PDA

View Full Version : Lightwave rendering engine



daros
07-21-2005, 04:14 AM
hi, I'm David Rossmann from Stack Studios.
I'm Lightwave user since 1995. I relized about 1000 projects and 4 hours architectural animation. Our clients are Zaha Hadid, Daniel Libeskind, Arata Isozaki, UN, Pirelli, Fiat, DeBeers, and many others.
Lightwave is a fantastic program, easy to use, high rendering quality, stable...
But now you can reach lghtwaves quality with many, many different rendering engines at 1/5 of the rendering time. In my company we are 3 Lightwave users, and a total of 180 3Ghz CPU's.
That means 60 Cpu per user? yes.
Why?
Because lightwave's rendering engine is slow and we have almost 6 to 16 hours frames. This is to much for a PAL frame.

The rendering engine needs to be revrited completly or Lightwave will disappear. The rendering engine of lightwave is almost the same since 1996.

Best regards

www.stack-studios.com
www.arkadin.biz
www.animax.it/works

mattclary
07-21-2005, 05:22 AM
You should be clearer in what you post. It took me several reads and some digging to actually figure out what you are advertising.

trick
07-21-2005, 05:52 AM
You should be clearer in what you post. It took me several reads and some digging to actually figure out what you are advertising.


I think it is VERY clear what he is talking about. And because this is his first post after 10 yrs using LW, I do NOT think he is here for advertising. I'd rather see someone complaining about LW with a 1000 projects record and a lot of references, then someone with a 2672 post record and NO references...

coremi
07-21-2005, 06:42 AM
Hopefully in about 2 weeks will found out what's next for Lightwave 3D. Maybe Newtek will improve render, now they hired that guy from Electric Image, and maybe will get the most expected openSDK so will have Fprime and Kray and maybe some other great renders working for Lightwave too.

Kurtis said that all will be revealed at Siggraph so hopefully, now that Luxology will come forward showing to the users what they have in mind, maybe Newtek will show some kind of a roadmap, so will know what to expect in the future, it is time to do that, they have a good team, stable, so they will have to give a sneak peak of what's next, so users will keep faith in a much better product to come.

MooseDog
07-21-2005, 07:25 AM
that's a very impressive collection of websites and firms daros. congratulations

agreed that 6 hrs a frame is entirely too much. with that much talent, surely you guys have figured out ways to reduce that time. passes, light set-ups, baking, etc etc.

agreed as well that the renderer needs speed improvements. i suspect it's coming, let's hope it arrives w/o sacrificing that special "lightwaveness" to the renders that does seperate them from the crowd.

digital verve
07-21-2005, 09:07 AM
Someone from Newtek (Chuck?) did post a while back that the renderer was being worked on. So maybe this Siggraph we will see the fruits of their labour. ;)

Edit: Daros, you have a great portfolio. Very nice work.

JML
07-21-2005, 10:52 AM
improving the rendering engine would be good of course, but I think a lot comes from scene optimization from the user too..

mattclary
07-21-2005, 10:53 AM
I do NOT think he is here for advertising.

http://www.arkadin.biz/

Then you should look here and ask yourself what "Remote Matador" is and who is making money from it.

Lightwolf
07-21-2005, 10:59 AM
http://www.arkadin.biz/

Then you should look here and ask yourself what "Remote Matador" is and who is making money from it.
So, that would mean that everyone who has a sig pointing to a (possibly) commercial service or tool is advertising? Do you also imply that is the only reason commercial users post here?
In that sense, every freelancer showing his work is advertising as well.

I don't really see any advertising in daros' post, I do however see that he and his company have done a lot of work using LW and developed for it as well. Heck, I didn't know Matador got that far in the past few years, very cool. Still not advertising though.

Cheers,
Mike - who ain't changin' his sig

mattclary
07-21-2005, 11:08 AM
I guess I should have just said, "Thanks for pointing out the obvious". Yeah, it's slow, thanks for cluing Newtek in, I don't think they or anyone else here knew that. :twak:

JML
07-21-2005, 11:10 AM
http://www.arkadin.biz/

Then you should look here and ask yourself what "Remote Matador" is and who is making money from it.

yeah, it looks like advertising to me too.

JML
07-21-2005, 11:12 AM
the thing I don't understand, why is it supposed to be a super fast rendering engine when it's powered by maxwell render ? ;)

mattclary
07-21-2005, 11:29 AM
So, that would mean that everyone who has a sig pointing to a (possibly) commercial service or tool is advertising? Do you also imply that is the only reason commercial users post here?
In that sense, every freelancer showing his work is advertising as well.

I don't really see any advertising in daros' post, I do however see that he and his company have done a lot of work using LW and developed for it as well. Heck, I didn't know Matador got that far in the past few years, very cool. Still not advertising though.

Cheers,
Mike - who ain't changin' his sig

I have absolutely no problem with anyone who advertises here concerning LightWave or 3D related things. I just get irritated when it's thinly veiled as something else. Maybe the guy isn't advertising, in which case see my last post. ;)

trick
07-21-2005, 11:45 AM
...Then you should look here and ask yourself what "Remote Matador" is and who is making money from it.

Don't you think it would be a better strategic move to put a more clear advertisement under the "third party" section, instead of telling people:

"hy, you guys with that awfully slow render engine, just click on the following links, read all about it and you will find a huge renderfarm which can help you LW-fanatics keep using your favorite software for the next couple of years..."

People should not try moderating posts if they are not moderators !!!

mattclary
07-21-2005, 11:52 AM
Don't you think it would be a better strategic move to put a more clear advertisement under the "third party" section, instead of telling people:

"hy, you guys with that awfully slow render engine, just click on the following links, read all about it and you will find a huge renderfarm which can help you LW-fanatics keep using your favorite software for the next couple of years..."

People should not try moderating posts if they are not moderators !!!

THUS my response. I'm chalking it up to his not being a native English speaker. :thumbsup:

trick
07-21-2005, 11:56 AM
I guess I should have just said, "Thanks for pointing out the obvious". Yeah, it's slow, thanks for cluing Newtek in, I don't think they or anyone else here knew that. :twak:

I don't think this is about "pointing out the obvious". This is all about experienced users are seriously looking for alternatives: and for architectural viz there are a lot of these, and they are all getting seriously better and faster. I know a lot of people in this business (all using LW ATM) which are all on the edge of using other solution or are already using them. Of course this may be obvious to you too, but that surely does not mean it should not be mentioned...

Tyler007
07-21-2005, 11:58 AM
I agree NewTek defiantly needs to make some optimization in the render engine. I remember the slow down in performance when they went to version 6. Also a lot more bugs worked its way into the render and many have not been addressed.

Matt
07-21-2005, 12:46 PM
So, that would mean that everyone who has a sig pointing to a (possibly) commercial service or tool is advertising? ...
Cheers,
Mike - who ain't changin' his sig

Such a shameless plug Mike, how dare you! ;) Now go and stand in the corner you bad lad!!!

:D

Lightwolf
07-21-2005, 12:50 PM
the thing I don't understand, why is it supposed to be a super fast rendering engine when it's powered by maxwell render ? ;)
Well, years ago Matador was a (a bit cludgy but workeable) single frame distributing render manager for LW. It would basically render tiles on remote nodes and assemble them.
It seems that they souped up the front end, made it stand alone and it now allows you to render on a remote renderfarm using maxwell. Quite a neat concept actually, a shame they didn't stick to lwsn though.

Then again, I still don't think advertising this service was the intent of the post.

For me his main point is:


But now you can reach lghtwaves quality with many, many different rendering engines at 1/5 of the rendering time. In my company we are 3 Lightwave users, and a total of 180 3Ghz CPU's.
That means 60 Cpu per user? yes.
Why?
Because lightwave's rendering engine is slow and we have almost 6 to 16 hours frames. This is to much for a PAL frame.

And, as others pointed out, it never hurts to tell your fav software vendor that there is competition that is becoming a feasible alternative.

Cheers,
Mike

Lightwolf
07-21-2005, 12:52 PM
Such a shameless plug Mike, how dare you! ;) Now go and stand in the corner you bad lad!!!
:D
:lol: You're not doing much better my friend, are you? :twak:

Cheers,
Mike :bowdown:

Chuck
07-21-2005, 01:03 PM
I think we're clear that this is just a concerned LightWave user expressing a request for improved performance in an area that affects the overwhelming majority of users. As has been mentioned, we brought Mark Granger, author of EI's Camera, on board earlier this year. Camera was known for speed and ability to handle high poly counts, and Mark and the team are of course devoting a great deal of attention to renderer improvements.

Intuition
07-21-2005, 01:04 PM
I've been using Maxwell Render and F-Prime all the time for about a year now.

I still bust out Lightwave's native tracer for things like Hypervoxels and some basic stuff I can comp in Digital Fusion or After effects.

I can't say I've really used Lightwave's main renderer seriously for some time now. With all the better radiosity solutions available I am discouraged from using LW's.

I believe newtek wont let us down though. They really pushed the envelope with the earliest radiosity solution. I think that they will come up with something great again.

