only one that find the OGL performance in modeler absolutely pathetic? Yes it's great we can tumble large object now but try moving that object across the screen and see what happens. Drag. Wait 15 seconds for update. Drag. Wait 15 seconds for update. etc.......

I am working on a project where I am building an entire city and even with splitting everything up into small little chunks as soon as I hit 100k or 150k the model becomes unworkable. To me this should be the #1 priority with the next update. What good are new tools if you can't use them because the OGL is so slow. Grrrrrrrrrrrrrrrr.

Andyjaggy
06-14-2007, 11:34 AM
It better or I might start looking elsewhere for most of my modeling.

Matt
06-14-2007, 11:52 AM
Yes, same here on my 8800GTX 768MB GFX card, another thing (that seems to be an issue with this card, and not the one at work) is when switching apps, it takes an AGE for LW / LWM to refesh, talking a good 4-6 secs.

Andyjaggy
06-14-2007, 12:20 PM
I have a 7950 and it is slow slow slow. I think you could have the best card on the market and it wouldn't make much of a difference with Lightwave.

sean hargreaves
06-14-2007, 01:09 PM
This is why programmers need a damn good thrashing. They program something and ALWAYS forget that it leads to a veritable TREE of other issues that need to be solved and answered. Its like inventing a car with wheels that can't be removed or a plane where the door does'nt open. Great!

G3D
06-14-2007, 01:34 PM
OGL was faster in some iterations of the beta release. Same with overall stability. This is true for me on two separate PCs. Go figure. :shrug:

Andyjaggy
06-14-2007, 02:03 PM
It might have been a little bit better with some of the betas but still wasn't anywhere near where it should be. I was watching a guy throw around a 1 million poly model in Max the other day. It made me jelous.

Andyjaggy
06-14-2007, 03:00 PM
I have been thinking the same thing, I've been expecting 9.3 to fix this issue. If it doesn't I will be very dissapointed.

sean hargreaves
06-14-2007, 04:11 PM
That cat, is enormous!

Dodgy
06-14-2007, 04:19 PM
As far as I understand it, max ONLY JUST had a rewrite of it's poly core. A friend of mine uses max and he says he was shown current max and the previous version and the previous version was dog slow in comparison. Hopefully we'll get a similar boost in this area soon, though I imagine it'll mean re writing all the tools to make use of it.

Lightwolf
06-14-2007, 04:23 PM
That cat, is enormous!
Please go away if you have nothing productive to add to this thread.

Sheeesh... have you seen the size of that cat? :D

Erm, em, errr... how many polys is it?

Cheers,
Mike - ;) <- please note!

P.S. Max seems to be very decent in that respect (it also skips displaying parts of the mseh to keep up the framerate). I've seen people work on 5 million polygon objects easily (CAD based cars).

Andyjaggy
06-14-2007, 04:58 PM
It's the Utah kitty of death.

Life would be good if I could edit 5 million poly objects. Please Newtek, make it so.

sean hargreaves
06-14-2007, 05:30 PM
Sorry Andy about Lightwolfs last comment, he's just annoyed that his modelling of a cat ended up looking like a moose.

Andyjaggy
06-14-2007, 10:21 PM
I think he was joking :)

Or if we can't get decent OGL give us geometry instancing. Please. I am trying not to get too upset because I am hoping there is going to be some love for modeler with the next point release, but grrrrrrr.

ShawnStovall
06-14-2007, 11:02 PM
I'm having trouble on even some of the simplest meshes I produce(and believe me, when I say simple I mean SIMPLE). I know that they are doing work on the OpenGL, but please, dedicate some more time to it.:hat:

Chris S. (Fez)
06-14-2007, 11:31 PM
The Max developers blog mentions that their main goal for Max 10 is performance and that viewports are much, much, faster than in 9.

I hope to see similar gains at some point in Layout and Modeler in Lightwave 9.x.

Lightwolf
06-15-2007, 02:09 AM
Sorry Andy about Lightwolfs last comment, he's just annoyed that his modelling of a cat ended up looking like a moose.
:ohmy: you just made my day :D

Cheers,
Mike

zardoz
06-15-2007, 05:15 AM
I work with 10 more guys and they all work with max...I see them moving millions of polygons around and I can't do the same in Lightwave.
They do archi stuff and they always have hundreds of trees and other stuff...if I import only one of their houses, Lightwave is so slow...why is it this way? all the computers there are the equal.

Matt
06-15-2007, 06:39 AM
ILightwave is so slow...why is it this way?

Because Autodesk have a ton more money and coders to throw at it! Remember NewTek are a fraction of their size (only 10 coders!)

We'll get there, just look at what they've done so far! People said it would take a TOTAL rewrite of OpenGL code to be able to improve the speed of anything!

I have faith in them! (More now than ever before!)

Andyjaggy
06-15-2007, 07:11 AM
But at the same time I think that tends to make Max modelers very sloppy. I was working with a mesh modeled in Max and it was so sloppy and had so many more polys then it needed to. I guess I am a little old school in thinking that you should keep your meshes clean and your poly count low. I wonder if that method of thinking is getting out of date.

I think good modelers will always think that way. The computer power may not require that it be so but it still makes it so much easier to work with the mesh, to edit the mesh, and to texture the mesh, and even now any seconds you can save off a render is worth it.

JGary
06-15-2007, 09:18 AM
So, is Max the only 3d app that has the upper hand over Modeler when it comes to working with complex models? My latest project has me importing large IGES files from ProE and working with some of these objects in LW 9.2 has been a true pain. One object was close to 1 million triangles and really brought modeler to a crawl, so I moved over to Maya 7 to see if there was a improvement. Honestly, Maya seemed just as slow at manipulating the same object. Whenever I would right click to get to the menu to select faces mode, I would have to wait about 30 seconds for any response from Maya. Maybe there's a big improvement in Maya 8.5.

My desktop is a AMD dual core 3800+ with a Geforce 6800 GS card and 3gb of ram. I also have a laptop with a Intel dual core T7200, a Geforce 7950 GTX and 2gb or ram. Not really much of a difference with either of these machines.

MSherak
06-15-2007, 09:48 AM
Hmm,

Maya is a lot slower than LW when it comes to GL.. Maya can't do 1/2 the things you can can in LW's GL.. About the only thing that Maya has over LW is the ability to view normal maps in GL.. Course you have to use a Nvidia shader or write your own and use the Hardware render which slows it to a crawl..
Try doing particles in Maya then LW.. For drawing functionality LW blows Maya out of the water.. Heck even LW's particles can be played backwards unlike Maya..

JGary
06-15-2007, 09:58 AM
Hmm,

Maya is a lot slower than LW when it comes to GL.. Maya can't do 1/2 the things you can can in LW's GL.. About the only thing that Maya has over LW is the ability to view normal maps in GL.. Course you have to use a Nvidia shader or write your own and use the Hardware render which slows it to a crawl..
Try doing particles in Maya then LW.. For drawing functionality LW blows Maya out of the water.. Heck even LW's particles can be played backwards unlike Maya..

Fortunately, most of my work experience with Maya concerns lower-rez game assets, so I don't have to use it with dense geometry often. It seems to me that modeler's abitlity to handle complex meshes has gotten worse over the past couple of versions.

-EsHrA-
06-15-2007, 10:07 AM
its dog slow here too, modeling on a big object is no fun.
im always hidding as much as i can.
we need more resources focussing on this subject.

mlon

calilifestyle
06-15-2007, 10:15 AM
you have alot of control in max also u can easliy drop a modifer to lower poly count and it does a good job of it u can even pick which areas you want to lower the count my last version was 5 so not sure what they have add since. imean u can work with low ploy then drop a mesh smothmodifer and see what you will get at high count, supress the modifer and go back to lowpoly.

Chris S. (Fez)
06-15-2007, 10:22 AM
I have not used Maya 8 and up, which supposedly received some performance inprovements, but Maya SUBD viewport performance was (is?) terrible.

Lightwave OGL is already tolerably fast. XSI and Max developers dedicated a lot of time and energy to optimizing subds and viewport performance and it is paying off bigtime.

Newtek has already acknowledged that they are replacing Modeler's core, so I think we can safely expect greatness :thumbsup: .

Andyjaggy
06-15-2007, 10:26 AM
Mmmmmm greatness. I didn't realize Maya was so bad with OGL. I am currently writing this message as modeler thinks for about 30 seconds about repositioning my object. Fun fun.

MSherak
06-15-2007, 03:06 PM
Mmmmmm greatness. I didn't realize Maya was so bad with OGL. I am currently writing this message as modeler thinks for about 30 seconds about repositioning my object. Fun fun.


Which mode are you using in modeler?? and what options do you have turned on??

Andyjaggy
06-15-2007, 03:17 PM
Multitextured. Buffered. With most of the eye candy turned off. No transperency etc....

-EsHrA-
06-16-2007, 04:16 AM
smoothshading multitexture with most options off, glsl is a no go.


mlon

darkChief
06-16-2007, 04:32 AM
Opengl is getting some big changes soon (the new version should be out this summer). There's currently two versions being worked on, one for current gen hardware and one for next gen. I'm guessing newtek are probably waiting for that to come out because a big focus of the new Opengl's is performance.
Just my thoughts.

AbnRanger
06-16-2007, 05:30 AM
Yeah, the difference between Max 8 and 9 (performance wise) is very dramatic. Instead of droping another big plugin/feature or two, as they had been doing, they pretty much focused on improving overall performance.

I'm sure with XSI's and Modo's impressive Sub'D handling, they saw "the writing on the wall." Let's hope Newtek does too.

What's ironic is that Modeler is a "Lighter" application since it doesn't carry the overhead of animation, particles, rendering engine...so, logically, it SHOULD be the fastest modeler among the major applications.

Chris S. (Fez)
06-16-2007, 08:48 AM
I hope we will see Modeler's new core some time before Siggraph.

vadermanchild
06-16-2007, 01:48 PM
Do Newtek have an opinion on this? Im confused - while there was a speed increase over 9.0 its not the speed increase that was I felt was implied.

I also saw the release info on the new 9.2 update and the bugs that have been quashed looked quite thin and mostly unimportant considering some of the big bugs mentioned here and elsewhere.

THe 9.2 OGL is very poor - it would be nice to hear from Newtek that everyones concerns are being delt with. Is it a priority task? or is there yet another pointless wax look shader in the making?

Newtek needs to demonstrate to us they know where their priorities are.

Modeler used to be a strong point of LW - Its been one of the reasons its pushed me and most of the other people I know doing modelling work away from modeller - Im a big fan of LW but you got to do whats best for your project or employment prospects.

At present its not lw modeler you choose.

Lightwolf
06-16-2007, 05:38 PM
Do Newtek have an opinion on this? Im confused - while there was a speed increase over 9.0 its not the speed increase that was I felt was implied.
Well, it seems the speed ups in both apps only occur with static meshes. Obviously, a mesh in Modeler that is being changed is not static.
I suppose speeding up danymic or deforming meshes needs a lot more changes under the hood, probably adapting base functionality to perfom tasks in a way that is friendlier to openGL... thus the delay (no pun intended).

Cheers,
Mike

P.S. Love your avatar, love the album!

hrgiger
06-17-2007, 06:27 AM
P.S. Love your avatar, love the album!

Which album is that? I haven't seen that one, I thought I had most of his. Is it newer? He was just here a few weeks back in Columbus but I had to choose between him and George Carlin. I had to go with George.

I thought that translation speed was one of the things that Newtek was supposed to improve with 9.2. Hopefully it will be addressed in 9.3 or whatever the next incantation is. XSI's gigapolygon core sure is swell.

Lightwolf
06-17-2007, 07:25 AM
Which album is that? I haven't seen that one, I thought I had most of his.
:hijack: You Are the Quarry - his second last - released two years ago if I remember correctly. Ringleader of the Tormentors is his current one.

Cheers,
Mike - who's got nothing to say about OpenGL for a change...

alifx
06-17-2007, 08:50 AM
Because Autodesk have a ton more money and coders to throw at it! Remember NewTek are a fraction of their size (only 10 coders!)

I was going to say the same thing as matt said.

when we talk about Autodesk we talk about 3 of the biggest companies in the 3d world, Alias, Discrete, and AD itself.

if we have 10 coders working on it, they have "100", if NT have 1 Million $ they have 10 Millions $.

and that's the problem, maybe if they let lightwave to be Opensource it would solve all the problems that we have <<< lightwender hahaa :P

Exception
06-18-2007, 03:28 AM
I doubt it is the OpenGL code that is causing modeler to be slow. Tumbling is a purely OpenGL operation (which is fast) but dragging a point is a mesh editing operation (which is slow) that is then afterwards reflected by OpenGL. It doesn't mean OGL is poor, it means the mesh editing routines are slow. This is being worked on as modeler receives a major overhaul. It's already been stated by Jay Roth that 9.2 focused its attention on Layout, and I think most of us are very happy with what they did there, and that the next major 9.x release will focus on modeler and CA. This will include the mesh editing engine, which, by the way, has been decoupled from modeler to allow layout to access it and do modeling operations. With all these complicating tasks, it cannot be easy. It's unfortunate we don't have the speed now, but we should have it in the forseeable future.

colkai
06-18-2007, 04:22 AM
I have found, the fastest way (if you'll pardon the pun), to slow down modeller is to drag / move points when I have an active background layer. It is painful!

Ok, I'll grant, I'm not running a high end system, but with the details in the foreground and using something like C_Worm, I have no problems. Editing a spline with the same geometry will bring my system to its knees.

Captain Obvious
06-18-2007, 05:40 AM
I just loaded up a 1,100,000 polygon archi viz object in modo, and it's reasonably smooth to tumble. Editing all those polygons is a bit slow, though...

AbnRanger
06-18-2007, 08:22 AM
I was going to say the same thing as matt said.

when we talk about Autodesk we talk about 3 of the biggest companies in the 3d world, Alias, Discrete, and AD itself.

if we have 10 coders working on it, they have "100", if NT have 1 Million $ they have 10 Millions $.While that may be the case, I doubt XSI's staff is significantly larger than Newtek's. And then look at Modo/Luxology...even smaller, yet look at the phenominal output. I have a good deal of confidence that we'll see a very modern LW, from top to bottom, by v10-10.5

Personally, I would expect Newtek to unveil LW 10 at Siggaph this year, even though we are currently at 9.2
Three reasons:
1) It's been 2yrs since LW 9 was shown at Siggraph '05 and it wasn't released til almost a year later.
2) At that time, we were at LW 8.3, and 8.5 was subsequently came around December/January
3) Market-wise, I think it would hurt them to let another Siggraph go by without being able to announce an upcoming full upgrade, like their competitors.

wavk
06-18-2007, 09:13 AM
While that may be the case, I doubt XSI's staff is significantly larger than Newtek's. And then look at Modo/Luxology...even smaller, yet look at the phenominal output. I have a good deal of confidence that we'll see a very modern LW, from top to bottom, by v10-10.5


i dont, although i really hope so.

too many promises not fullfilled.

now its the time for nt to prove me and allot of other users wrong.


mlon

beverins
06-19-2007, 09:18 PM
If they were to demo an alpha of LW 10 at Siggraph I would be very happy indeed. Unfortunately, the only things I can hope for are simple things - I hope you guys are cranking out lots and lots of demo discs... I hope you get Andy Bishop this year... please tune your presentations to focus on your audience (I'd better not be hearing "are you scared of 3D" from a presenter!!... etc.

Oh, right, we're talking about the bad OGL performance in modeler. I find using weightmap mode for pretty much everything (save texturing, obviously)speeds things up for me a bit.

jin choung
06-19-2007, 09:34 PM
While that may be the case, I doubt XSI's staff is significantly larger than Newtek's. And then look at Modo/Luxology...even smaller, yet look at the phenominal output. I have a good deal of confidence that we'll see a very modern LW, from top to bottom, by v10-10.5

Personally, I would expect Newtek to unveil LW 10 at Siggaph this year, even though we are currently at 9.2
Three reasons:
1) It's been 2yrs since LW 9 was shown at Siggraph '05 and it wasn't released til almost a year later.
2) At that time, we were at LW 8.3, and 8.5 was subsequently came around December/January
3) Market-wise, I think it would hurt them to let another Siggraph go by without being able to announce an upcoming full upgrade, like their competitors.

actually, i can really do without frequent full point updates. one of the reasons i like lw and stick with it is because of low cost of ownership.

if newtek were to up their paid upgrade schedule to something like maya (ridiculous), i would be one unhappy customer.

and some of maya's last few paid upgrades were pretty fing anemic and i would have been superfingpissed if i were a direct customer.

every few years is just fine by me.

jin

Andyjaggy
06-19-2007, 09:41 PM
I would be very surprised to see LW 10 announced at Sigraph. I am just hoping that we will be on our way towards our next point release by that time.

alifx
06-19-2007, 10:36 PM
While that may be the case, I doubt XSI's staff is significantly larger than Newtek's. And then look at Modo/Luxology...even smaller, yet look at the phenominal output. I have a good deal of confidence that we'll see a very modern LW, from top to bottom, by v10-10.5

Personally, I would expect Newtek to unveil LW 10 at Siggaph this year, even though we are currently at 9.2
Three reasons:
1) It's been 2yrs since LW 9 was shown at Siggraph '05 and it wasn't released til almost a year later.
2) At that time, we were at LW 8.3, and 8.5 was subsequently came around December/January
3) Market-wise, I think it would hurt them to let another Siggraph go by without being able to announce an upcoming full upgrade, like their competitors.


I would like to see at siggraph what Newtek will do or what they did for the 9.x of lightwave more than to see how is LW10 will be..

and I would like to see also something about the interactivity of tools in modeler, and anything about new modeling features.

I hope also that one day newtek adds python to LW.

AbnRanger
06-19-2007, 11:58 PM
actually, i can really do without frequent full point updates. one of the reasons i like lw and stick with it is because of low cost of ownership.

if newtek were to up their paid upgrade schedule to something like maya (ridiculous), i would be one unhappy customer.

and some of maya's last few paid upgrades were pretty fing anemic and i would have been superfingpissed if i were a direct customer.

every few years is just fine by me.

jinI can't disagree with you there, but first of all; them unveiling LW10 would be simply letting us take a sneak peek at what they have planned and are working on. Obviously they were still working on 8.5, when they previewed 9 at Siggraph...which means they were working in parallel (on both 8.5 and 9 simultaneously). A word that I remember seeing in one of Jay Roth's letters, regarding LW's future.

Secondly, the longer they go without a full (paid) upgrade, the more it strains their overall budget. Sure we'd all like them to keep giving us point updates that are comparable to full upgrades with other programs, but I realize that these guys have families too, and need to put food on the table. I say two years without a full upgrade is more than enough time.
You wouldn't want "Little Timmy" going hungry, now would you? :D

jin choung
06-20-2007, 01:44 AM
i'm certainly not advocating 5+ years between upgrades. timmy does need to eat.

BUT,

let him eat at mcdonald's! kidding... that's a deathwish. subway. alright, subway.

but i would indeed be a very satisfied newtek-head if they kept right on doin' what they've been doin' in terms of schedule.

jin

p.s. yeah, the open gl thing/mesh operation thing. also, BLACKBOX it so that we can swap in DX etc.... not for the sake of DX but simply so that it is THAT blackboxed so that it can be upgraded without a whole bunch of dependencies which slows down iterations.

also, since we got such rockin' cards now, howabout we revive RENDER FING GL?!?!?!

don't make me come back there!

sammael
06-20-2007, 02:07 AM
As far as I understand it, max ONLY JUST had a rewrite of it's poly core. A friend of mine uses max and he says he was shown current max and the previous version and the previous version was dog slow in comparison. Hopefully we'll get a similar boost in this area soon, though I imagine it'll mean re writing all the tools to make use of it.

Same, my mate uses max and he can easily work with objects 2mil polys+ on an old machine.

zardoz
06-20-2007, 02:27 AM
this last week the company where I work had a huge project...it had 5000 trees with several thousand polygons each, plus the buildings, thousands of cars, people, trains, trainstations, railways, etc...it was really huge, and they used vray meshes to do this in max. If I had to do this in Lightwave and render the amount of frames (with GI) they did in 2 days (the project took a week to complete, 2 days to render) I would simply say that I couldn't do it. The way each one of them manipulated this huge scene in max...I'm pretty sure I couldn't do it in lightwave and we all have the same hardware configuration. A lot of work has to be done in this area...opengl and memoty handling for huge projects.

zardoz
06-20-2007, 03:02 AM
to be quite honest I don't care about the version number...every year I see max change version and the changes in the software are minimal...so to me Lightwave could even go back a few versions...if they are better than the previous release.

Yog
06-20-2007, 06:55 AM
So, is Max the only 3d app that has the upper hand over Modeler when it comes to working with complex models?

I'm afraid both XSI and Modo knock spots off Modeler for VP interaction on large models.

On the flip side, whilst MAX can interact well with very large models, it is a real memory hog when it comes to rendering. There have been a few scenes lately that I have brought into Lightwave because MAX just couldn't render the high amount of detail.

Andyjaggy
06-20-2007, 07:00 AM
I would like to see at siggraph what Newtek will do or what they did for the 9.x of lightwave more than to see how is LW10 will be..

and I would like to see also something about the interactivity of tools in modeler, and anything about new modeling features.

I hope also that one day newtek adds python to LW.

New modeling features.............. screw that i just want everything we have allready to work correctly. Interactivity would be nice. If all they did was add interactivity, faster and more efficient geometry handling, and fix CC's I would be pretty happy.

Exception
06-20-2007, 08:30 AM
New modeling features.............. screw that i just want everything we have allready to work correctly. Interactivity would be nice. If all they did was add interactivity, faster and more efficient geometry handling, and fix CC's I would be pretty happy.

:agree:

So terribly true.
Some small smart workflow enhancements, and a truckload of performance increase: one smart move. Modeler has proven power the way it is now. Except that it just becomes a snail for medium meshes, alla 60k polygons and more. Millions? Not an option.

All those weird hickups need to be smoothed out. How often doesn't modeler just 'stall' for no reason at all?

-EsHrA-
06-20-2007, 10:16 AM
i dont use any lw native im/exporter coz they all s * c k !

:)

mlon

mattclary
06-20-2007, 01:26 PM
And then look at Modo/Luxology...even smaller, yet look at the phenominal output.



Can you actually animate anything in modo yet? :ohmy:

What's it been now, 4 years? 5?

jin choung
06-20-2007, 01:46 PM
yah but modeler is slow and it doesn't animate either. oh... nevermind... you're commenting on the comment about output...

yah, modo has a more limited scope so they can concentrate on a smaller target. agree.


jin

vadermanchild
06-20-2007, 01:56 PM
Lightwaves animation is ok for low end work.

Lws animation is universally derided. I dont expect to see LW animation to come close to other packages for a number of years.

claudioarchi
06-20-2007, 03:18 PM
i completely agree with you guys, It's a painfull experience. I have a quad core and nvidia quatro but still very slow. I was playing with the Ivy generator (http://graphics.uni-konstanz.de/~luft/ivy_generator/) and as soon as i send the result to modeler it takes ages to redraw. It is strange but in my laptop VISTA ULTIMATE OGl is faster!!!!!

Hope next upgrade of modeler will do as good as 9.2 is to layout

Please NEWTEK!!!!!

jat
06-20-2007, 03:21 PM
It's like working in molasses.......pathetic!!!!!

Stooch
06-20-2007, 08:16 PM
I would be very surprised to see LW 10 announced at Sigraph. I am just hoping that we will be on our way towards our next point release by that time.

hah i remember walking through siggraph last year and they were demoing the features that still havent gone gold with the promise that they will be "available shortly after siggraph"

only now they are finishing them up... just in time for the next siggraph yeah. confidence inspiring indeed.

Andyjaggy
06-20-2007, 08:53 PM
Okay I am depressed now. I'm not going to participate in this thread anymore.

Mr Big
06-20-2007, 09:07 PM
Do you get the same performance bottleneck in wireframe mode?

zapper1998
06-20-2007, 09:50 PM
It's like working in molasses.......pathetic!!!!!