Oh, BTW Wolf, I've been back at EdenFX for a little bit. Still utilizing your infinimap but I am stuck trying to convert the images I have to ECW cause they are over 500MB in size. I may have to buy some ECW software.

The animation is neat so far though.

Lightwolf
07-21-2005, 01:48 PM
Oh, BTW Wolf, I've been back at EdenFX for a little bit. Still utilizing your infinimap but I am stuck trying to convert the images I have to ECW cause they are over 500MB in size. I may have to buy some ECW software.

The animation is neat so far though.
Nice to hear (darn, this is turning into an ad now... sorry for that...).

Cheers,
Mike - intuition, I'll PM you ;)

Pavlov
07-21-2005, 02:59 PM
I've been using Maxwell Render and F-Prime all the time for about a year now..

Same here too, i'm sticking to Fprime and Kray. I almost never use LW's native renderer unless i've baked GI with Kray before, so i can use MB and other tools.
My group is not as heavy-sized as Stack studios, but i'm an old LWer, i did hundred projects and a lot of viz movies too: i confirm each single word of original post.
I'm sure something render-wise will happen; hope it wont be a "mock-up" or some added features but more a brand new concept engine, with implementation of actual rendering features. Volumetric technologies have gone far beyond HV; Good GI means Photon Mapping. A fresher/deeper lighting toolset, subpixel displacement, IES, are also needed.
It's strongly adviced not just to improve Montecarlo performance, but to implement also alternative GI methods, like photon mapping (Kray integration?)or/and Lightscape recursive tessellation.
Last, give 3rd parties a full access to each glitch of rendering pipeline, let us land on a continent instead of an island.

BTW, good work.

Paolo Zambrini

Vision1
07-21-2005, 03:00 PM
yep i think render speed is very important
but ik think the quality is even more important

look here http://www.zaon.com/company/articles/3d_rendering.php

there can you see that the Lightwave's native built-in renderer is the only native renderer actually suitable for production film work

so when newtek make's a better render you must keep that in mind !!!!!

JML
07-21-2005, 05:54 PM
yep i think render speed is very important
but ik think the quality is even more important
...
so when newtek make's a better render you must keep that in mind !!!!!

YES :thumbsup:

Muad'dib
07-21-2005, 07:56 PM
I've been aware of Stack Studios and their quality work for a long time now and I'm with Chuck & Lightwolf on this one. I see Davids post as a point of concern and a call on NT to raise the bar once again. In terms of the links - I don't see them as any sort of advertising, to me they just show what is possible from third party / outside solutions and the quality clients/people now expect when presented with still renders and animations. These are world class Architects / clients Stack has shown to be dealing with here so I can totaly understand the point David makes.

We have had LW quality and accuracy for a while but now others have them as well as speed to boot (allbeit at the sacrifice in accuracy). Quite simply ... we need solutions to these issues or in the least a glimpse of the way ahead and what we can / should look forward to.

Emmanuel
07-22-2005, 09:40 AM
A 400% speed boost would be fine :)

daros
07-22-2005, 12:08 PM
Sorry, i wasn't smart enough to suppose that the first answeare to my post be from a LightWave Loyalist.
No this was not an adwertising post.
All the works from the animax web page are made with Lightwave. I repeat, Lightwave is fantastic and i'm simply try to explain the risk to see disappear this software and that only because the rendering engine.
The best Modeler for polymodeling, the very good and simple Layout, good plug-ins, but a old render engine.
Since every user of Lightwave uses Lightwave to finally heve a rendering this last step is the most important. I'm too know many LW users that are looking around for other Render solutions.
No need, for mee, to go hire electricimage programmers, since many young programmers around the world have made better rendering engines a Electricimage one.
for example
http://rise.sourceforge.net (this is not an advertise since it's a free software)
sorry for my bad english (i'm not a native english speaker).

David Rossmann

pixym
07-22-2005, 12:44 PM
Dear David Rossman,
I clap my hands to your reply.

I am one those who are looking for an another solution (Max+Vray for example) to render my architectural scenes

doimus
07-22-2005, 01:20 PM
OK, we all generally agree LW needs (much!) faster rendering engine, but some of the posts here seem like it's 2003 all over again... haven't you guys noticed FPrime at all.... neat little thingy that is. :p

Or perhaps KRay ....

Or Microvawe baker for that matter...


:lwicon:

pixym
07-22-2005, 01:28 PM
OK, we all generally agree LW needs (much!) faster rendering engine, but some of the posts here seem like it's 2003 all over again... haven't you guys noticed FPrime at all.... neat little thingy that is. :p

Or perhaps KRay ....

Or Microvawe baker for that matter...


:lwicon:

I currently use FPrime for still... that can be render with my amount of memory. Fprime is ram consuming
Kray, mmm it is promising but no multithread support and I have no scene that can be render with it (more than 1 100 000 polygons)
Microwave Baker, mmm I do not stay in front of my conputer to bake, too much probability to have to change something at the last minute for the final rendering...

Chuck
07-22-2005, 01:31 PM
No need, for mee, to go hire electricimage programmers, since many young programmers around the world have made better rendering engines a Electricimage one.


David, having programmed one state-of-the-art renderer already doesn't turn a good creative analytical mind into mush incapable of creating another state of the art product based on the latest technologies. I'm sure that there are younger folks out there doing what you do, and I'm sure you'd agree that this doesn't mean you should be put on the shelf with such a dismissive comment as you've made about Mr. Granger. I'm sure in your field that you feel your experience, skill and expertise mean something and have value to the customer. The same applies in programming.

Camera was an effort that he did some years ago and it was state of the art for the technologies current at that time, just as LightWave's renderer was state-of-the-art, in very different ways. There is, naturally enough, much new art and science today in this field, and every reason to understand that an experienced hand at creating a professional rendering system will be able to create advanced rendering capabilities based on this new art and science, and mold them into professional level tools as only the skill and knowledge gained by previous experience can do.

I'm certain that you understand this concept with reference to your own profession, and I would hope that on reconsideration you'll agree that it applies to other professions as well, including programming.

Emmanuel
07-22-2005, 03:34 PM
I would say I agree 100% with that.
It was for a reason Camera was so good back then, when there was competing products, and I think that people who have that kind of experience will probably use it to adapt new technologies in an intelligent manner.
Experience is much more important than youth.
I for one am not such a fond fan of third party solutions, I haven't even tried Kray, Maxwell or Fprime.
I need my Hypervoxels, shaders and volumetrics, and I love LWs rendering engine.
If NT makes it better and faster, to **** with anything else :)

wacom
07-22-2005, 03:36 PM
yep i think render speed is very important
but ik think the quality is even more important

look here http://www.zaon.com/company/articles/3d_rendering.php

there can you see that the Lightwave's native built-in renderer is the only native renderer actually suitable for production film work

so when newtek make's a better render you must keep that in mind !!!!!

Um...yeah if you send it off to a farm or have a large farm at home. Some would say that there are other renderers that are better suited for a small farm or simple two core system. I don't do production, but motion graphics, where more than 1min a frame is going to kill me- so LW for GI isn't really cutting it. Still I send stuff to farms... Fprime works to some degree, but the way it produces noise is a problem. That artical is also fairly old now...and some renders have gone "native" that can do quite a bit on just ONE machine. Such as MR...

Lightwave's saving grace is its free render nodes...

romrom74
07-22-2005, 04:37 PM
hi all,
interesting thread indeed...makes me want to share my experiences...
I've been using lw for many years (since 5.0 actually) and I recently realized that the only thing that changed my workflow in the past ~2 years is frprime...I even shortcut my f9 key to the frprime renderer. all the project Ive done since frpime was out, are strictly fprime rendered. I havn't touched lw F9 or F10 renderer for 18 months or so...so for me , no more HVs, fancy shaders or things like that, i can't afford the time to a frame to get rendered anymore ! At the begining I thought I couldn't make "good" images without the help of all the aditionnal technologies available through third-party shaders (sss, brdf, anisotropy, occlusion, displacement, volumetrics...etc) but with fprime I rediscoevred art of simulating the reality with simply modeling-texturing-lighting.
In a sense i agree with with daros: lw renderer is obsolete . But fortunatly other renderers are available. I own maxwell as well for 2 months but guess what :
i didn't use it for the last project i've realized (3 high res architectural viz).too unstable..to many crashes..to many constraints..I had 4 days to do the job, entierly with modeling-texturing-lighting etc..., a complete block with dozens of buidings and so on...at the end I had only one night left to get my 3 frames rendered (at a 4k res)...with montecarlo radiosity, raytraced reflections I couldn't afford waiting for a purely and "shader'd" lightave render . so i've fprim'd the scene and everybody was happy (including me and client)...So for now all I really need is a FAST and STABLE GI renderer .
so what's really the point here...is lw renderer outdated and "not suitable for nowadays production" or worst, (and I mean it) has NT left this part of the software behind for years ?
I personally think that both are true but not fatally mean the death of lw... i consider myself as a lw "loyalist" but it' s getting harder and harder these days to convince other people (max and maya users, at teh moment I work for a company full of this kinf of people...) that lw is still a reliable soft "suitable" for production works. Anyway i'm still confident in the futur of Lw and hope this EI guy will lead the developper team to an astonishing renderer...hopefully not to late. Nt, we poor lonely users need to be amazed, please, show us a part of our futur...We (I) need to actually see that things are going in the right direction. I can't afford trying to improve myself in the rendering field on an obsolete basis. What good are all the old lighting-texturing tricks if anybody can do that with the brand new "push-button" renderer in no time. We (I) need LW renderer to evolve and still exist so all my experience is still valuable! I need to be able to trick and fake my images the way Ive learnt to, but f - a - s - t - e - r.... I think Sigraph is a perfect timing to tease us with smthing...
hmmh enough blabla...need to get back to work.
rom

daros
07-22-2005, 04:41 PM
If you think Mr. Granger is the solution, ok...you are the NewTek Marketing Team.
I'm only your customer.

The last time i visited the EI gallery was in 1995 and i have such a bad impression about the rendering engine that the first time i visited it again was 10 minutes ago only because you have insisted about Mr. Granger.
I think the next time i visit the EI gallery will be 2015.

Best Regards

David Rossmann

Chuck
07-22-2005, 05:12 PM
If you think Mr. Granger is the solution, ok...you are the NewTek Marketing Team.
I'm only your customer.

The last time i visited the EI gallery was in 1995 and i have such a bad impression about the rendering engine that the first time i visited it again was 10 minutes ago only because you have insisted about Mr. Granger.
I think the next time i visit the EI gallery will be 2015.

Best Regards

David Rossmann


Artists create gallery entries, and we're very privileged that LightWave's renderer has seen use by so many talented artists. Any programmer, regardless of background, who works on the LightWave renderer is certainly going to have as a goal to insure that artists will continue to be able to make the kind of amazing images that fill our sample gallery, that any image rendered before could be rendered again identically, and that even better imagery is possible - pristine photo-realism or sublime fantasy perfectly matching whatever the artist has been able to imagine. Our development team understands that requirement and will abide by it. Please rest assured that no one on the team wants to turn LightWave 3D into a recreation of some other, older product.

My apologies if you feel that I've overstepped what is proper for a company employee talking to a customer. That certainly wasn't my intent - you've asked for improvements in the product and I've tried to reassure that those are coming. I've also tried to reassure you that your skepticism that the people our management has chosen can do the job is unfounded, and that the product will develop as you need it to. If you can't agree with that point of view, I hope you will at least be willing to reserve judgement and see the results of the work underway, and continue with LightWave if you find that it does indeed continue to meet your needs with our planned improvements.

We appreciate the concerns you and others have expressed, and are doing our best to address them.

daros
07-22-2005, 05:27 PM
Fantastic! but please not with this AA tricks... :)
I trust your guys.

Best regrards

David Rossmann

toby
07-22-2005, 06:04 PM
If you want to judge EI, watch Star Wars Episode 1, not the gallery. I've seen some questionable things done in LW too. Also see Titan AE. The giant crystals at the end are awesome.

I'm pretty sure that EI's anti-aliasing and motion blur is better than LW, they had problems with LW during Star Trek Nemesis and Serenity.

However, as far as workflow goes, EI is a royal pain to use and is much more limited feature-wise. Poor preview options, few keyboard shortcuts (uneditable), it's what you'd expect from a 3D program from the early 90's.

Looking forward to what they can come up with for Lightwave.

daros
07-22-2005, 06:41 PM
With Motion Blur and AA i agree. We need to use always AA Enhaced Extreme and some times is not enough.
Nice your "confirmation" link.

Ade
07-22-2005, 09:26 PM
Can I say you guys all have bought up some really great points, so has Chuck on the direction of Mr Granger and his talents and possibilities.

May I give my opinion now.
Ok judging by what everyone wants in a renderer, current trends and current marketing competition, wouldnt it be best for everyone that say lw 8.5 is released but it doesnt spend its countess hours of R&D on trying to get a newer renderer instead lw8.5 is the update that intergrates and supports all the best 3rd party renderers.
intergrated is Vray, Brazil or maxwell.

Trend is split apps are advancing more than all in one apps.

C4d- doesnt look like theyre going to update their default renderer which is based on cebas because cebas renderer would be in competition with them.
C4d announced final render just some days ago.

3dmax- bought and incorportated lightscape into its renderer but now 3ds has gained market due its Vray as intergrated as u would want.

Maya - mental ray sucks for archi work, but they have a choice as well.

Lightwave - I can see maxwell, or brazil being part of lightwave as it has things our artists use like SSS. Vray would make lightwave soo popular in its growing archi world, but it has to be intergrated so you dont have to import everything into a seperate render app.



All the money saved on a new renderer can go towards modeller and new particle effects.

CB_3D
07-22-2005, 11:58 PM
Theres a lot of stuf that coud be improved in lightwave, just as in any program i know. I see no problem in having an experienced programmer as oposed to a younger and more revolutionary one. One of the main concerns i have for future LW renderer revisions is stability and full integration.

I am really looking forward to future FP development, more so than LWs native renderer even.

And i wont change my sig,LOL.

toby
07-23-2005, 12:19 AM
Ade's right, allowing more 3rd party renderers to work with LW is very important, and they've said many times that they're working on that, but LW's renderer is much closer to being an excellent renderer by today's standard than Max or Maya's default renderer, and it's 'native' as opposed to 'integrated' like Vray, so with some work it could seriously compete with the freshest-latest renderers.

duke
07-23-2005, 12:56 AM
If ever there were circumstances in which Lightwave would allow external renderers to operate within Lightwave with decent access, and Vray were to become a fully compatible LW render plugin, I don't know anyone who wouldn't be happy :P

trick
07-23-2005, 02:03 AM
...wouldnt it be best for everyone that say lw 8.5 is released but it doesnt spend its countess hours of R&D on trying to get a newer renderer instead lw8.5 is the update that intergrates and supports all the best 3rd party renderers.
intergrated is Vray, Brazil or maxwell...

Surely not for me.
I'm allready using most other engines, but still prefer LW because of some things these engines still do NOT support; HD_Instance as an example. Try this thing with VRay, Brazil, Maxwell, fRender or MRay and you'll see the difference in handling...The bad thing about standalone renderers is the lack of supported features of the host application. Many people talk bad about fRender (for MAX) because it is slower, but actually this is my favorite because it is the most integrated one and this can be a big advantage in many cases. You'll have to realize that LW has still the best NATIVE render engine around that can render on UNLIMITED nodes. Some people would argue about C4D's AR, but you can definitely skip MAX's and Maya's. These are purely engineered for speed, not for quality. XSI even does not have one and goes straight to MRay: if you have a small studio doing animations you'll meet large investments and long setup times !!


If ever there were circumstances in which Lightwave would allow external renderers to operate within Lightwave with decent access, and Vray were to become a fully compatible LW render plugin, I don't know anyone who wouldn't be happy :P

Of course, but seeing the "integration" of 3rd party renderers in MAX (like previewing in the material editor for example) this will still take a VERY long time.


...All the money saved on a new renderer can go towards modeller and new particle effects.

But then again these have to be supported by the 3rd party renderer: this takes at least 2-3 revisions and this can take a very very long time for a renderer which supports many many host applications ;)

CB_3D
07-23-2005, 02:54 AM
There are several issues which need to be resolved.

1. LWs native renderer needs to be improved

2. LWs sdk needs to be opended up completely for external renderers like FP and the like

As far as we all know these issues are being adressed. I guess well know more in two weeks. :rolleyes:

Lightwolf
07-23-2005, 03:40 AM
I do see a problem when integrating third party renderers, especially if you make good use of the options available in the surface editor.
To date, FPrime is the only third party renderer that can actually access all surface attributes (disregarding shaders here), and this is because it runs within LW and not as a standalone.
So, be prepared to loose a lot of options (like for example all procedurals until they have been ported) in an external renderer.

The only way NT could remedy that situation is by letting third parties replace parts of the rendering pipeline (or all of it) _within_ LightWave.

Cheers,
Mike

CB_3D
07-23-2005, 04:58 AM
yes,thats exactly what i meant. opening up the sdk for complete access INSIDE the program.

operation
07-23-2005, 05:20 AM
hi,

Keep in mind that Fprime can't replace Lw native render engine ... ( for now, maybe in futur)

Fprime is a good tool for radiosity preview, to render a frame .... but there is too much limitations with this plugin . When you do 'broadcast' as me, you must render with the native render engine (interlace, antialiasing, complexe shaders with plugins ,etc...)

Fprime sounds faster .... but it's not ! It takes more time to render in Fprime than the Lw native engine ( with the same quality: Fprime refresh the screen, make disk access,etc...).I bought Fprime, but I am not satisfied ( this tools need more developement but I am sure this plugin will grow in the good way).


Because we don't have the same use ( architecture,commercial,TV,Print...) : we don't have the same needs.

Lw render engine is slow (everybody knows), of course the quality is very good ( you need to push to the maximum all the parameters).