I agree 150%

Really Slow OGL bummer.....

zardoz
06-21-2007, 03:37 AM
On the flip side, whilst MAX can interact well with very large models, it is a real memory hog when it comes to rendering.

if you use vray and vray meshes (it's like instancing I guess) that doesn't happen...

in the project we did last week we used thousands of trees, cars, people...it was about 25 million polygons (it was more, but at least 25) and we had no memory problems...I'm the only one there using lightwave and I'm pretty sure I couldn't do this project in the same time.

vadermanchild
06-21-2007, 05:47 AM
Do Newtek read these threads? I hope they do.

wavk
06-21-2007, 06:27 AM
we didnt expect, we were promised.


mlon

Dodgy
06-21-2007, 07:08 AM
if you use vray and vray meshes (it's like instancing I guess) that doesn't happen...

in the project we did last week we used thousands of trees, cars, people...it was about 25 million polygons (it was more, but at least 25) and we had no memory problems...I'm the only one there using lightwave and I'm pretty sure I couldn't do this project in the same time.

Maybe if you used HD instance? I'm guessing it's the equivalent of vray meshes.

Andyjaggy
06-21-2007, 07:14 AM
But then it's all volumetrics and renders slooooooow, but I guess it would render slow anyways.

Captain Obvious
06-21-2007, 09:18 AM
I just rendered 952 million polygons in modo... *cough*

Lightwolf
06-21-2007, 09:30 AM
I just rendered 952 million polygons in modo... *cough*
Cool... how many of them were modelled? :D

Cheers,
Mike

Andyjaggy
06-21-2007, 09:55 AM
What kind of work do you do in LW neverko?

Lightwolf
06-21-2007, 09:57 AM
That's not the point, the point (I guess) is that it's possible and possible within a modest memory footprint.
Yup, just look at just about any renderer but LW.
But what is the relevance related to openGL?

I rarely need millions and millions of polygons, unless I render displacements.
...
Luckily instancing is among the nice things promised for 9.x :)
Yup, that would solve a lot of my issues (I rarely used displacements, but quite often cloned high poly meshes).

Cheers,
Mike

Dodgy
06-21-2007, 10:01 AM
Umm, I've just rendered 2,980 million polys in Lightwave...

Those figures really don't mean anything. I did mine with HD instance, in about 4 minutes 21 secs, I wouldn't call that quirky old tech if you need instances.... At a res of 2048x2048, you'd have 4million polys at a poly per pixel, unless you had layers of transparency, and at 10 layers, and that would be 40 million. I can barely see my instances at that kind of density.

What we need is a base mesh memory optimization, not just random instances.

Captain Obvious
06-21-2007, 10:50 AM
Yup, just look at just about any renderer but LW.
But what is the relevance related to openGL?
Did I mention that modo was silky smooth displaying the 3,000,000 polygon base object?

zardoz
06-21-2007, 10:57 AM
what is your hardware Captain?

Captain Obvious
06-21-2007, 11:10 AM
Pretty darned high-end; a QX6700 and a Geforce 8800GTX.

But Modeler gets really slow at about 100k polygons...

Lewis
06-21-2007, 01:28 PM
YES we need modeler update ASAP :). And yes it needs to be good one ;).

Andyjaggy
06-21-2007, 02:09 PM
Yes, yes, and yes. Come on Newtek. Modeler update!!!! I want it NOW!............... going off to sulk in the corner.

Red_Oddity
06-21-2007, 05:43 PM
Strangely enough, during beta, i had the feeling the OpenGL was better, but right now it's pretty slow, compared with for example Modo.

even when running LW9.2 64bit on XP64 on a computer with 2x Xeon 5355 processors with 8GB memory and GeForce 880GTX, it's still way too slow, this machine can push so much more.

Lightwolf
06-21-2007, 05:57 PM
Did I mention that modo was silky smooth displaying the 3,000,000 polygon base object?
Lucky you, last time I tried 2.4 million it crashed after eating up my RAM ;)

Cheers,
Mike

Captain Obvious
06-22-2007, 05:21 AM
Well, it was a subdivision surface mesh.

Lightwolf
06-22-2007, 05:24 AM
Well, it was a subdivision surface mesh.
How many polygons on the cage then?

Cheers,
Mike

Lewis
06-22-2007, 06:13 AM
Hmm currently I'm thumbling 324000 SubDs poly object in modeler with very nice FPS (not smooth as baby's but but smoothih enough considering that I have 20+ surfaces and 25 layers) and it's at Subpatch level 4 which equals to 5.1 milion polygons if i freeze it at that level (or render in layout). Problem is with Editing not thumbling :(.

Andyjaggy
06-22-2007, 07:47 AM
Hmm currently I'm thumbling 324000 SubDs poly object in modeler with very nice FPS (not smooth as baby's but but smoothih enough considering that I have 20+ surfaces and 25 layers) and it's at Subpatch level 4 which equals to 5.1 milion polygons if i freeze it at that level (or render in layout). Problem is with Editing not thumbling :(.

Yeah viewing the mesh isn't a problem, editing it is. You either have to divide everything up into a ridiculous amount of layers or hide stuff all the time. Of coarse it takes forever to select stuff to hide it as well.........

Animapper
06-22-2007, 07:58 AM
whoaaa on this one. It depends on the card you have. I saw Maya 8.5 at NAB this year doing realtime wind, particles and full shaders on a nVidia 4500. I would not make this assumption at all.

Regards,

Animapper
06-22-2007, 08:07 AM
Msherak said that maya was a lot slower - not from what i saw at NAB. The open GL performance was outstanding on the quadro 4500 HP workstation.

Animapper
06-22-2007, 08:18 AM
Hmm,

Maya is a lot slower than LW when it comes to GL.. Maya can't do 1/2 the things you can can in LW's GL.. About the only thing that Maya has over LW is the ability to view normal maps in GL.. Course you have to use a Nvidia shader or write your own and use the Hardware render which slows it to a crawl..
Try doing particles in Maya then LW.. For drawing functionality LW blows Maya out of the water.. Heck even LW's particles can be played backwards unlike Maya..
I'm commenting on what Msherak said. He talked about openGL - i was editing in real time on this box with wind, particles and four shaders - We were editing a sign with streamers blowing in the wind in real time. It was amazing. The editing of the mesh was the topic. We don't get this kind of feedback in LW. That is my point.

IMI
06-22-2007, 08:21 AM
Modeler's "tumble" OGL is just fine, even with a dense mesh with hi-res textures, but use something like Drag Net on that same model and watch the performance go way downhill, quickly.
I'd gladly trade all that tumble-ability for some serious realtime performance dragging points around, particularly in symmetry mode.
LW 9's recommendations call for an Nvidia Quadro 4000, IIRC. Why should that be? Should a $2500 video card be required to do the same thing in Modeler that my GeForce 7950 can do in modo?

IMI
06-22-2007, 08:49 AM
I didn't know that, neverko. I always assumed it was the OGL and the hardware acceleration that results in either speed or lag regardless of what action is being taken on my screen.
Makes perfect sense though.

theo
06-22-2007, 09:10 AM
New modeling features.............. screw that i just want everything we have allready to work correctly. Interactivity would be nice. If all they did was add interactivity, faster and more efficient geometry handling, and fix CC's I would be pretty happy.

Absolutely agree!!

1) CC's and edge functionality that actually perform as expected
2) Increased geometry manipulation speeds

This list would suit me perfectly at the moment while other aspects of Modeler would continue to undergo advancement.

Captain Obvious
06-22-2007, 10:20 AM
How many polygons on the cage then?

Cheers,
Mike
About half a million.

Andyjaggy
06-22-2007, 11:31 AM
Well I just hope Newtek is reading this thread and realize what a serious problem this is for all of us. I am sure they are aware of the problem and working hard on it right this moment. It would just be nice to hear from them, all has been pretty quiet for a while now. It would be nice to have an update on the situation.

Stooch
06-22-2007, 11:54 AM
lol sometimes i think NT should just throw out modeler and admit defeat and focus on layout and include a license of modo like they did with vue. clearly these guys dont get it as far as modeling tools go. all the good tools are made by third parties. think about that for a bit.

Andyjaggy
06-22-2007, 01:47 PM
It's not that bad but in all truth I don't see much of a difference from modeler in 7.0 and modeler we have today. There are a few fancy little new tools and stuff but it's basically the same. It needs some serious loving to get it up to par.

Lightwolf
06-22-2007, 01:59 PM
all the good tools are made by third parties. think about that for a bit.
That's an indication of what problems the developers have with the old code. They need concentrate on revamping it, replacing the core - which leaves no time to add tools but paves the way for new tools in the future.

Then again, unlike third parties they _can_ actually change the core (I wish I could ;) ).

Cheers,
Mike

jin choung
06-22-2007, 02:00 PM
modeler is great.

tools wise, it totally rocks. if it's a matter of optimizing the geometry handling core, that is manageable.

let's not throw out the baby with the bathwater.

jin

Andyjaggy
06-22-2007, 02:43 PM
well I am obviously still using it so there are enough great things about it to keep me coming back. It's just that I spend most of my time in modeler and some of those other options are looking mighty tempting at the moment. I feel it safe to say that the 9.x cycle will make or break Lightwave for a lot of people. I feel confident that good things will happen and I have hope for the future, but if it all comes crumbling down I am out of here.

Stooch
06-22-2007, 03:25 PM
what i mean by "they dont get it" has nothing to do with core optimization at all. infact this could have been done a long time ago. for example, look at the tools like pictrix, lwcad etc. you can clearly see that they can and do make use of whats already there but are more concerned with WORKFLOW.

tool consolidation could have been done a long time ago without the core updates, and there is a huge demand for this that has been going on for years. so whats the problem? is it that hard to combine some similar tools and make them context sensitive? no one is asking for miracles here. the whole speed of the core could be more tolerable if you can work more efficiently to begin with. afterall some of us have been putting up with LWs lack of speed all this time.

everyone marvels how modo is getting all the glory. and what does it offer? WORKFLOW. front and center.

vadermanchild
06-22-2007, 03:43 PM
This forum is revolting.

Seriously.

Im hearing a lot more once content LW forum members sounding more glum with each passing week of using 9.2

Newtek should be worried. If this loyal area of the community can be shook by a poor release like 9.2 they need to speed things up or face the consequencies. The 9.2 PR release does not seem address any of the serious problems.

Andyjaggy
06-22-2007, 04:32 PM
No. 9.2 was awesome. It just didn't do much for Modeler or animation. I know Newtek promised that was next, we are all just a little impatient I suppose.

Lightwolf
06-22-2007, 04:43 PM
what i mean by "they dont get it" has nothing to do with core optimization at all.
Erm, it does. Of course, they could waste years hacking together consolidated tools on top of the current core only to see them break once they get the core up to par.
Or, work on the core first, with consolidation and the tons of other _core_ issues in mind (i.e. speed of certain operations, CCs, edges, edge weights...) and then add on top of that. That way they'd at least not waste valuable ressources of internal developers.

everyone marvels how modo is getting all the glory. and what does it offer? WORKFLOW. front and center.
Well, if you consider that the design of modo has probably started almost 10 years ago, the new core was started to be written... 5,6,7(? guessing here) years ago - and we on the other side have a new team here that is just getting used to the systems they are (currently) forced to develop and think in...
Yes, it is asking for a lot.
Now, from a users perspective, I absolutely see your point. However, since I code, I can also guess at the pain involved from the dev teams side.

Having said that, 9.0 and especially 9.2 have a lot of good things going for it (especially in Layout). I might not always agree on the design decisions, but there's an aweful lot of nice stuff in there.

Cheers,
Mike

Stooch
06-22-2007, 04:59 PM
This forum is revolting.

Seriously.

then leave.

people here are posting with the intent to see changes. if you are happy with LW then shut up and keep using it.

Stooch
06-22-2007, 05:05 PM
Erm, it does. Of course, they could waste years hacking together consolidated tools on top of the current core only to see them break once they get the core up to par.
Or, work on the core first, with consolidation and the tons of other _core_ issues in mind (i.e. speed of certain operations, CCs, edges, edge weights...) and then add on top of that. That way they'd at least not waste valuable ressources of internal developers.

Well, if you consider that the design of modo has probably started almost 10 years ago, the new core was started to be written... 5,6,7(? guessing here) years ago - and we on the other side have a new team here that is just getting used to the systems they are (currently) forced to develop and think in...
Yes, it is asking for a lot.
Now, from a users perspective, I absolutely see your point. However, since I code, I can also guess at the pain involved from the dev teams side.

Having said that, 9.0 and especially 9.2 have a lot of good things going for it (especially in Layout). I might not always agree on the design decisions, but there's an aweful lot of nice stuff in there.