It could be great to have the choise between 2 algorithms ( raytrace and reyes ( Renderman)). Most of the time we can tweak the scene ... the Reyes method could be faster for most of our work. And if someone needs more , he could activate the -raytrace,radiosity,global illum- mode ? aka (Renderman + BMRT: In 'Bugs life', Pixar used BMRT for all the 'raytrace shots' ).

---------------

About Electric Image: Don't judge the user gallery...

In the past, EI had the reputation of a very good and fast renderer ( used in many movies ...) of course the technology is obsolete now ... but the use of the knowledge of Mr. Granger is a good idea. He can bring a lot to the NT team.
Why to not judge the user gallery ?
Make a search on the net: Renderman+gallery ...
You will be desapointed, you will find "hugly" pictures from 'classic' users (sorry for them :( ) but if you watch movies, you can see how people use the maximum of it( Matrix,Titanic, Starwars,Fianl fantasy, Pixar movies,etc...).


Keep in mind it's hard to find a compromise between Speed and Quality 8~

My solution :D is to have more parameters in the Render panel to tweak , tweak ...
Look at Finalrender,Vray, Mentalray ... you have a large choise of parameters to speed up your render.(Mentalray and Renderman are the most tweakable)

In Lw ,there are not enough parameters . :thumbsdow

cheers!

Operation

ufo3d
07-23-2005, 05:32 AM
It would be good if NT hire Mr Henrik Wann Jensen, or integrated his incredible renderer Dali with LW, or as a third-party renderer, just my dream :D

Pavlov
07-23-2005, 06:11 AM
Imho NT shoul take the time to do both a good natibve engine and allow external ones.
Native rendering needs a major revamp; this can be done improving current one (or rewriting it), or integrating seamlessly a 3rd renderer (Like Maxon did - and therefore reprogramming all pipelines in order to allow access to every LW's function, plugins comprised).

Beside this, NT surely understand LW cannot be any more a "film tool" only. Market is way more segmented than years ago, it's simply not possible to "conquer" a whole production area and live with it only. If NT keeps paying more attention to film/FX industry than Viz/gaming, LW will have more problems than now.
Viz area will soon overwhelm film one, if it has not done already. Maybe not in America, but as i stated several times, here in europe there are 100 Viz studios for every FX/film one.
They cannot make Lw strong all around unless they "freeze" for 3-4 years, so at least let users choose an appropriate rendering solution. I'm positive about Mr. Granger, but i really doubt NT will be able to make its rendering as good for Viz as Vray or others. So, develop a good native renderer, but also allow us to choose the most suited for our needings.
Fortunately we have Fprime and Kray, which is not Vray (yet) but is a very powerfull photon mapping engine; allow these guys to make their tools better. Most of these engine's limitations are due to SDK limitations.


Paolo Zambrini

daros
07-23-2005, 07:18 AM
I think that are good ideas.

Lightwolf
07-24-2005, 05:10 AM
yes,thats exactly what i meant. opening up the sdk for complete access INSIDE the program.
...that will again mean that standalone renders aren't likely to be ported over (i.e. Brazil, FR, etc...).
Then again, a network renderable FPrime wouldn't be too shabby at all ;)
Cheers,
Mike

JML
07-24-2005, 11:37 AM
Then again, a network renderable FPrime wouldn't be too shabby at all ;)
Mike

a network renderable FPrime would be great ! , then I could finally use it at work, instead of just using it at home...
let's hope it would be soon..

wacom
07-24-2005, 05:47 PM
The funny thing is that in order to make a renderer that does everything everyone wants, from arch viz to animation, they renderer is going to have to get more complicated. That however takes away some of the advantages of LWs renderer- how easy it is to setup and get going. mental ray, renderman, vray- these are powerful renderers, but you have to spend way more time with them than the LW one to get what you need out of them though the rendertimes can be amazingly fast.

It's funny, because on another board, for another application, the arch viz people were saying that the company needed better arch viz render tools, or Vray support. The application is used more for motion graphics and animation. Well the people who use it for animation and motion graphics were up in arms over it- they just want faster animation and motion graphics and see the arch viz R&D as a waste of time.

The way I see it is that arch viz is a very special market and to render in it requires a special renderer that may not be the best for animation (Vray isn't really that great for it nor for creating skin etc). I'd rather have newtek work on making the current renderer more able to handle animation better (faster native GI solutions, SSS, nodal shader etc.) and open up the SDK etc. for renderers like Vray and Kray which seem mainly for use in arch viz. The arch viz market makes a lot of money, but is smaller in terms of users when compared to the game, flim, and motion graphics markets (and lets not forget web/flash). Maybe if NewTek could get things to bake as fast as they do in Kray 90% of the arch viz market could be satisifed with using LW...

The other thing to consider is this- how much can NewTek afford to put into a renderer where they give away the render nodes for free? The only other good one that seems to have free ones is PMG...and that isn't used that much. Are people ready to buy render nodes in packs?

Muad'dib
07-24-2005, 08:41 PM
The funny thing is that in order to make a renderer that does everything everyone wants, from arch viz to animation, they renderer is going to have to get more complicated

I think that would only apply if you approach it from from Vray/Kray point of reference. On the other hand you have things like FP and Lightscape which are very easy to set up. It all depends on the approach as far as I see it. So more powerfull does not have to imply taking hours to set up.


The way I see it is that arch viz is a very special market and to render in it requires a special renderer that may not be the best for animation

I think the things we're asking for are general improvements and will stand to benefit everyone not just the Arch Viz crowd. Pretty much all the suggestions you make are applicable to Arch Viz (faster native GI solutions, SSS, nodal shader etc...and open up the SDK etc) ;) yes even SSS.



The other thing to consider is this- how much can NewTek afford to put into a renderer where they give away the render nodes for free?

That's the model NT have been using since day one so I don't see any problem with continuing that apprioach. It is one of the biggest LW drawcards as far as I'm concerned.

It's not much longer till we find out where they are heading. Assuming they will give us that information :D ;) I'm quietly optimistic.

Nu Visual Scien
07-24-2005, 11:22 PM
EI has just been updated to 6 FYI

I recently bought an old copy of EIAS off ebay just for the render engine, It is that fast and very high quality.

Having said that i was very very excited to hear the guy who invented the render engine is now working at NT, If he is working on the render engine i expect fully that it will be an unbelievable solution, Very fast and very high quality.

I hope he doesn't get his mits on the sizing system inside LW though (Granger units indeed :rolleyes: )

Pavlov
07-25-2005, 02:26 AM
The arch viz market makes a lot of money, but is smaller in terms of users when compared to the game, flim, and motion graphics markets (and lets not forget web/flash).

This is a common mistake, because it's that way just in America. I can assure you that in Europe, Viz market (not only arch, but also industrial, scientific and design) is WAY larger that FX/Cinema/game market, and it's only going to grow. Maybe NT should take some time for some European market analysis, too. Afterall, Cinema has become what it is now (you know...) just because it properly surfed Viz wave. Max is a mayor seller because of thousand Viz studios using it, too.
So, my advice is NT has to consider Viz on the very same plane of Cinema/gaming/Fx.
Native engine cannot be perfect fo all purposes, in fact, as i stated, the major need is: let us choose the engine we need, so make that hooks and open that SDK.
I can predict with a good knowledge that if LW will remain the only tool unable to hook properly 3rd party engines, it will have (another) major drop in its sales, with obvious consequences.

Paolo Zambrini

3dworks
07-25-2005, 11:30 AM
If you think Mr. Granger is the solution, ok...you are the NewTek Marketing Team.
I'm only your customer.

The last time i visited the EI gallery was in 1995 and i have such a bad impression about the rendering engine that the first time i visited it again was 10 minutes ago only because you have insisted about Mr. Granger.
I think the next time i visit the EI gallery will be 2015.

Best Regards

David Rossmann

david,

first, big respect for your work! but as a longtime EIAS and LW user i cannot completely agree with your statements. agreed, a cgi software company's web gallery should reflect the quality of it's render engine, but not always the marketing team or the web designer is aware of what is excellent and what is not.

we are still using EI because of it's very good render quality and speed. the native LW renderer and EI's camera produce a very similiar 'photo-quality' like output. we choose LW or EI depending on the project and are very happy with the results of both renderers. our very demanding clients don't bother which software we use, but never complained about render quality. just as a game, have yourself a look at our website gallery at http://www.3dworks.com and try to find out which software was used on what project... ;)

that's why, of course, i'm very excited to see mark granger at work for the next LW render engine!

cheers

markus

wacom
07-25-2005, 04:02 PM
This is a common mistake, because it's that way just in America. I can assure you that in Europe, Viz market (not only arch, but also industrial, scientific and design) is WAY larger that FX/Cinema/game market, and it's only going to grow. Maybe NT should take some time for some European market analysis, too. Afterall, Cinema has become what it is now (you know...) just because it properly surfed Viz wave. Max is a mayor seller because of thousand Viz studios using it, too.
So, my advice is NT has to consider Viz on the very same plane of Cinema/gaming/Fx.
Native engine cannot be perfect fo all purposes, in fact, as i stated, the major need is: let us choose the engine we need, so make that hooks and open that SDK.
I can predict with a good knowledge that if LW will remain the only tool unable to hook properly 3rd party engines, it will have (another) major drop in its sales, with obvious consequences.

Paolo Zambrini

Arch viz and ID work seem to need slightly different things though. LW is fine for ID in terms of rendering sans the fact that it lacks good antroscopic effects and nurbs. Good snapping tools on the modeling side could help each side though. And Arch Viz REALLy could use some instancing (as could everyone else). I do think that if people want the speed of Kray and Vray though that they're going to have to except a more complex render engine. Can anyone name another Vray like renderer on the current market that is as fast yet as easy as lightwaves to setup?

I wasn't in anyway saying the arch viz market nor viz market is lesser than- only that the user base might be smaller than the broadcast, film, and game sectors (Maya?) combined. After looking at XSI very closely there are a lot of things that are great for film, video and games, but they don't really aid Arch Viz work. There is only so much money to go around, and LW still needs more than a better renderer to help layout move to he next level and beyond. A nodal based shading system with things like ambient occulsion passes built in would really help a lot of people...but arch viz? Is sub-pixel displacement on the top of your list?

You also have to realize that NewTek is based on broadcast and film. Think of how LW used to be a plugin for VT! Is it really worth it for them to try and compete with Max or would it be better to focus that effort and money on doing what they do well? Is the film or television industry hurting? Video games not making money? Hmmmm...

And the Viz market is big here too BTW so don't think that it has to do with any kind of cultural bias/regional difference.

Pavlov
07-25-2005, 04:56 PM
I agree about most of the points above, but i'd like to point out a different message.
I use LW for Viz moslty, but lets forget this now: i'm saying that Lw could gain a huge userbase paying attention to Viz market in the same way the care about Movie industry.
This means more more money for NT, therefore more power for us.
NT has born that way, but as i stated market is not the same of late 80s anymore, and a company simply cannot be a "field-only" one anymore, even more if it is a small company.
So, since we didnt see much attention toward Viz market until now, my suggestion is: pay Viz market more attention, in order to give a bettere tool fo all users.
Put a good Osnap/precision toolset (integrate LW-CAD !), take some time for a good DWG/DXF IO toolset (or make some strategic partnership wit a large CAD developer like Maxon did), and some other time to open SDK enough so Viz users can happily integrate Vray, Brazil, or Kray in an almost-seamless manner.
SSS and microdisplacement can be leaved to 3rd party developers, if SDK is clean and open enough.
NT efforts cannot be all-around exhaustive, they must focus on some areas release after release. Imho, doing the steps above by now would be the most effective focusing among all others. I agree LW users are mostly VFX oriented, but Viz market (not LW only, i mean the whole Viz market) is several steps larger than all others . If LW could gain even a small section of this area (by now a Max mostly area), NT would make a huge step ahead in term of income and software diffusion, and this is what we all hope to get in the end a better all-around, long-life software.

Paolo Zambrini

wacom
07-25-2005, 11:20 PM
Put a good Osnap/precision toolset (integrate LW-CAD !), take some time for a good DWG/DXF IO toolset (or make some strategic partnership wit a large CAD developer like Maxon did), and some other time to open SDK enough so Viz users can happily integrate Vray, Brazil, or Kray in an almost-seamless manner.
SSS and microdisplacement can be leaved to 3rd party developers, if SDK is clean and open enough.




Snap tools! How hard are they to put in? And everyone would get something out of it. Now here is the catch 22- many people NEED sub-pixel displacement. Anyone who owns Z-brush and wants to get the most out of it doesn't use LW.

So in and odd way I disagree about which is more needed, but I do agree that trying to stuff it all in might be impossable. Still, the arch viz market seems like it would love LW's straight forward approach to things and so maybe it should be the way to go. But I just have a feeling they might lose the people waiting for Sub-pixeld displacment, nodal shader trees, fast OGL previews, and better animation tools. There are people in the LW community that want to have a direct hook into renderman- and that seems to be the opposite dirrection you'd like things to go.

I know I'd be look elsewhere if sub pixel displacement, SSS, edges, etc. aren't in 9...**** they should have been and lets hope they are in 8.5.

Sorry to get slightly off topic- it just there is more to a good 3D app then nice renders...

Pavlov
07-26-2005, 02:09 AM
There are people in the LW community that want to have a direct hook into renderman- and that seems to be the opposite dirrection you'd like things to go.

Nope, it's the same thing.
Lightwave is not perfect, now more than ever... so its saving grace can really be if NT let users choose whatever engine. Just make this possible, by opening SKD and building hooks, and let people be happy with whatever engine. LW's engine will never be "the best" as in the past, some engine are really too powerful for NT trying to catch. Vray or Renderman, it makes no difference; if engine is a plugin class you can use the old but good LW modeler and then work in something really new and up to date. If you want Renderman and SDK is open, someone will be able to build a comminication tool wothout problems.
If we are costricted to stay within LW pipeline, NT has better to make it a new and actual one, but i really doubt it will happen soon, too much work needed since LW is still by all means an early '90 tool.
So, if it is still unclear, ****open that SDK pelase****.

Paolo Zambrini

Intuition
07-26-2005, 10:21 AM
I think that F-Prime made a huge leap for Lightwave's productivity and could be made even better (and probably is).

Since Worley maybe working on the new hooks in the SDK for implementing (net gossip) shaders, voxels, etc. F-Prime will become more of a direct replacement for Lightwave's native until Newtek updates it. This will become true as soon as the network render ability of F-Prime comes to reality along with the shader/plug in implementation.

As I stated in an earlier post in this thread I still believe Newtek will pull a rabbit out of the hat for a later release. Maybe not 8.5 but perhaps when 9.0 is made. I think by that time there will be major immprovements if not a whole new engine.

But here is something to consider.

Maybe Newtek is concentrating on the 64bit versions of Lightwave. If I were boss or product development manager/director I certainly would want to move full speed ahead on the 64bit evolution of computing. Although it may be a year away before we start seeing 64bit as common as 32 it is inevitable (Matrix pun intended) and all of us will migrate that direction. Heck, I would have already but things are still buggy and in test mode in the 64bit world, someplugs work, some don't etc.

Sub pixel displacement and all these other wish list features will be alot better afforded and implemented if the time is spent making a 64bit version that utilizes all the High RAM capabilities. Right now, I can't really get a scene as big as I like with huge textures without maxing out the 2GB wall.

Why add in sub pixel displacement if the texture maps aren't that big? Developing too much in the 32bit world is like backing ones self into a corner.

64bit should be the new direction for Newtek's Lightwave as well as its strong 3rd party developers like Worley.

This way they will not be moving with the crowd or following suit with trends they will be again leading the way and being the innovative company we have made our businesses with.

:D

EmperorPete
07-29-2005, 05:40 AM
Not everyone can afford to shell out for an app like FPrime, me included. I'm a home user on a limited budget; and as much as I would love to get hold of FPrime, it's not likely in the near future due to money considerations. As far as I am concerned, any increase in LW's renderer's speed (with no sacrifice to quality) would be a Godsend.

Lightwolf
07-29-2005, 06:42 AM
Sub pixel displacement and all these other wish list features will be alot better afforded and implemented if the time is spent making a 64bit version that utilizes all the High RAM capabilities. Right now, I can't really get a scene as big as I like with huge textures without maxing out the 2GB wall.

Why add in sub pixel displacement if the texture maps aren't that big? Developing too much in the 32bit world is like backing ones self into a corner.

I think, solving some of those problems inherent to the current architecture of LW by demanding a 64bit port is a bit like saying: Oh, the renderer is too slow? Get a faster CPU then.
Most of these issues can be solved on a current 32bit system (sub pixel displacements and high res image maps) as other vendors prove (Suprisingly, LW 5.6 had less problems with huge images (~200MB) then every version after, starting with 6.0).
Yes, I too think 64bit is the way to go, no doubt about it... But LW has in some areas not even reached the limits of 32bit with clever programming (it has reached the limits of 32bit due to some really brute force coding techniques).
I still think that buying more memory because a renderers architecture is not up to par is a bad idea. (print res renders come to mind as well).

For example:
* Why keep the complete frame in memory when rendering huge images for print? If you sacrifice some of the image filters (something you tend to do in post anyhow) you can let LW render in stripes and save those stripes consecutively to disk without using large amounts of memory.
* Whay load a complete image texture in memory if only a part of it is visible in the final render at any time?

Then again, I assume NT know fairly well where they stand and what they need to do to keep selling their product.

Cheers,
Mike

Intuition
07-29-2005, 10:54 AM
Good points Mike.

I agree that 32bit hasn't really been fully exploited. I think I am just equating that development time focuesed on 64bit will mean a better more complete final product next year.

Yet, I also agree that there are alot of things still left to do with the current 32bit frame as well.