Cheers,
Mike

lol you are telling me these things like i dont know them. I have coded before. I see a bunch of modeling plugins that modeler loads. so im guessing that an additional .p file can be added that can merge the function of some of these redundant tools. and there is a whole crapload of software out there that focuses on modeling and has sprang up in the last couple of years. they seem to be doing just fine as far as creating a tool from scratch, not just combining a FEW EXISTING TOOLS in a logical fashion.

Lightwolf
06-22-2007, 05:12 PM
lol you are telling me these things like i dont know them.
Well, looking at your post that was the conclusion I drew...


I have coded before.

Then you should know better ;) Unless you haven't touched the LW SDK, then I forgive you ;)


...and there is a whole crapload of software out there that focuses on modeling and has sprang up in the last couple of years...
Oh, out of nothing, just like that? Obviously not. How long did it take them to get to a V1.0? Which is almost the situation we seem to have here. We have a V1.0 that probably needs to be trimmed down to a V0.5 (throwing out garbage) before you can bump up the version again and build new stuff.
The problem seems to be though that NT probably can't afford to duck and hide for a few years while they rewrite, what they do now seems to be the only way to actually stay in the game. A tough, bumpy, painful ride... Layout has shown a glimpes of what can come of it in a good way... I hope Modeler will soon as well.

Cheers,
Mike

Stooch
06-22-2007, 05:24 PM
so basically you are arguing with me because you dont agree that they should consolidate tools? i dont understand where you are going with this.

im here voicing my opinion to NT. i am not telling anyone how to feel and would appreciate if i was afforded the same luxury.

thanks.

Andyjaggy
06-22-2007, 05:43 PM
Yes. Let's not turn this thread into a pissing match. Although that is almost gauranteed to happen after 100 posts or so. Even less if Oddity is part of it.

Lightwolf
06-22-2007, 05:44 PM
so basically you are arguing with me because you dont agree that they should consolidate tools? i dont understand where you are going with this.
No, I'm not even arguing. All I'm saying is it isn't reasonable to expect miracles.
I want tools consolidated as much as anybody else (likely more so) - but I want it done properly (because, imho, NT can't afford to take a sloppy development approach as they don't have the ressources for that).
Which means: planning, design and most likely top to bottom design but bottom to top coding. The disadvantage is you won't see a lot happening at the surface initially, but that is essential work to be done. And I think we are in a transitional period like that.

im here voicing my opinion to NT. i am not telling anyone how to feel and would appreciate if i was afforded the same luxury.

Am I telling you how to feel? If I do sorry for that, wasn't my intention. (I only said you should know, not feel better ;) <- joke! ).

But you did write "so whats the problem?" and was trying to give you some (more or less speculative) opinion on what might be the problem.
Didn't work as intended, did it? ;)

Cheers,
Mike

zareh
06-22-2007, 06:17 PM
Am I the only one that find the OGL performance in modeler absolutely pathetic? Yes it's great we can tumble large object now but try moving that object across the screen and see what happens. Drag. Wait 15 seconds for update. Drag. Wait 15 seconds for update. etc.......

I am working on a project where I am building an entire city and even with splitting everything up into small little chunks as soon as I hit 100k or 150k the model becomes unworkable. To me this should be the #1 priority with the next update. What good are new tools if you can't use them because the OGL is so slow. Grrrrrrrrrrrrrrrr.

According to my benchmarks the OpenGL performance in 9.2 is orders of magnitude faster than 9.0 and 8.5. In Multitexture/Buffered(VBO) mode. However you're absolutely right in that moving objects,polygons or points is very slow. I think this is due not to the OpenGL implementation but to the internal structures used to represend the polygonal objects.

Best Regards,
Zareh

DogBoy
06-22-2007, 06:26 PM
Yes. Let's not turn this thread into a pissing match. Although that is almost gauranteed to happen after 100 posts or so. Even less if Oddity is part of it.

<kvetch> Well it has already sunk into a thread for personal whinging not actually relevant to the OP.

I'm with LightWolf here, we seem to expect miracles: "Look how good Modo is!" Well I should bl**dy well hope so! They spent enough time on LW to have got an idea of where to go, and they had the luxury of losing most of the baggage, and half-thought out solutions LW is plagued with, along the way. Lux still have issues wi' shipping dates and stability, let's not forget.

I was pretty impressed with where 9.2 improved. Yes, tool consolidation is a must. Yes it would be nice if modeler had a better/faster transformation system, but that is not the fault of the guys who have made the changes to the OpenGL implementation. The updates it contained made my life easier.

NT are trying to be focused on development. They have a small dev team and as LightWolf mentioned, they do not have the luxury of doing a a full overhaul in seclusion. They/we want LW 9.x to have a (relatively) steady release schedule. That means they are tied into working in a way that may not get the fastest results.
We are promised a bug-squash release next, to quieten the those who complain that 9.2 didn't focus clearly enough on that area. Only then will they move onto the next area of work.

These thing, unfortunately, take time. Our constant p!ssing on doesn't help in the slightest. If you want LW to improve get on the OB & crush bugs, so we can move on to the next point release, and the much wanted modeler updates, sooner.</kvetch>

vadermanchild
06-22-2007, 06:32 PM
then leave.

people here are posting with the intent to see changes. if you are happy with LW then shut up and keep using it.


Wow - calm -

I agree ENTIRELY that LW is weak in many areas.

My point was that even the normal happy to overlook LW problems fanbase that reside here are revolting - the normal happy to wait and happy to wait for basic features other programmes have had for years members are beginning to speak up.

THat what I was getting at - Keep your hair on.:) Im with you that LW is fast becoming a sinking ship.

Scazzino
06-22-2007, 10:27 PM
I must say that LightWave's OpenGL is MUCH faster now than it was back in 8.5 on the Mac. I can spin around models and scenes now that were like molasses before. As has been mentioned before, editing in Modeler is not an OpenGL problem, it's much more and goes to Modeler's old core. It will take some time to rebuild Modeler's core to handle more modern editing requirements of course.

Just look what NewTek did with their new core changes in Layout so far. As an example the new cameras, motion blur / DOF, AA, node based surfacing and interpolated radiosity are brand new areas built on a new foundation. It'll take time for them to get to each area of both applications and replace all the rickety old code with a more modern foundation.

Based upon what I've seen in the new areas of LightWave so far, I'm confidently looking forward to see what they can do with the remaining areas. LightWave is indeed being reborn, but it will take time and patience before it's completely reborn... ;)

Andyjaggy
06-22-2007, 10:56 PM
Well I have learned one thing from this thread. Editing meshes isn't really an OGL problem. I guess that does make sense. It's just so frustrating when you are working with 200,000 poly and up models everyday and you can hardly do anything because LW starts moving slower then molasses. Whine whine complain moan grumble. Suck it up and make it work Andy.

vadermanchild
06-22-2007, 10:59 PM
I must say that LightWave's OpenGL is MUCH faster now than it was back in 8.5 on the Mac.



When Newtek said 9.2 would offer imporvements I really did think they meant it would be comparable with other software and not just an imporvement over 8.5 which was one of the most emabarrassing thing I have ever seen.


9.x is better but you do realise its still almost as emabarrasingly slow? (and yes yes yes - it pushes a few hundred polys around fine but my objects are all too large to use modeler to edit) So lets not compare 9.2 to 8.5 - compare it to something people enjoy using then the discussion is more realistic.

scenicdave
06-22-2007, 11:33 PM
When Newtek said 9.2 would offer imporvements I really did think they meant it would be comparable with other software and not just an imporvement over 8.5 which was one of the most emabarrassing thing I have ever seen.


9.x is better but you do realise its still almost as emabarrasingly slow? (and yes yes yes - it pushes a few hundred polys around fine but my objects are all too large to use modeler to edit) So lets not compare 9.2 to 8.5 - compare it to something people enjoy using then the discussion is more realistic.

Hey Guys,
The one thing I see emerging from this thread is that Modeler compared to other softwares is lacking, apparently. But this comparison, as regards to modeling, is irrelevant. Can Modeler produce models quickly? Yes.

'My Ferrari can do 340kph!!' (Well, yes, but we're just doing the grocery shopping today.)

''My Ferrari can get to 100kph in 2.6 seconds!' (Yesss, yknow, we might just catch a bus back from the shops)

'My Ferrari is really really red!' (Yep; see ya)

If Modeler slows down at 100,000 polys then start a new layer.

Max has to throw huge numbers of polys around because Max models are incredibly fat.

There is a lot of modeling software available now that is too complex. Most good modeling is done by using simple tools cleverly, not by clever tools.

And at the risk of annoying our dissenters, I have this question: what on Earth are you making that requires 3 million polys? Do you think high poly counts are a good thing? Any object or scene that gets that big needs to be split up. It's just not necessary to have the whole world in one layer.

Modeler is a perfectly good, very fast modeling package in the right hands. It's strength is it's simplicity. Calls for more complex tools are a complete misdirection.

regards
Dave Tracey
Sydney

mattc
06-23-2007, 12:53 AM
so basically you are arguing with me because you dont agree that they should consolidate tools? i dont understand where you are going with this.


Err, no, what Mike is saying is that it isn't so easy to consolidate everything and do what you want in a short amount of time. Given that NT is essentially pursuing the parallel changeover strategy (and really, I don't think they had much choice from a business standpoint) there's going to be some pain as we all sit round waiting for this to occur.

FWIW, I agree with you. I'm sure Mike does as well. Getting there is the issue.

M.

jin choung
06-23-2007, 01:35 AM
re: tool consolidation -

man... can't argue with that. even back when i heard that they were going to introduce EDGES as selectable entities, i was sounding the warning alarm saying that there's probably gonna be a sh!tstorm of redundant tools if they're not careful...

alas, they were not careful.

but as others pointed out back then, it's a matter of clean up. unfortunately, it's simple but unglamorous and probably more complicated than it should be... so it'll probably take a while.

anyhoo, can't agree more.

jin

alifx
06-23-2007, 01:55 AM
just have some patience ..

we are still in 9.2 ........Nine Point TWO0o

we have been promised for CA animation improvements and they're coming in the next release.... maybe it's 9.3 or 9.4 or 9.whatever

we have been promised in Modeling also, it should be improved in the same series of 9.x

9.2 was a the first step to achive all these improvements, we still have some more versions coming up before LW go to R10


If NT reneged they promise (I don't think they do!)... then do anything you want!

we should trust NT to make LW to survive from whats going in the community these days.

Red_Oddity
06-23-2007, 05:41 AM
Well, most off us use LW in our pipelines, and patience is not an option then.

To use the darn car analogy again...sure, i drive a Ferrari, but the last 6 months it has been in the car shop because one of the wheel came off, one seat is missing, and for some reason the battery drains really fast when i put too all lights on.

Anyhoo. It doesn't mean we don't appreciate what has been done so far and where it is going, but for some off us a broken tool means a broken pipeline. (And yes, there are always workarounds, but they take time to figure out, and a pipeline should be solid yet flexible, bet never stuck together with duct tape.)

Matt
06-23-2007, 06:16 AM
Well it has already sunk into a thread for personal whinging not actually relevant to the OP.

I'm with LightWolf here, we seem to expect miracles: "Look how good Modo is!" Well I should bl**dy well hope so! They spent enough time on LW to have got an idea of where to go, and they had the luxury of losing most of the baggage, and half-thought out solutions LW is plagued with, along the way. Lux still have issues wi' shipping dates and stability, let's not forget.

I was pretty impressed with where 9.2 improved. Yes, tool consolidation is a must. Yes it would be nice if modeler had a better/faster transformation system, but that is not the fault of the guys who have made the changes to the OpenGL implementation. The updates it contained made my life easier.

NT are trying to be focused on development. They have a small dev team and as LightWolf mentioned, they do not have the luxury of doing a a full overhaul in seclusion. They/we want LW 9.x to have a (relatively) steady release schedule. That means they are tied into working in a way that may not get the fastest results.
We are promised a bug-squash release next, to quieten the those who complain that 9.2 didn't focus clearly enough on that area. Only then will they move onto the next area of work.

These thing, unfortunately, take time. Our constant p!ssing on doesn't help in the slightest. If you want LW to improve get on the OB & crush bugs, so we can move on to the next point release, and the much wanted modeler updates, sooner.

Quoted for total agreement and the only sensible post in this thread!

:hey:

I would also like to add, NewTek do, despite what people think, actually READ these threads, so can people try and keep comments constructive and on topic, just ranting helps no one, check your post before casting dispersions on what NT are and are not doing.