Lightwolf
07-29-2005, 11:42 AM
I agree that 32bit hasn't really been fully exploited. I think I am just equating that development time focuesed on 64bit will mean a better more complete final product next year.
The hardest part is making the current code 64bit safe. Since the actual OS APIs don't change (except for only being callable from 64bit code), it doesn'T change much in development after at. I'd assume that 99.9% of the codebase is identical in both versions, if not 100%.

Cheers,
Mike

Nemoid
07-30-2005, 03:27 AM
Well, Lw renderer can be slow compared to other rendering engines but I would not bash Nt for its choices about new people to program and enhance what we have. how the heck can you guys judge people from what you've not seen yet. i really dunno.

IMO, Nt could move into different directions as for giving us better rendering solutions.

first one, they're working on SDK to give f prime full access to shaders, and integrate it better with lw. F prime means faster previews and faster rendering with a good level of accuracy so, doing this work is necessary.

secondly, with an updated sdk, other rendering engines will be ready to work easily with Lw. since there are great and fast rendering solutions on the market, good possibilities os working seamlessly with Lw will give us great benefits.

finally, improving what we have now and enhancing it its a good thing to do. so far, Lw rendering can be slow, but the light quality of it is noticeable. it also gives good results with not so much efforts.

these solutions will bring us alot of choice and power so that there will be no great problem on the rendering area.

i agree that going 64 bit and base rendering efficiency only on CPU power isn't a solution. the above points are good solutions and i think Nt is working on them all.

Lightwolf
07-30-2005, 04:04 AM
Well, Lw renderer can be slow compared to other rendering engines but I would not bash Nt for its choices about new people to program and enhance what we have. how the heck can you guys judge people from what you've not seen yet. i really dunno.
I 100% agree with that. As they say, the proof is in the pudding...
That goes both ways though and accounts for the pressure that the NT dev team must be under as well...
Cheers,
Mike

wacom
07-30-2005, 05:43 PM
instances-

I don't think they require 128 or 64bits to get them going. These simple things really help make large scenes a reality even on a system with limited RAM and CPU power- and as far as I know every competitor of LW has had them for at least several years.

I remember using them in MAX about...seven years ago...

XSI has great instancing support...

But all we need is a native version, even if it's bare bones. **** it would even solve most of the problems with using clip maps on things like grass, trees etc.

lo-fi yes...but very powerful!

Akryls
07-31-2005, 08:22 AM
Well, personnally I still use Lightwave.. mainly for the good and fast rendering engine :) . Of course, for Architectural work, it is a very slow renderer, because you need radiosity... and lw's radiosity is slow. But when you need 3D motion blur on 2k images with 2millions polys and raytraced reflections, lightwave's native raytracer is very fast (of course, no radiosity, no arealights, only spinning lights rigs..). When I say fast, I mean between 5 and 20 minutes per image on dual 3ghz, with dithered motion blur and ehanced medium AA.
The speed of the renderer depends of the kind of work... for film sfx, rides, etc.. it is a very good fast and cheap renderer.. not very modern, but simple and effective (work like a charm on a renderfarm, easy to set up, unlimited nodes licenses... ).

pixym
07-31-2005, 08:24 AM
"The speed of the renderer depends of the kind of work... for film sfx, rides, etc.."

I am not agree with that, Speed in rendering is something absolute I think.

Akryls
07-31-2005, 08:35 AM
Some renderers are faster than others on simple scenes (max for exemple), others shine on very complex scenes (renderman for example is slower than other renderers on a simple cube.. not on a higly detailed nurbs scene with displacement and motion blur ) .. this is what I mean , a person who render simples cubes everydays will say than renderman is slow, and the same thing will happen with another who try to render complex landscapes with displacement on max.

pixym
07-31-2005, 08:40 AM
OK :thumbsup:

toby
07-31-2005, 01:26 PM
He's got a point, I'm told that Renderman will render a 720x480 frame at the same speed as a 2k frame. I've been working with Vray lately, and while you can do noise-free radiosity test-renders in a minute or two, a plain renders will take 10 times longer than LW (and twice as much ram).

Akryls, do you know Farid?

Akryls
07-31-2005, 03:15 PM
Yes, if we talk about the same Farid :) .. the one who work at D2?

wacom
07-31-2005, 05:33 PM
Some renderers are faster than others on simple scenes (max for exemple), others shine on very complex scenes (renderman for example is slower than other renderers on a simple cube.. not on a higly detailed nurbs scene with displacement and motion blur ) .. this is what I mean , a person who render simples cubes everydays will say than renderman is slow, and the same thing will happen with another who try to render complex landscapes with displacement on max.

Very good point indeed! XSI's NPR rendering, while very flexable is sooooo slow compared to LW's or any of the LW 3rd party plugs. To say something a renderer is "fast" or "slow" is way too much of a generalization. LW's basic raytracing IS fast IMHO. And while Vray might be ideal for arch viz and some ID work I still to this day haven't seen it produce a better "skin" shader/render for a human than many other packages.

Many people have the same issues with Fprime as well...which is another story all together!

toby
07-31-2005, 05:35 PM
That's where I met him, but he's not at DD anymore -

Feel like coming to the US? I've seen your work, I can almost guarantee you a place at DD.

Hey, if Farid can do it, you know it must be too easy! :p

:hijack:

pixym
08-01-2005, 06:20 AM
...a plain renders will take 10 times longer than LW (and twice as much ram)...

Toby, What do you mean by "plain"?

pixym
08-01-2005, 07:26 PM
I hope the "more to come" from the LW 9 announce will include radiosity speed improvement. This will be really make LW safe for architectural animation renderings.
In addition the external rendering core surely means other third party renderers could come to LW...
:thumbsup:

toby
08-01-2005, 11:59 PM
Toby, What do you mean by "plain"?
No radiosity or raytracing, like test renders. I don't think Vray is capable of rendering anything, even a blank scene, in less than 4 or 5 seconds -

trick
08-02-2005, 12:28 AM
...I don't think Vray is capable of rendering anything, even a blank scene, in less than 4 or 5 seconds -


I am glad I use it myself so nobody can make me believe such nonsense :eek:
...oh...unless you are still on a P2 450 MHz :D
Since you've been using VRay yourself lately I think you should really dive in and start tweaking those parameters !!

toby
08-02-2005, 12:36 AM
You're not counting the time that it takes to gear up before every test render. And gear down afterwards.

trick
08-02-2005, 12:58 AM
Except for FPrime (that even has gear-up of some seconds...maybe gone in 2.0) every renderer has gear-up time, especially in MAX. With intelligent Prog.Pathtracing parameters I get almost instant feedback for tweaking my scenes...

daros
10-27-2006, 05:02 PM
Finally we are here one year later :)
As espected, thoose big promised improvementes in Lightwave 9.0, are more as disappointing.
The rendering engine is apparently faster but the cost is the AntiAlising quality during animation. No way to render complex scenes with Lightwave's 8.3 or prior quality.
Flickering...artifacts, moirees even at the higest AA without adaptive sampling.
or it's only my opinion?
Last year we rendered 12 minutes broadcast animation for the italian Winther olimpics.
With Lightwave 9.0 we are not able to reach the same quality even at maximum camera parameters.
The only option we have is to re-install Lightwave 8.3.
Best regards

David Rossmann

Exception
10-27-2006, 05:18 PM
The only option we have is to re-install Lightwave 8.3.
Best regards

David Rossmann

David, I assume you have tried rendering with the classic camera and the classic AA filters? Should produce the same result.

daros
10-27-2006, 05:33 PM
Classic camera is a little bit better as the perspective camera,
but not suitable for flickerfree animation in complex scenes.

Earl
10-27-2006, 09:56 PM
Strange... I haven't noticed this myself. Could you post an example comparison?

toby
10-28-2006, 12:10 AM
I'm sure you're doing something different, here's a comparison btwn 8.5 and 9, they're identical

Puguglybonehead
10-28-2006, 12:52 AM
Daros:

I've found LW9 to be identical in quality to LW 8.3. And, the Perspective Camera is definitely faster. I think something in your scene files must be corrupted. Rebuild your scenes from scratch in LW9 and I think you'll be pleased with the results.

trick
10-28-2006, 04:12 AM
...Flickering...artifacts, moirees even at the higest AA without adaptive sampling...

For me there is still only one Lightwave that is compatible with the work I do and the plugins I definitely need, and that is 7.5. I stopped waiting for better because there are far better and integrated options. Besides a lot of workflow improvements in 8 and 9 nothing has changed in the render and lighting area: bad AA, bad network rendering, slow GI, no instancing. I can NOT instance materials, I can NOT instance surface layers, I can NOT instance objects, I can NOT reference objects, etc. Is this too much to expect after 10 years ? I know they say a lot of work has to be done and they are doing it now, but that's what I have heard over 2 years now. My clients don't wait, so why should I ?

Pavlov
10-28-2006, 04:31 AM
Sadly it's true. Implementing KD trees is not that big improvement, in the ocean of things to do in render area. GI, lighting and AA shouldbe trashed and rebuilt from scratch, or even better they should implement actual solutions (MR, Kray, Prime, FR, whatever) which they will never and never be able to match, if not other for time reasons.
Sorry, but if there's almost no pro user out here using LW's engine there's a reason.
Fprime and Kray are a "sine-qua-non" to keep using LW.
I dont expect that much, luckily we have people like Worley and Grzegor Tansky serving actual rendering solutions, with tons of heavy limitations caused by SDK. Oh well, let's stop or i'll write pages.