They are working damn hard with very old code, trying to drag it kicking and screaming into the 21st Century, from what I can gather it's an uphill struggle, one they are winning, but it will take time.

So let's help them out rather than fighting against them, they are on OUR side you know!!!

lino.grandi
06-23-2007, 07:01 AM
Looking at what the NT dev team has done to LW render, I really don't understand how we shouldn't be confindent about the fact they're going to make the whole Lightwave better than ever.

I think it's just a question of time, and every aspect of the software will be "evolved".

What they're doing is great and greatly difficult to do.

I admire the work that has been done in Layout, and I'm sure that Modeler is in the works.

It's ok to ask (and never as this time Newtek is listening to the users) but I really think the coders deserve respect, they're working hard, and Jay knows what we need.

Andyjaggy
06-23-2007, 07:11 AM
Hey Guys,
The one thing I see emerging from this thread is that Modeler compared to other softwares is lacking, apparently. But this comparison, as regards to modeling, is irrelevant. Can Modeler produce models quickly? Yes.

'My Ferrari can do 340kph!!' (Well, yes, but we're just doing the grocery shopping today.)

''My Ferrari can get to 100kph in 2.6 seconds!' (Yesss, yknow, we might just catch a bus back from the shops)

'My Ferrari is really really red!' (Yep; see ya)

If Modeler slows down at 100,000 polys then start a new layer.

Max has to throw huge numbers of polys around because Max models are incredibly fat.

There is a lot of modeling software available now that is too complex. Most good modeling is done by using simple tools cleverly, not by clever tools.

And at the risk of annoying our dissenters, I have this question: what on Earth are you making that requires 3 million polys? Do you think high poly counts are a good thing? Any object or scene that gets that big needs to be split up. It's just not necessary to have the whole world in one layer.

Modeler is a perfectly good, very fast modeling package in the right hands. It's strength is it's simplicity. Calls for more complex tools are a complete misdirection.

regards
Dave Tracey
Sydney

That is the stupidest post I have read in a long time. You are obviously completely clueless. Yes in my experience Max modelers use way more polys then necessary, although I am sure that is not a universal truth.

Trust me when I say I am a good modeler and know how to keep my poly counts low. By saying that anyone who is working on a model that has 3 million polys is a lazy and bad modeler is pure ignorance.

Andyjaggy
06-23-2007, 07:22 AM
This is what happens when you try to point out a flaw in LW. A bunch of people start yelling in agreement and then a bunch of people get start yelling in defense of Newtek and trying to pretend like all is just swell and great.

The truth is Modeler is fastly becoming an outdated app.
Is Newtek working hard. Yes. Is Newtek promising to fix this. Yes. Is it going to get fixed. Yes. Do we all need to excercise patience with Newtek while they fix the old code. Yes. Do all of us have the time and ability to sit around for a year while they fix it. No.

I'm all with splitting objects up into manageable and reasonable files and sizes. But having to split an object into 300 seperate object just to get Modeler to work at a useable speed doesn't sound like good workflow to me.

jameswillmott
06-23-2007, 07:38 AM
Does pointing out a flaw that's already been identified and promised to be fixed, actually help to fix it any faster? :D

Andyjaggy
06-23-2007, 07:45 AM
Probably not no :D Some of use need to vent though. I'm sure Newtek is doing what they can but in the meantime I am going to poind my fists on the table and scream like a little baby till it gets fixed :)

Scazzino
06-23-2007, 08:26 AM
Well I have learned one thing from this thread. Editing meshes isn't really an OGL problem. I guess that does make sense. It's just so frustrating when you are working with 200,000 poly and up models everyday and you can hardly do anything because LW starts moving slower then molasses. Whine whine complain moan grumble. Suck it up and make it work Andy.

Yes, I understand completely. The OpenGL "is" much improved, but of course Modeler is still a very long way from where it should be and where it will be in most other aspects. They haven't gotten to Modeler much yet other than improving the OpenGL so far. I don't think the editing operations have been worked on much at all yet. Most of the focus of the update so far has gone into Layout and the renderer.

All I'm saying is that based on what I've seen that they have done in the new areas of Layout, that I have confidence that they will be able to improve Modeler dramatically as well, when they get to it... It just takes time, that's all... There's an awful lot of very old code that was not designed for modern technology. Rather than trying to patch that up here and there, they are systematically replacing the underlying foundation with a more modern one, one module at a time. That takes time to do right. We now have a new camera system which is basically an entirely new render engine which has enabled new features like the photorealistic motion blur and DOF. They will do the same with Modeler I'm sure, and probably have been working on designing the new foundation behind the scenes already.

I feel your pain if you work primarily in Modeler since they started with updating modules in Layout first. I primarily use Layout myself, so I've been impressed with the progress so far. I have no doubt that Modeler will be drastically improved as well, but it can't happen overnight unfortunately.

Keep the faith... :beerchug:

jat
06-23-2007, 08:41 AM
just my two cents............. I really need to be able to work quickly and efficiently with my models and not worry about spining beach balls, redraws and setting patch dvisions and dealing with OGL buffers and what not. I don't really care what they do to fix this problem as long as its fixed and fast.They all know what they have to do but I've been using LW for a while and its always wait, hold on, almost there, a couple more months, next upgarde etc and its just getting old............just sucks that's all................MOLASSES!!!!!!!!!!!!!!!!!!!

Karmacop
06-23-2007, 08:43 AM
Didn't someone link to an interview with David Ikeda recently where he mentioned he spent 90% of his time is maintenance work in modeler? To me that sounds as if they are addressing this editing slowness along with other problems with a fair bit of effort, it's just taking a long time.

So yes it's slow, but they are (seemingly) putting a big effort into it. This is why you should buy the software for the features it has, not what they promise it will have :)

Andyjaggy
06-23-2007, 08:52 AM
Yes, I understand completely. The OpenGL "is" much improved, but of course Modeler is still a very long way from where it should be and where it will be in most other aspects. They haven't gotten to Modeler much yet other than improving the OpenGL so far. I don't think the editing operations have been worked on much at all yet. Most of the focus of the update so far has gone into Layout and the renderer.

All I'm saying is that based on what I've seen that they have done in the new areas of Layout, that I have confidence that they will be able to improve Modeler dramatically as well, when they get to it... It just takes time, that's all... There's an awful lot of very old code that was not designed for modern technology. Rather than trying to patch that up here and there, they are systematically replacing the underlying foundation with a more modern one, one module at a time. That takes time to do right. We now have a new camera system which is basically an entirely new render engine which has enabled new features like the photorealistic motion blur and DOF. They will do the same with Modeler I'm sure, and probably have been working on designing the new foundation behind the scenes already.

I feel your pain if you work primarily in Modeler since they started with updating modules in Layout first. I primarily use Layout myself, so I've been impressed with the progress so far. I have no doubt that Modeler will be drastically improved as well, but it can't happen overnight unfortunately.

Keep the faith... :beerchug:

I am prepared to be amazed, it's just a matter of how long I'm going to have to wait for that to happen. I love Lightwave, and I love the tools in modeler, I just don't want to wait 15 seconds everytime I drag an object across the screen or go to select some points. It makes me mad when you get people on here that say, oh that isn't a problem, the slow editing is actually a benefit because it forces you to keep your poly count low. Any real modeler would never have poly counts over 100,000 (which is the limit where LW becomes unusable now.) and telling everyone something isn't a problem when it is a serious problem for a lot of people here.

Matt
06-23-2007, 08:57 AM
I had the opportunity to witness one of the NewTek coders working with the old code, now I understand completely what they're up against!

;)

lino.grandi
06-23-2007, 09:01 AM
just my two cents............. I really need to be able to work quickly and efficiently with my models and not worry about spining beach balls, redraws and setting patch dvisions and dealing with OGL buffers and what not. I don't really care what they do to fix this problem as long as its fixed and fast.They all know what they have to do but I've been using LW for a while and its always wait, hold on, almost there, a couple more months, next upgarde etc and its just getting old............just sucks that's all................MOLASSES!!!!!!!!!!!!!!!!!!!

We already can do great things with Lightwave at its current state. Saying that the application "just sucks" is really a weird sentence.

If you can say things such as "and its always wait, hold on, almost there, a couple more months, next upgarde etc"...well, I just think you're blind, since we have a brand new render and radiosity.

I'm using Lightwave 9.2 for my project NOW, I really don't need to wait anything, even if any time the application gets better my way of working gets better too.

Andyjaggy
06-23-2007, 09:05 AM
I had the opportunity to witness one of the NewTek coders working with the old code, now I understand completely what they're up against!

;)

Nice.

JensD
06-23-2007, 09:37 AM
A NewTek Developer said in a german Forum that they are now working on some relics of LightWave, like meshedit and the animationsystem which both highly limit the possible speed.

Cheers

Jens

vadermanchild
06-23-2007, 09:43 AM
Hey Guys,
Can Modeler produce models quickly? Yes.

If Modeler slows down at 100,000 polys then start a new layer.

And at the risk of annoying our dissenters, I have this question: what on Earth are you making that requires 3 million polys? Do you think high poly counts are a good thing? Any object or scene that gets that big needs to be split up. It's just not necessary to have the whole world in one layer.regards
Dave Tracey
Sydney

It slows down a lot sooner than 100K....

Im not even going to begin to explain why more polys are needed, perhaps your comment was a throw away one - but its not about Ferrari speeds - Im after a Lada that can do zero to 60 mph in 35 seconds.... unfortunately modeler isnt even in that range.

If your not doing anything that needs more than 10 K then great - if your happy selecting an eye socket and hiding the cheek and jaw just to make a small tweak at a reasonable speed then great - seems plenty people are unhappy - lets hope newtek dont work to please the lowest common denominator.

vadermanchild
06-23-2007, 09:45 AM
trying to pretend like all is just swell and great.

The truth is Modeler is fastly becoming an outdated app.



Nail hit on head.

vadermanchild
06-23-2007, 09:53 AM
Does pointing out a flaw that's already been identified and promised to be fixed, actually help to fix it any faster? :D

I agree it wont help fix it any faster - although im not sure Ive seen the same info as you - Newtek said we would be impressed with the 9.2 modeler changes - we`re clearly not - if there is a press release from Newtek detailing what to expect with some reasonalble info in it then Im eager to see the link :D very eager as it will keep me calm as I try to edit what I consider simple meshes....or I load the object into "another" app and get my modeler editing calmness there.:thumbsup:

Andyjaggy
06-23-2007, 10:01 AM
There is a lot of modeling software available now that is too complex. Most good modeling is done by using simple tools cleverly, not by clever tools.

And at the risk of annoying our dissenters, I have this question: what on Earth are you making that requires 3 million polys? Do you think high poly counts are a good thing? Any object or scene that gets that big needs to be split up. It's just not necessary to have the whole world in one layer.

Modeler is a perfectly good, very fast modeling package in the right hands. It's strength is it's simplicity. Calls for more complex tools are a complete misdirection.

regards
Dave Tracey
Sydney

It's not even about new clever tools. If you look at my previous posts you'll see that I said I could care less about getting new tools. I just want what we have to work and work fast. I can make anything I want with the tools we currently have. Sure some new fancy tools and functionality might make things a little easier (better rounding, freezing levels, interactivity, instancing, subpatched UV export, C-planes, etc... I could go on and on :)) but even without those tools I still find Modeler very powerful. SLOW but powerful.

DogBoy
06-23-2007, 10:26 AM
Probably not no :D Some of use need to vent though. I'm sure Newtek is doing what they can but in the meantime I am going to poind my fists on the table and scream like a little baby till it gets fixed :)

:D

I don't think anyone is saying it's perfect. We all agree it's not, but I just think this went from a "why isn't OpenGL doing what I think it should?", to being a bit of a whinge (I'm sorry for my earlier rudeness BTW, I'd had a few drinkies last night). We all need to vent, I agree, but I really hope that the deformation slowness will get looked at soon.

If NT were to step up and give us an idea of it's running order for upcoming work, maybe it would calm folks down.

Dan

Lewis
06-23-2007, 10:27 AM
It slows down a lot sooner than 100K....

If your not doing anything that needs more than 10 K then great - if your happy selecting an eye socket and hiding the cheek and jaw just to make a small tweak at a reasonable speed then great - seems plenty people are unhappy - lets hope newtek dont work to please the lowest common denominator.

Hmm !

Are you sure you are using MTS + VBO mode (not GSLS + VBO) in modeler ? I'm not saying it's super fast (but NT is working on it for sure) but 10k model is piece of cake in modeler 9.2 even on my old P4 3Ghz+Nvidia 5950Ultra which is 4+ years old by now ? New one Dual core + 8800 GTX handles polys (in modeler) around 2x times faster from what i can see every day when i compare my old and new machine ;).

I just finished 324000 subDs Aircraft model (level 4 subpatches while redrawing/modeling). Before that I made 1.8million polygons Building and before that several other "heavy" objects. I'm really using hi-poly models every day and i agree that low-poly/optimize times are behind us (ofcouse game stuff needs to be low poly but I'm talking about Achitectural stuff/broadcast/print res models) One thing i noticed is that Modeler tools are resonably faster if you aren't doing it in subDs (TAB) mode.

cheers

Andyjaggy
06-23-2007, 10:30 AM
A little communication would be nice. We might have to wait till Siggraph though. We all complain about the lack of communication but we have tons more then if we were using an 'Autocash' product. That's part of the appeal of LW though is the community and the open friendly communication we sometimes have with the NT staff.

Castius
06-23-2007, 11:28 AM
Deforming mesh in modeler is a lot slower than the competition. Modeler version 9.2 does not deform mesh any slower than 9.0. If your work requires you to deform mesh more than the software is capable of then find a way to work around it. It's no different anywhere you go.

In Max or Maya i can't paint weight maps on a mesh with 100k poly without pulling my hair out. Some tools are not designed to handle a certain amount of raw data and they need to get updated. So right now we are just waiting.

I think this discussion is over I'm my opinion. The openGL is not slow it's the deformation and it's been stated by Newtek that the deformation system is part of the revamping of the animation system. We just have to wait and see how that turns out.

Andyjaggy
06-23-2007, 11:44 AM
No the discussion is not over. I told you I would bang my fists and scream until I got my way, and dang it I am going to do just that. J/K yeah I think all that could be said has been said. Not much else to say. You've got two choices, patiently wait it out or go somewhere else.

vadermanchild
06-23-2007, 12:04 PM
You've got two choices, patiently wait it out or go somewhere else.

If there is public news from Newtek about how succesfull they are being on this and other important issues with some idea of timescales it might stop your some people from taking up your second option.


I feel once a user leaves modeler for another appsit will be hard to win them back which is very bad news for all of us.

sean hargreaves
06-23-2007, 12:08 PM
Guys, just don't p--s off Andys cat, its a killer. I mean, look at it! :D

Lightwolf
06-23-2007, 01:23 PM
If there is public news from Newtek about how succesfull they are being on this and other important issues with some idea of timescales it might stop your some people from taking up your second option.
As long as people don't complain when a certain development target can't bet hit in time (just think of the modelling within Layout fiasco). Come to think of it, that is probably another reasons why things might just take a bit longer.

Cheers,
Mike

stevecullum
06-23-2007, 02:40 PM
Its really not that suprising that the feature development and code updates are taking so long. Think of the brief the coders must have got :

"We want you to totally upgrade every part of LW as if its a new product, but your not allowed to break anything or knacker backward compatibilty...and you have to make all this new code to work with sloppy out of date code, so folks can continue to work in LW....and we want it yesterday!"

Thats a pretty tough brief IMO. Personally, I think they should have thrown caution to the wind and created a totally new LW, regardless of backwards compatibility... :rolleyes:

-EsHrA-
06-23-2007, 02:45 PM
Personally, I think they should have thrown caution to the wind and created a totally new LW, regardless of backwards compatibility... :rolleyes:

*cough* ... m*d* .. *cough*..

:)

mlon

stevecullum
06-23-2007, 02:57 PM
Haha...I was wondering how long it would take to throw that one in...

vadermanchild
06-23-2007, 03:05 PM
*cough* ... m*d* .. *cough*..

:)

mlon

you missed out a couple ooohhs. 8)

lino.grandi
06-23-2007, 04:39 PM
No the discussion is not over. I told you I would bang my fists and scream until I got my way, and dang it I am going to do just that. J/K yeah I think all that could be said has been said. Not much else to say. You've got two choices, patiently wait it out or go somewhere else.

If I were a coder, I wouldn't be happy to work with a screaming baby in my room.

And I wouldn't be happy to read on a forum I'm not doing a good work.

If the problem with high poly mesh editing will be solved, it will be because it is already in the works, not because you're "banging your fists and screaming".

Newtek is well aware of the problems present in Modeler...and, I repeat myself, IS LISTENING NOW.

What you're doing is just disturbing noise at this point, nothing else.

lino.grandi
06-23-2007, 04:46 PM
Thats a pretty tough brief IMO. Personally, I think they should have thrown caution to the wind and created a totally new LW, regardless of backwards compatibility... :rolleyes:

Writing a brand new application from scratch requires a lot of time as well (see Modo....is growing, but in how many years of development?).

If LW is still alive it's because it's renewing itself, with the dev team solving problems standing there for so long.

Lightwave it's changing...and every change it's going to be deeper. Yes, it's not an easy and fast process...but it's already started and leading somewhere I have faith in, on the basis of what I've seen till now.

Andyjaggy
06-23-2007, 05:04 PM
If I were a coder, I wouldn't be happy to work with a screaming baby in my room.

And I wouldn't be happy to read on a forum I'm not doing a good work.

If the problem with high poly mesh editing will be solved, it will be because it is already in the works, not because you're "banging your fists and screaming".

Newtek is well aware of the problems present in Modeler...and, I repeat myself, IS LISTENING NOW.

What you're doing is just disturbing noise at this point, nothing else.

Being sarcastic there. Although from what I hear that's largely an American thing and lost on a lot of cultures.

lino.grandi
06-23-2007, 05:07 PM
Being sarcastic there. Although from what I hear that's largely an American thing and lost on a lot of cultures.

Sorry...:D

However....I'm pretty sure our :lwicon: is growing and going to grow well. You do really good things using it!

Lightwolf
06-23-2007, 05:20 PM
Being sarcastic there. Although from what I hear that's largely an American thing and lost on a lot of cultures.
Being sarcy is? Oh boy... _Now_ I could rant ;)

Cheers,
Mike

sean hargreaves
06-23-2007, 05:38 PM
One name Andy........Oddity!

jat
06-23-2007, 06:07 PM
We already can do great things with Lightwave at its current state. Saying that the application "just sucks" is really a weird sentence.

If you can say things such as "and its always wait, hold on, almost there, a couple more months, next upgarde etc"...well, I just think you're blind, since we have a brand new render and radiosity.

I'm using Lightwave 9.2 for my project NOW, I really don't need to wait anything, even if any time the application gets better my way of working gets better too.


lol.....the thread is about OGL isn't it? The comments were specifically about OGL because it does SUCK on my Mac waiting for redraws and beach ball updates and so on...............read carefully before you post....

Andyjaggy
06-23-2007, 06:39 PM
Oh no I am being compared to Oddity. Never thought that would happen :D

Lightwolf
06-23-2007, 06:47 PM
Oh no I am being compared to Oddity. Never thought that would happen :D
Yup, I find it surprising as well, after all, your avatar is totally different :D

Cheers,
Mike

Andyjaggy
06-23-2007, 07:45 PM
Totally different. Instead of a guy looking like he is about to attack someone, its a cat that looks like he is about to attack someone.

sean hargreaves
06-23-2007, 08:55 PM
No guyd, Andy said that maybe sarcasm was an American thing, and I disagree and my example was Oddity.....see! :D

And, I love Andy's fat cat!

Andyjaggy
06-23-2007, 09:08 PM
I highly suggest getting a cat avater. It will help with your 3D alot. And now this thread has been totally derailed. Which is okay it was about done anyway.

scenicdave
06-23-2007, 10:09 PM
That is the stupidest post I have read in a long time. You are obviously completely clueless. Yes in my experience Max modelers use way more polys then necessary, although I am sure that is not a universal truth.

Trust me when I say I am a good modeler and know how to keep my poly counts low. By saying that anyone who is working on a model that has 3 million polys is a lazy and bad modeler is pure ignorance.

Hey Andy,
Good tanty you throw there!! Did widdle bumpkins get upset??

Crying and screaming is childish behaviour; as is rudeness.

For the record, I don't trust you,
regards
Dave Tracey
Sydney

scenicdave
06-23-2007, 10:13 PM
No the discussion is not over. I told you I would bang my fists and scream until I got my way, and dang it I am going to do just that. J/K yeah I think all that could be said has been said. Not much else to say. You've got two choices, patiently wait it out or go somewhere else.

There, there Andy; hush child...try to play nice with the other kiddies...

regards
Dave Tracey
Sydney

sean hargreaves
06-23-2007, 10:16 PM
Get 'im Andy, get 'im.....!! :twak:

Andyjaggy
06-23-2007, 10:35 PM
Whatever.

jin choung
06-23-2007, 10:41 PM
the baby screams because it has needs that are not being attended to and is not certain that it is indeed being heard.

i think part of the decibelage of certain rants is because we have no indication that something has registered with the developers.

newtek has developed PHENOMENALLY when it comes to user feedback and openness in communications. the discrepancy between now and what it used to be is indescribable.

but i would say that they could reduce the "noise" of threads such as this simply by a developer showing up and posting "YOU HAVE BEEN HEARD".

it is impossible... IMPOSSIBLE to address every thread and concern in depth.

but it IS possible simply to assure us that our concerns have been noted.

jin

scenicdave
06-23-2007, 10:45 PM
Whatever.
Hi Andy,
Excellent! Much better!

Now, what are you trying to build with 3 million polys?

regards
Dave Tracey
Sydney

DogBoy
06-24-2007, 02:49 AM
Totally different. Instead of a guy looking like he is about to attack someone, its a cat that looks like he is about to attack someone.

Andy, no offense, but that cat can hardly walk let alone mount an attack. it looks like it wishes it could attack someone :D
As opposed to oDDity who attacks constantly ;)

lino.grandi
06-24-2007, 03:08 AM
lol.....the thread is about OGL isn't it? The comments were specifically about OGL because it does SUCK on my Mac waiting for redraws and beach ball updates and so on...............read carefully before you post....

OGL is not slow. Manipulating very high poly count meshes is. And that's not related directely to OGL, but to the software core.

Rotating the view of very high poly objects is smooth, both in Modeler and Layout.

I think something is wrong from the start in this thread. I'm not talking about what has been said, but HOW it has been said.

Get informed before you post, and read carefully all the posts before you do it.

dballesg
06-24-2007, 03:37 AM
I do not mind complains! :) I am a big complainer! :)

But if we keep saying that the team it is not doing a god job because they do not implement 1 million changes per second on the code maybe we will waste their infinite patience and they will take revenge for example implementing Blenders interface on the next LW! :)

I found modeler editing capabilities a bit slow as others noted. But I had other problems as well related to OPenGL. Like having a bunch of crossing strange polygons using LWCad curves an Polygons on two simultaneously active layers. But looks like latest build has corrected that.

Other problem I found with LWCad (even 2.5) it is hangs up on my machine a lot using Background layers on the SAME object to snap to points or curves.

So those are slowing me on a building I am doing.

But I would like as well a bit of information of what changes they're doing, only that, I think a communication from Jay on what they are doing on a much more frequent basis, will keep the users more happy, knowing which road their favourite application is taking.

They can even keep the communication on the OB forums only, so it is tied to the NDA an the competence will do not know what they are doing! :)

Best regards,
David

scenicdave
06-24-2007, 04:50 AM
As a small example, I had a 864k poly object I wanted to deform in places - and these were continous polygons, so no layering possible. When you hide polygons in Modeler they're still evaluated for selections and operations. So even if you hide 98% of your model, the remaining bit is still just as slow to work on, as the entire object.



Hey Neverko,
Why not cut the part of the model you're working on, paste into another layer, continue working, when you're done cut and paste back; merge points. If you have to do it all the time, make a quick point selection set to control the point merge each time.
Not perfect, but it works fine.
regards
Dave Tracey
Sydney

scenicdave
06-24-2007, 06:10 AM
I know there's tons of little workarounds for all these kinds of problems, but that's what we want to get rid of, yes? :) The eternal workarounds and patched up workflows have to go. And I'm sure they will. It just can't be soon enough.
Hi Neverko,
Now that's optimism!