Paolo

daros
10-31-2006, 09:20 AM
just tried to render a 5000x2500 image on a dual opteron with 2GB of ram.
The geomtry scene is about 2.5M polygons.
lightwave begins the render and runs for more as 4 hours, and then writes a
1 kb black EXR file. Nice!
I tried with many other formats but it dosn't save the image to disk.
The PC has about 400 MB free memory during rendering that means it should be able to complete the rendering without problems.
I runned the test about 12 times.
The same scene at 3200x1600 pixel works very well by using about 1.2 GB of ram.
cheers.

Intuition
10-31-2006, 09:49 AM
I have to say that for BG radiosity only with millions of polygons LW can still make nice images.

Yet, I've gotton so desperate for a cleaner more photoreal image that I've been budgeting in extra money for doing stuff in Maxwell and having it sent to a farm for quick rendering. Also, I've chased down a few things for animating the camera in Modo like GwennR's script which allows for Camera mot files to be saved from lightwave and used in Modo. I do this because Modo's render engine is about as good as it gets right now for quality vs speed. Nearly hitting Maxwell but in mintues instead of hours.

I went back and read this whole thread and noticed my optimism last year. I do think Newtek is fighting to get better things out but the next GI upgrade will be a make or break for me.

I am beginning to break down and use XSI more often and am considering v-ray as well. Even though V-ray and Maxwell renders are slow you get such a nice image that it is admissable.

I can't get that fine of an image with multiple hours in Lightwave native engine using all the bells and whistles (GI, caustics, etc).

Again, I must state that I am very happy with LW9 in an overall sense but, as stated in other posts, even though Newtek is working on catching up right now in certain areas I have had to extend my other software usage to get more then the usual "element".

I have high hopes for the next GI engine for LW but M*D* is probably going to make an animation announcement in January which may be tangible/purchasable by Siggraph 07.

I am using 9 in production. I fire up F-Prime and use BG-radiosity. Then I set up my scenes and send to the farm using BG-rad in LW native. The results are consistent. Interpolated RAD is fast but getting it to be smooth over animation can be tricky when you are trying to make it keep a fast render setting.

Pavlov
10-31-2006, 11:29 AM
Intuition, i understand you very well but wait for Kray 1.7.
Maxwell is good but you need a huge farm to use it for animation, and setup is not immediate - light emitters, materials and so on.
Kray will be one of the fastest - if not the fastest - render engine on the market, and with a Vray quality.
Its instancing and smart caching will allow to get few minutes/frame in animation, even on complex scenes (you can see in some example posted here in arch and viz section).
It's our real alternative, and a luck for NT which will have some more time to (re)code render engine.

Paolo

Intuition
10-31-2006, 02:04 PM
Intuition, i understand you very well but wait for Kray 1.7.
Maxwell is good but you need a huge farm to use it for animation, and setup is not immediate - light emitters, materials and so on.
Kray will be one of the fastest - if not the fastest - render engine on the market, and with a Vray quality.
Its instancing and smart caching will allow to get few minutes/frame in animation, even on complex scenes (you can see in some example posted here in arch and viz section).
It's our real alternative, and a luck for NT which will have some more time to (re)code render engine.

Paolo

Actually I have been using Kray a little last week. Its come a long way. I check in every few months over the last year or so.

Kray could be a nice solution. Thanks for the reminder.

I know that Maxwell is slow but I have a workflow setupo now that I am confortable with even though the renders are slow. I can get a frame an hour at SL13 on my quad core system which works really nice. So I can turn around ideas quickly. Then I can just apply that as on option to my clients and budget in a renderfarm. I am not giving up on Lightwave but for the next 12 months I have been seeking out alternative routs to get better results.

I'll jump on Kray at 1.7 as soon as its available. Hopefully it will become a nice option for future Lightwave use in the case that the next GI engine for LW is not up to par with current competing engines.

Again, I may sound dismal but I do love LW and hope for the best but must create high quality content in the meantime to keep up competition.

JML
10-31-2006, 03:55 PM
we use lw9, and we now use radiosity backdrop only and montecarlo 1 bounce on heavy (exterior) scenes (1-2-3millions polys), and rendertimes are fine.
(before lw9, we couldn't, it was taking too long)

the only problem is with interiors, where you usually need more than 1 bounce, and more than 1 bounce in LW is suicide...
so I hope they will work on that.

we have maxwell and I think it's a waste of rendertime, looks beautiful but
quality VS rendertime is horrible. Much more interested in Vray or Kray right now..

trick
11-01-2006, 02:40 AM
...Even though V-ray and Maxwell renders are slow you get such a nice image that it is admissable...

5-10 minutes on a 3GHz machine (and you can render with only 1GB RAM on your slaves) per PAL-frame for multi-million poly scenes with full GI, animated GI-baked people, instanced trees, etc. in VRay....I don't call that slow. Now Maxwell is another story: I guess if you don't want to tweak your renderer and just want an unbiased one-button solution you have to pay the price.

Pavlov
11-01-2006, 07:32 AM
Vray is fast, indeed. anyway it's nice to know we (LW users) have one of the best tools on the market soon available, since Kray 1.7 will be almost on pair with Vray, and thus it will save many of us from jumping on other boats for Viz work, until NT offers something new.
BG radiosity can work nice, but it's not like 8 bounces... you know, result is different. Having a solution which really does not care about interior/esterior issue, bounces, and renders n-bounce images at 2-4 minutes per frame, whithout having to export anything, is another thing ;)
Maybe times will get higher when Motion blur will be introduced (1.7 will ship with MB) but i guess they will remain around Vray times, which means the best you can get today.
Said this, i'm curious to see what Newtek will come out with in next update, hopefully there's something GI related.

Paolo

Pavlov
11-01-2006, 08:16 AM
nice to hear there are some scheduled improvements; i agree this "filtered mode" does not look too much a revolutionary step ahead, but let's see it in action.

Paolo

daros
11-01-2006, 08:29 AM
like the 200 AA filters that are totally useless instead of a single good one :)

Captain Obvious
11-01-2006, 01:43 PM
Doesn't sound too revolutionary though, at least judging from that small snip - particularly the "providing faster results in some situations" part.
It's a bit odd, that. In most of the 3D industry, it's always "OMG NOTHIGN IS FASTER THIS ROCKS!!!1," but in newsletters regarding Lightwave or Lightwave plugins, it's more like "this is generally faster, in some situations, than the other methods," even though it's like a gazillion times faster. Like this, on Worley's site:


FPrime's is particularly adept with complex scenes with many lights, transparency, reflections, high polygons, radiosity, and area lights. Like all 3D programs, FPrime's rendering speed is highly scene dependent. For simple low polygon scenes with just a few lights, both FPrime and LightWave render at roughly same speed. As scene complexity increases, (especially as transparency, reflection, area lights, and radiosity are added) FPrime's renderer tends to outstrip LightWave.
Seriously, tends to outstrip Lightwave? It's like forty times faster!

bryphi7
11-01-2006, 01:54 PM
I am the first one to bash, but lets at least get the program before we start!

Chris S. (Fez)
11-01-2006, 02:26 PM
I think it all sounds great, just as long as it is released in a reasonable amount of time.

I like the Lightwave AA modes. AA can ALWAYS be better, but I love having all the options.

Please Newtek, give us a GI pass and fix the frickin PSD exporter.

toby
11-01-2006, 11:52 PM
5-10 minutes on a 3GHz machine (and you can render with only 1GB RAM on your slaves) per PAL-frame for multi-million poly scenes with full GI, animated GI-baked people, instanced trees, etc. in VRay....I don't call that slow.
No that is fast, but I'm curious, how many radiosity bounces are there and is it cached or using an irradiance map?

Panikos
11-02-2006, 12:49 AM
Nothing beats FPrime. Its even 10 times faster than Maxwell :thumbsup:

trick
11-02-2006, 04:31 AM
No that is fast, but I'm curious, how many radiosity bounces are there and is it cached or using an irradiance map?

VRay does not use bounces as in LW or FPrime; you can choose a calculation method for the first bounce - which should be the most accurate one - and you can choose another calculation method for all secondary bounces (depending on the scene and the details in it, this one does not have to be that accurate). Of course the mentioned times do not include the Irradiance and/or Lightcache precalculations, but for a minute of animation these are mostly done within a a few hours on todays multicore/proc machines. All moving things must either be pre-baked or calculated in a seperate pass.

Regarding Kray: from what I tested it is still not near VRay's quality, especially with AA, Instancing, Light Quality, Host Integration, Quality customization, and some other minor things, but I'm always keeping an eye on it. Since I do 99% animations this could be different for stills though...