I think though that workarounds will always be here. Space Shuttle with a tube of silicone on board is a workaround. 'Check the gate' after every shot is a workaround. Everytime you turn around to see what's behind you is a workaround.

Modeling won't ever be perfect; finding an *acceptable* workaround is quite normal.
regards
Dave Tracey
Sydney

Andyjaggy
06-24-2007, 08:14 AM
Man this thread needs to die already. First off I'm sorry if I seemed like I attacked you Dave. It just kind of irked me that you seemed to be telling the 50 or so people on this thread who were agreeing with me that the editing speed in modeler needed some serious help that they were all wrong and bad modelers because of this. But anyway.

As to your question I have several projects going on that are requiring high poly counts but the main one is I am building a model of ancient Jerusalem. As you can imagine a whole city starts to take alot of polys. I have split the city up into little chunks (seperate objects) and each of those chunks into layers but I am still constantly running into problems with some of the bigger and more detailed building just simply having too many polys to be manageable in Modeler.

Am I making it work with "workarounds" yes of coarse, I can find ways to work around it. Will there always be workaround, yeah of coarse. Would it be nice to have editing speed at least on par with most other major apps today. Absolutely. Which was the point of my original post, which I guess has gotten completely lost a this point.

SmEsh
06-24-2007, 09:39 AM
If you build the same city in Max you would probably:

1) split it in a grat numbers object (= LW layers); it's not important if it's only one max file (= LWS), I doubt that you would build the city as a single polymesh. And it's s GREAT difference, because if you load a 5 million polys city and edit a 12 polys column, you aren't editing 5 million polys, but 12. And if you move a 200K polys building, you aren't moving 200k polys, but only 1 pivot point.
Instead, if you load a city in modeler, you are editing all of its points, and if you move an object, you don't move its pivot, but all of its points. It's a shortcoming of the separation of modeler\layout.
Try joining all of the city in Max in one single polymesh, see if it's still superfast!(tested in v.8)
Try dividing your city in modeler in layers, then load it in layout, and move objects around, see if it's still superslow!
This is a reason LW needs modeler tools in layout.

2) hide unnecessary objects

2) use istances for trees etc... Lw needs istances...

Andyjaggy
06-24-2007, 09:51 AM
I'm not saying I would even want to put all the city in one object, that would be stupid for a variety of reasons. But I can't even work on some of the larger more detailed individual buildings without having to wait 10 seconds after every click I make.

I see what you are saying about the way Max handles object. It sounds interesting. I've noticed I can move large objects around in Layout fairly easily. Is this because of the way Layout converts things to a bounding box when you move them? If so would it be possible to have an option like that in Modeler?

SmEsh
06-24-2007, 10:04 AM
I'm not a programmer, but I think it's because when you move an object in in layout you're editing only the coordinates of its pivot poit (1 point), while the coordinates of the single points are inaltered (if you open the object in modeler is still at its place).
Instead if you move the object in modeler, LW is editing the coordinates of all of it's points, while the coordinates of the pivot are inaltered.

So, you will never be able to move an object in modeler the way you do in layout: in modeler the pivot of the object is always at 0:0:0 : you need to wait for modeler integrated in layout.

Then, Lw should surely be a lot faster at editing large meshes.

scenicdave
06-24-2007, 10:10 AM
Man this thread needs to die already. First off I'm sorry if I seemed like I attacked you Dave. It just kind of irked me that you seemed to be telling the 50 or so people on this thread who were agreeing with me that the editing speed in modeler needed some serious help that they were all wrong and bad modelers because of this. But anyway.

As to your question I have several projects going on that are requiring high poly counts but the main one is I am building a model of ancient Jerusalem. As you can imagine a whole city starts to take alot of polys. I have split the city up into little chunks (seperate objects) and each of those chunks into layers but I am still constantly running into problems with some of the bigger and more detailed building just simply having too many polys to be manageable in Modeler.

Am I making it work with "workarounds" yes of coarse, I can find ways to work around it. Will there always be workaround, yeah of coarse. Would it be nice to have editing speed at least on par with most other major apps today. Absolutely. Which was the point of my original post, which I guess has gotten completely lost a this point.
Hey Andy,
Jerusalem would be tough; all that rock, jaggy edges. Thatch. Ouch! (But at least the interiors would be out of shot.)

So, your modeling problem isn't actually a modeling problem at all.

I repeat, so that all can understand;;;;;;; Your problem here is not with Modeler. You are trying to solve the wrong problem.

The problem you need to solve is with instancing. Do you have HDInstance?

To build something like this you need to reuse as much geometry as you can and you need to have good texturing skills because you ..just.. ..can't.. ..built.. it ...all..!

So it becomes- build what you have to, use it as often as you can, texture the rest.(How are your texture painting skills?) For example; build a really good wall of rough rock, with breaks, debris at the bottom etc; instance that. Use that to build other buildings. Stick different instanced roofs on each.
(I once built Cheops pyramid with very detailed instanced blocks; stupendous polygon count, but it rendered in 1 gig of ram)

Also consider this; building something in it's entirety is somehow gratifying and feels complete but it's a total waste of time in 3d. Waste Of Time!!!
No-one is ever, ever going to see all the detail you put in simply because you will never get around to rendering all the angles and if you did it would be the most boring, dull monotonous piece imaginable.
Waste of your life I tells ya!!

Instead, spend a weekend and do an animatic. Then build to that. Every single shot will look better, render faster and be more exciting because of it. Take each shot and flesh it out. Building the whole world, the whole environment is *not* the way to do it.

For the record; LightWave is as good as Max, XSI, Maya for this task. Easily as good; maybe better because it's sooo much faster to set this sort of stuff up. They all have to instance, XSI gigapolycore included, and they all struggle.

You'll end up with hundreds of objects in Layout; use nulls as heirachies. Learn about UV clip maps. Learn about subtractive surfacing. Start new layers as soon as Modeler slows down for you. Stop bitching.

:-))
Happy to help anytime,
regards
Dave Tracey
Sydney

Andyjaggy
06-24-2007, 10:12 AM
That makes sense I guess. The Newtek staff sure has a job ahead of them :D Too bad I love everything else about Modeler so much. It's been a while since I used Max but I remember hating modeling things in Max. I've also heard Maya isn't much better. I guess it just depends what you like. I love the straight forward, almost raw approach that you take in Modeler. It's simple, and very fast if you know what you are doing.

SmEsh
06-24-2007, 10:27 AM
Unfortunately the raw approach is not the best when you work on complex projects, because it eat s your system resources very fast.
For example: you are modeling a temple with 24 very detailed columns of 1000 polys. In modeler you would clone one column 23 times= 23000 more polys. Instead in max you would istance the column 23 times= 0 more polys.
And if the client ask you to change a detail in the column? In Max you only have to change 1 column, in LW instead you need to delete all of the other columns and reclone them.
I also hated organic modeling in Max, but when it comes to architecture...

Lightwolf
06-24-2007, 10:41 AM
Unfortunately the raw approach is not the best when you work on complex projects, because it eat s your system resources very fast.
For example: you are modeling a temple with 24 very detailed columns of 1000 polys. In modeler you would clone one column 23 times= 23000 more polys. Instead in max you would istance the column 23 times= 0 more polys.

I guess that is why I lay out my objects in Layout, and not in Modeler. Of course then you can't model in context anymore...

Cheers,
Mike

AbnRanger
06-24-2007, 10:49 AM
Even though Max 9 is much faster than 8, I still separate most objects into layers...so I can hide or freeze certain one's quickly.
LW's layer panel is similar to Max's Layer Manager, so the workflow is not entirely different. It helps, regardless of which program we use, to isolate just the object(s) we are working on. That's not an attempt to be casual about the Mesh-editing weaknesses of Modeler.
It is markedly slower than Max (and Modo and XSI, I'm sure). I brought in a model from Max (for a dental animation) to 9.2, and even though I could tell a big difference between 9.0 and 9.2, it was still slower than tumbling around and editing in Max.

Nevertheless, I still think that having FPrime helps to make up for this annoyance. I'm confident Newtek will soon transform Modeler by the same measure they have with the renderer. If we don't have the patience, then Hexagon or Silo would be good stand-ins during this transition period, since they are relatively inexpensive yet powerful modelers.
To get you up and running fast, you can find a video training series for Hex at:
http://www.vtc.com/products/Hexagon-2-Basics-tutorials.htm

dballesg
06-24-2007, 11:26 AM
If you build the same city in Max you would probably:

1) split it in a grat numbers object (= LW layers); it's not important if it's only one max file (= LWS), I doubt that you would build the city as a single polymesh. And it's s GREAT difference, because if you load a 5 million polys city and edit a 12 polys column, you aren't editing 5 million polys, but 12. And if you move a 200K polys building, you aren't moving 200k polys, but only 1 pivot point.


Sorry to interrupt :)

But that "you are moving 1 pivot point" theory it is wrong.

When you have geometry on ANY 3D app. And you move it, you are applying a transformation matrix to EVERY point on that geometry, at least on the redrawing part of the application.

In Layout case, I think this it how it works, if your object it is under the value indicated by Bounding Box Threshold, LW will use the Bounding Box for your object so it is applying this transformation matrix to the pivot point + 8 points on the box, and redrawing that box on each redraw pass of the application interface. That is why you have the "sensation" it is moving fast. You are "temporally" only moving 8 points. NOT 1.

But when you stop moving the object, Layout need to apply the same transformation matrix to EVERY point of your geometry so it will able to redraw your "full" object again on its new location.

So the speed you really will have depends a LOT on how the original team coded those transformation matrix and applied them to the geometry.

It can be done with a simple loop going point by point applying same transformation matrix to each one. I left someone with better math knowledge than me explain that! :)

And I shortened my explanation without take in account world and screen coordinates transformations. If you choose a different coordinate space (left or right) than you main display library (read here OpenGL or Direct X) you can even slow down things more.

And if you look to the OLD examples on the SDK you will see that the "old" dev team was not very carefull designing OR optimizing things.

About HD instance it is a good plugin, but it is a "hack" (a real clever one) using the volumetrics in LW. Almost implementing it's own raytracing routines to reuse geometry data to render the "instanced" objects.

A real instancing system will use the original "pointers" in the memory of the application to "reuse" the geometric data of your original object when it is draw or rendered. So has the advantage of reduce the memory consumption during those tasks.

Best regards,
David

Lightwolf
06-24-2007, 11:43 AM
Sorry to interrupt :)

But that "you are moving 1 pivot point" theory it is wrong.

When you have geometry on ANY 3D app. And you move it, you are applying a transformation matrix to EVERY point on that geometry, at least on the redrawing part of the application.

But the question is... where and when does the transformation of the points happen.
If the geometry is on the graphics board, you do only compute a new transformation for it... everything else is handled by the GPU.
Also, when you render, only then are the points transformed.

Obviously, if you add deformations, the whole thing will break down and get slow... as can be witnessed in Layout.

Cheers,
Mike

dballesg
06-24-2007, 12:07 PM
But the question is... where and when does the transformation of the points happen.
If the geometry is on the graphics board, you do only compute a new transformation for it... everything else is handled by the GPU.
Also, when you render, only then are the points transformed.

Obviously, if you add deformations, the whole thing will break down and get slow... as can be witnessed in Layout.

Cheers,
Mike

Ha, my master to the rescue! :)

You are right, but I wanted to keep my explanation simple as I could! :)

And of course depends if you are using double buffer techniques for example as well! :)

That is why when the "new" team say that they are doing "under the hood" changes, many of us we do not "understand" the "mayor" implications of that sentence! :)

jat
06-24-2007, 01:45 PM
[QUOTE=lino.grandi]OGL is not slow. Manipulating very high poly count meshes is. And that's not related directely to OGL, but to the software core.

Do you use a Mac? If not then listen........the OGL SUCKS banner I am using here is to speak of the implementation of this set of functions on the Mac and how Newtek along with Apple haven't done nearly enough to make it work as it should. Perhaps the call functions have to be refined and that's slowing it down, maybe the draw methods need to be refined along with the optimized use of vertex arrays, or vertex buffer objects. Maybe there are new and better ways to send vertex data for the command buffer....they know what the bottleneck is and it just needs to get fixed. BUT WHEN? WHEN?


Rotating the view of very high poly objects is smooth, both in Modeler and Layout.