Pavlov
11-02-2006, 04:58 AM
Trick, which Kray release have you tested ?
1.7 has great AA and quality is very near Vray - if not the same.
Btw you need to get the grasp of it before, now it's easy to get good results but to match Vray you need to go a bit deeper in setting it.
I assume you tested 1.62, which is a totally different beast both for quality and speed... even if it had not instancing. So, can you clarify this a bit ?

Paolo

Erwin Zwart
11-02-2006, 05:14 AM
Nothing beats FPrime. Its even 10 times faster than Maxwell :thumbsup:
Kray is 10 times faster than FPrime :boogiedow

Panikos
11-02-2006, 06:12 AM
Kray is 10 times faster than FPrime :boogiedow

With baking :)
Without baking, considering events ?

Erwin Zwart
11-02-2006, 06:24 AM
With baking :)
Without baking, considering events ?

kray has GI per frame, time interpolation over frames and Shared GI cache for all frames, take your pick.
On top of that, with baking to UVmaps it gets to game-engine-speed, hunderds, thousands or millions times as fast ofcourse.

But we showed in a couple of tests in this forum (Architecture part), that you can do non-baked animations in a fraction of the time FPrime renders it, especially difficult GI solutions (like interior)
It sounds like you have to check kray again ;)

Pavlov
11-02-2006, 06:25 AM
Paniknos, what are you talking about ? Kray is several times faster than Fprime natively, without baking. It's natural, since Fprime is unbiased and Kray is biased.
Maybe give a look at all anims posted in this forums, they render at 2-4 minutes per frame without baking or caching.

Paolo

silverlw
11-02-2006, 06:41 AM
With baking :)
Without baking, considering events ?

This is one example without any cache/baking
http://www.newtek.com/forums/showpost.php?p=452551&postcount=178

And here is animation done of the same scene. http://hem.bredband.net/b223277/Studiohall1.avi
it's not cached or baked.

if you want to try it in Fprime the scene is here:
http://www.newtek.com/forums/showpost.php?p=452553&postcount=179

Panikos
11-02-2006, 06:47 AM
Cool :) I will check demo :thumbsup:

Jure
11-02-2006, 06:53 AM
I don't know about you guys but I love my kitchen! My kray kitchen to be exact! ;)
It can cook all kind of wonderful renders! Thanks to the chef Grzegorz LW users will not be hungry anymore. :P

Pavlov
11-02-2006, 07:49 AM
nice shot Juer, just abit splotchy in the corner but i love the mood. do you have moer images of this one ?

Paolo

Joscci
11-02-2006, 09:25 AM
This is one example without any cache/baking
http://www.newtek.com/forums/showpost.php?p=452551&postcount=178

And here is animation done of the same scene. http://hem.bredband.net/b223277/Studiohall1.avi
it's not cached or baked.

if you want to try it in Fprime the scene is here:
http://www.newtek.com/forums/showpost.php?p=452553&postcount=179

That's pretty incredible... I'm fully aware of Kray, but was not aware at how fast it has become. I, like Panikos, didn't think there was anything faster than FPrime. Thanks for pointing this bit of information out. Will definitively be keeping an eye on its development. I wonder... was the video test rendered on a single machine, a renderfarm? What were the typical machine specs?

Cheers. :)

tektonik
11-02-2006, 11:42 AM
here at work i have only a few weeks left before i am forced to use vray....

come on and release kray 1.7 already!!!!

i don't want to work with autodesk viz..

monovich
11-02-2006, 12:05 PM
Scanned this thread and just wanted to say I am very curious about Kray 1.7.

I tried 1.6, but was turned off to it by the lack of clear up-to-date documentation and tricky learning curve. I think it's workable and some of the renders I've seen have been jaw dropping... I just couldn't get much out of it while I took my first look.

Just don't forget that doing some of those manual settings is extremely intimidating for the more right-brained among us.

bryphi7
11-02-2006, 12:18 PM
Yes, i too will buy kray when it has good documentation. I tried it as well, and got frustrated by the lack of docs.(hint hint)
I am not a big fan of the trial and error method of learning.

silverlw
11-02-2006, 12:49 PM
That's pretty incredible... I'm fully aware of Kray, but was not aware at how fast it has become. I, like Panikos, didn't think there was anything faster than FPrime. Thanks for pointing this bit of information out. Will definitively be keeping an eye on its development. I wonder... was the video test rendered on a single machine, a renderfarm? What were the typical machine specs?

Cheers. :)

I rendered it on my amd dualcore 4400+. You can see the rendertimes of the frames in the format Hours:Minutes:Seconds:Hundreds of seconds

ops sorry if i got offtopic from this thread. Il shutup now.

lw3d23
11-02-2006, 01:13 PM
hi silverlw, when will 1.7 be available , approximately? is there a feature list of 1.7?

pixym
11-02-2006, 03:27 PM
EDIT: Erased because Silver has already told what I meant to Joscci.

Joscci
11-02-2006, 03:30 PM
I rendered it on my amd dualcore 4400+. You can see the rendertimes of the frames in the format Hours:Minutes:Seconds:Hundreds of seconds

ops sorry if i got offtopic from this thread. Il shutup now.

Sweet.

Yeah, I noticed the render time per frame as the animation ran through. Nice touch. :o)

Thanks, again for the info. :thumbsup:

Cheers.

JML
11-02-2006, 04:53 PM
I'm going to install Kray again to test it more..

I installed it a while ago but at that time, I did not have time to learn all the parameters to make a good render..

I saw some tutorials, but does anybody know a really good one that shows GI setting examples ?

toby
11-02-2006, 11:54 PM
This thread is very informative, with lots of images and animations posted.
http://www.newtek.com/forums/showthread.php?t=57914

I will be very tempted buy it once 1.7 comes out

silverlw
11-02-2006, 11:55 PM
Here is a good tutorial by Jure
http://www.kraytracing.com/forum/viewtopic.php?t=46&sid=0422abbc871e5a213d74b0c3415e90e8
and we are almost always online in Kray chat Effnet #kray or via this script http://www.kraytracing.com/irc/irc.cgi

Jure
11-03-2006, 12:17 AM
Silver that tutorial was written for 1.51 so it's a bit outdated. I suggest reading this thread that has more updated suggestions.
http://www.kraytracing.com/forum/viewtopic.php?t=229

trick
11-03-2006, 02:59 AM
Trick, which Kray release have you tested ?
1.7 has great AA and quality is very near Vray - if not the same.
Btw you need to get the grasp of it before, now it's easy to get good results but to match Vray you need to go a bit deeper in setting it.
I assume you tested 1.62, which is a totally different beast both for quality and speed... even if it had not instancing. So, can you clarify this a bit ?

Paolo

I tested the last downloadable demo (of course). Can you show me an animation, with what ever release you got, from a scene with LOTS of small details (trees, railings, detailed sharp textures) without flickering or any rolling lines ? (Of course withing the rendertimes I mentioned above) I still have not seen one !! This is not to prove you wrong: I'm really interested and would surely use Kray instantly if it could !!

Pavlov
11-03-2006, 03:26 AM
ok, so you tested 1.62 which is another thing form upcoming release.
Regarding animation, there's Howard's room posted by SilverLW, it has a good detail amount. Again, he posted a small sun animation in fryernder scene, dunno where these are, but they were posted recently. Hope silver jumps in and post some links ;)

Paolo

silverlw
11-06-2006, 01:13 AM
Here is a bunch of earlier posts
http://www.newtek.com/forums/showpost.php?p=448771&postcount=28
http://www.newtek.com/forums/showpost.php?p=452551&postcount=178
http://www.newtek.com/forums/showpost.php?p=452553&postcount=179
http://www.newtek.com/forums/showpost.php?p=453027&postcount=206
http://www.newtek.com/forums/showpost.php?p=453103&postcount=209
http://www.newtek.com/forums/showpost.php?p=455391&postcount=240

silverlw
11-06-2006, 01:22 AM
ah.. here is old Howard B room uncached/unprecalculated but DivX quality is poor. http://www.newtek.com/forums/showpost.php?p=448993&postcount=50

janoverust
11-06-2006, 05:23 AM
The Kray render looks VERY nice. I just might have to do some test with it...
Thanks for sharing.

Let me just add that I have faith in Newtek, and I'm shure they'll to come up with something just as fast... :)

otacon
11-06-2006, 09:28 AM
daros is right, visibly worse aa in 9 on animations.

Pavlov
11-06-2006, 09:58 AM
Is NT aware of this issue ?

Paolo

Intuition
11-06-2006, 10:31 AM
Yes, my hangups with Kray were needing to have moving geometry and not being able to get a non-flickering result. It always looked good when things were cached but everything had to hold still. If there was some sort of "smart" adaptive irradience that could minimize flicker then it would be more a part of my permanent pipeline.

The results are stunning though. :D

silverlw
11-06-2006, 12:58 PM
If you look at the posted animations you see thats exactly what 1.7 have.
http://hem.bredband.net/b223277/Studiohall1.avi
maybe i have posted the above before but in this quick anim the new stabilising algorithms are used.

toby
11-06-2006, 11:51 PM
daros is right, visibly worse aa in 9 on animations.
??
Can you post an example?
Did the example I posted? Stills, but identical... I'll try an animated one