:eek: NO, IT ISN'T!......maybe for you but not for me and many other Mac users like me. Do you consider a 166K model to be dense? I don't and yet it drags like an old dog. Don't make general statements unless you know what your talking about. And if you do have a Mac and LW is lightning quick then you are lucky....VERY lucky and I want to know what you're doing different. I will reserve my final judgements on when the Universal update is complete but for now I am not impressed. I do however think Chilton is doing a fantastic job in trying to address our concerns. And my machine is a dual core G5 with a good card and plenty of ram so it's not exactly outdated. And that's why you can understand that so many people are frustrated.

jat
06-24-2007, 01:49 PM
CHILTON SAVE US............lol

SmEsh
06-24-2007, 02:31 PM
But the question is... where and when does the transformation of the points happen.
If the geometry is on the graphics board, you do only compute a new transformation for it... everything else is handled by the GPU.
Also, when you render, only then are the points transformed.


But my question is:
if select 100k points and try to move them, LW is terribly slow, ... OK, I can understand it...
but if select only 1 point of these 100K, LW is as slow at moving it as I select them all!! Why?
Does it means that actually, every time you apply a trasformation in modeler, it recompute the position of all the points of the active mesh? Does it loop them all?

Call it, curiosity... I only studied some python, so maybe I am not able to understand, mine are only speculations...

Lightwolf
06-24-2007, 02:47 PM
But my question is:
if select 100k points and try to move them, LW is terribly slow, ... OK, I can understand it...
but if select only 1 point of these 100K, LW is as slow at moving it as I select them all!! Why?
Does it means that actually, every time you apply a trasformation in modeler, it recompute the position of all the points of the active mesh? Does it loop them all?
That is very likely... depending on the internal data structures. Which is what I meant earlier, if they haven't been designed with openGL in mind (which I doubt, openGL no option when Modeler started out - and has also changed a lot during the past few years) then they do need to rework the core.
And if we're really lucky, that will mean breaking a lot of plugins during the process as well...

Cheers,
Mike

jat
06-24-2007, 02:51 PM
Lightwolf

just wanted to ask your perspective on how close you think Newtek is to getting this issue nailed once and for all.....

jameswillmott
06-24-2007, 04:43 PM
Don't forget that when Modeler was designed, the polycounts we have today were a thing of fantasy. No one would have thought, 'Let's design this to push 1million+ polygons around'...

Lightwolf
06-24-2007, 05:00 PM
just wanted to ask your perspective on how close you think Newtek is to getting this issue nailed once and for all.....
Once and for all? Never, since the hardware and APIs change, there will be a faster way to do something in 5, 10 or 15 years time ;)

Quite frankly, I haven't the foggiest. I might know LW from the outside through the SDK, but that gives little insight in places, especially if the dev team is changing parts without making it obvious through the SDK.

Cheers,
Mike

Lightwolf
06-24-2007, 05:05 PM
Don't forget that when Modeler was designed, the polycounts we have today were a thing of fantasy. No one would have thought, 'Let's design this to push 1million+ polygons around'...
Not only that, but the old 5.6 Modeler, which didn't support openGL, was faster than the early openGL boards (at least in wireframe mode).
And it had one feature that you can not implement with openGL, that was interrupteable drawing.
Now, if you decide to move, pan, however modify your view, the code has to wait for openGL to finish drawing the current mesh (it might be possible to queue the next transformation, but the current drawing process still has to complete on the graphics hardware).
Back then, you could move the mesh any time during a redraw (which was completely software driven). You might not have seen everything (I think Modeler painted points first, then the connecting lines) - but you could navigate a large mesh quite quickly. It took a lot of years after 6.0 introduced openGL in Modeler to get that kind of interactivity back (due to the hardware not being up to it).

Cheers,
Mike

Stooch
06-24-2007, 06:43 PM
Andy, no offense, but that cat can hardly walk let alone mount an attack. it looks like it wishes it could attack someone :D
As opposed to oDDity who attacks constantly ;)

maybe andys cat is so fat it needs 3,000,000 polygons.

Andyjaggy
06-24-2007, 08:14 PM
Are you calling my cat fat!? You can say whatever you want about me, call me a baby for complaining about something I think isn't up to par in LW, call me a lazy modeler, tell me I don't know what I am doing, but I'll be d#$med if I sit around and let you call my cat fat. :D

Anyways I'm done with this thread. I've sat around bitching long enough, time to get back to work and figure out a way to make it happen.

AbnRanger
06-25-2007, 01:18 AM
Are you calling my cat fat!? You can say whatever you want about me, call me a baby for complaining about something I think isn't up to par in LW, call me a lazy modeler, tell me I don't know what I am doing, but I'll be d#$med if I sit around and let you call my cat fat. :D

Anyways I'm done with this thread. I've sat around bitching long enough, time to get back to work and figure out a way to make it happen.Say man, that is oooone fat cat.:D

jin choung
06-25-2007, 03:15 AM
Don't forget that when Modeler was designed, the polycounts we have today were a thing of fantasy. No one would have thought, 'Let's design this to push 1million+ polygons around'...

hmmm....

errrr....

ummmm....

uuuuu....

hahahahaha...

gotta say... if this has any relevance to what we're talking about, in any way whatsoever, it is HIGHLY distressing!

jin

jameswillmott
06-25-2007, 05:37 AM
It's only speculation.

mgcross
06-25-2007, 09:38 AM
I thought openGL was in modeler since 5.0? I was using a permedia-based 8MB FireGL card for openGL back then, lol. Maybe I'm mistaken, but I know that at least Layout had openGL in 5.0.


Not only that, but the old 5.6 Modeler, which didn't support openGL,

<snip>

Cheers,
Mike

Lightwolf
06-25-2007, 09:40 AM
I thought openGL was in modeler since 5.0? I was using a permedia-based 8MB FireGL card for openGL back then, lol. Maybe I'm mistaken, but I know that at least Layout had openGL in 5.0.
Actually, that may well be... I do remember an openGL viewer plugin for 4.0...

Hey, I'm getting old :confused:

Cheers,
Mike

mgcross
06-25-2007, 09:57 AM
I know exactly what you're talking about though - manipulating points in a large mesh felt much more interactive pre-6.0. And to keep on-topic, I just updated to 9.2 and the openGL does feel more 'laggy' than I thought it would.


Actually, that may well be... I do remember an openGL viewer plugin for 4.0...

Hey, I'm getting old :confused:

Cheers,
Mike

Scazzino
06-25-2007, 11:35 AM
:eek: NO, IT ISN'T!......maybe for you but not for me and many other Mac users like me.

Just to check. Make sure you're not trying to use GLSL... That has never worked for me on the Mac other than to turn everything into molasses. I get pretty decent rotations of my stuff when OpenGL is set to multitexturing and VBO. Editing or deforming the mesh however is still very sluggish, but as has been mentioned a number of times before, that's not an OpenGL issue.

mkiii
06-28-2007, 08:39 PM
Well that *was* an interesting read... I go away for a few months & when I come back....

...Everything is exactly the same ;)

Honestly. These same topics crop up every year, & for every version. Expectations always outstrip the capabilities of the app for some people.

IMO Modeler has had its day. Once it was the fastest thing since sh1t first slid of a shovel, now, you can get better performance out of a $100 app like Silo - and have pretty neat displacement painting built in.

Maybe one day (V20?) they will just quietly drop it after having moved all the modeler tools over to Layout...... Then the rose tinted brigade will have to take notice - because their precious layout will be just as slow as modeler.

Seriously though.... Newtek want more game devs to use LW... Well they better get used to the idea of shifting 3million poly objects around, because Next-Gen titles are going to need it. Until then, Max still has a stranglehold, and many more have taken up Maya & XSI. I have been looking for a new job for a little while now, and guess how many LW related modelling jobs I have seen in Games?

still.. it might all be fixed in 9.3 eh? :hey:

Ivan D. Young
06-28-2007, 10:33 PM
One thing that I find interesting is that some of the next big game engines are doing things in real time that LW struggles with. Has anyone looked at the next ID Software MegaTexture stuff? A 20 GB texture that runs in real time. LW can not even load such a texture. If that game engine does DOF and Motionblur, you could do Space ship stuff right out of a game enigne. Why could'nt something like the Battlestar style stuff minus effects not be run on a game engine like that in real time? GAme engines already run at higher resolutions than 1080p now. IF anyone has been watching the Crytek stuff the way that they can occlude stuff and keep framerates high is pretty amazing. When a game engine can do subpatch geometry and the game editor will allow for geometry maipulation like modeler, many 3D apps will be introuble possibly.

BeeVee
06-29-2007, 06:29 AM
Umm, with Infinimap (http://www.infinimap.com/) 20GB would be small beer.

Ivan D. Young
06-29-2007, 09:25 AM
yes that is true, but with the ID software engine you would still be talking about something that is running in real time and possibly as high as 2048 by 1600 something in resolution. and maybe with DOF and Motion blur.

Jarno
06-29-2007, 10:28 PM
Just about anything can be made realtime if you bake it enough.

---JvdL---

jin choung
06-30-2007, 03:56 AM
baking is next to godliness... that is true.

but a 20gig texture is a 20gig texture....

BUT, games have always done things in terms of real time rendering that software apps like maya, max or lw could not.

even now, not a lot of development software incorporates all the rt render features in top of the line games.

part of the reason that games can do it is that they are very SPECIFICALLY DESIGNED while development software has to be more general purpose.

things like aggressive binary space partitioning and occlusion culling that goes on with games may not be practical for someone who's actually WORKING the geometry... not just looking at it.

having said that - yes, there probably IS a lot of features we can take from gaming.

ILM incorporated rendering optimizations from a book on game graphics in order to not render more than is visible in a pixel for pirates of the carribean.

thing is, we have to KNOW about it in order to do it and i would imagine that newtek has a tough time (as anyone would) of keeping up on all the tech used in not only games but other competing apps.

jin

Stooch
06-30-2007, 04:49 AM
realtime rendering is all about working as efficiently as you can with one arm tied behind your back. the reason why its useable in games is because the graphical standard as to what "looks good" is merely passable in full blown cg. so i really dont see the realtime engines as a "bonus" - i consider it a handicap. infact its the real time engines that are trailing the full CG rendering innovations by coming up with shortcuts that always sacrifice quality for the sake of speed.

Ivan D. Young
07-02-2007, 12:20 PM
The thing to remember is that in some areas Game Egines are more capable than 3D apps. At some point Game Engines will get tools that are in 3D apps and the line will be blurred even more. What is stopping someone from licensing one of these new engines and building on, or modifying tools to add more functionallity. I realize that is not probable in the very near future, but there is no reason that this could not be done. And before a lot of people say this is not likely, remember going forward the multi processor thing is only going to get more and more. Next year Intel releases 8 cores, game engines will be utilizing this power faster than 3D apps will (except for rendering)and that trend does not look like it will change for awhile.

Stooch
07-02-2007, 08:25 PM
i havent seen a game that comes even close to a standard software render. dont get me wrong there are plenty of nice render engines out there. but they are not even in the same league. as far as leveraging thier power in a production? fat chance. whatever time you save with realtime rendering, will be eaten up by you setting the scene up to be realtime in the first place. and it wont be as detailed. its a pipe dream imo. will be for the next decade or so.

theo
07-02-2007, 08:54 PM
i havent seen a game that comes even close to a standard software render. dont get me wrong there are plenty of nice render engines out there. but they are not even in the same league. as far as leveraging thier power in a production? fat chance. whatever time you save with realtime rendering, will be eaten up by you setting the scene up to be realtime in the first place. and it wont be as detailed. its a pipe dream imo. will be for the next decade or so.

I was planning on posting a similar perspective- well, this isn't going to happen now since creative beings tend to revile repetition... curses! May the fleas of twelve hundred camels mercilessly suck every drop of blood from your zealous body.

Um, that COULD qualify as a slight over-reaction... anyways....

Realtime rendering works great with games because it is a one-off, quite final visual product, per rendered frame, that is. Problem is, this isn't ideal for compositing environments where the creative villagers tend to require much more data to manipulate than a simple image sequence.

jin choung
07-02-2007, 09:19 PM
doesn't have to be real time to leverage its speed... it's not all or nothing you know.

rendering something at a frame every 20 seconds through the gpu would still offer lots of speed benefits.

and gpus as of dx10 compatible cards are basically general purpose processors with awesome floating point performance and fat, optimized pipes with fast fast ram.

not a pipe dream.

and it's unfortunate that lw still doesn't have a rendergl that leverages the power on current vidcards.

jin