Whats the big deal with edge weighting?

[Original1]

A lot of talk has centred around edge weighting

Whats the big deal with edge weighting?

If we are talking about the same sort of edge wieghting as Cinema 4D then we already have it.

I can apply a weightmap to my SDS and get a simular result, it even puts a line in in Celshader mode.

One of the things I would like to see is integrated paint simular to bodypaint.

A Sub Division Spline Mode might be nice, like biuld a spine cage, patch it, and then be able to subtly adust the splines in real time to adjust the SubD's but for most stuff I can get away with useing Larrys (splinegod) technique, and pulling the polygon cage about.

Please explain to me where you think modeller lacks a killer feature it really needs.

Please I don't want this to become a thread of negative comparisions, What I am really looking for is an explanation and disscusion of what people think Lightwave modeller lacks and how you would improve the toolset.

[phrick]

The big deal with edge weighting is that you can adjust edge weights (unlike Lightwave, in which you can only adjust point/polygon weights) :)

Check this out: http://maxon.net/pages/products/c4d/movie_modeling.html

[hrgiger]

I already responded to this in the Modo in action thread but once again, just to clarify, Lightwave does not have edge weighting.

[js33]

Yeah we definately need to have edge weighting then. :D
Of course that electrical plug could easily be made in modeler with regular polys and beveling but adjusting the weighting with a SDS model is much easier and quicker. I say that if modeler only gets one improvement then it should be edge weighting. If a lowly app like C4D has it and LW doesn't that will be very embarassing going forward.

Cheers,
JS

[hrgiger]

Originally posted by js33
Yeah we definately need to have edge weighting then. :D
Of course that electrical plug could easily be made in modeler with regular polys and beveling but adjusting the weighting with a SDS model is much easier and quicker.

Well, I think the real benefit in the case of that plug is it uses a lot less polygons to achieve the same shape of a mesh that's a lot heavier in density. We're not dealing with real-time renderers yet so unfortunately, efficiency while modeling is still a concern. Not to mention, the less dense it is, the easier it is to work with and tweak.

[riki]

Nice video

What's the difference between adjusting Edge Weights and the Weight of two Points? Is it just an ease of use issue or are there results that you can't achieve without it?

[Elmar Moelzer]

the difference is that weighted points work on all adjacent edges while edges only work on a single edge.
I.e with edge- weighting you can make one single edge hard. I would also like to see polygon- weighting that was shown here. With that you could turn a box into a cylindrish shape.
You cant to that with point- weighting.
CU
Elmar

[riki]

Oh good point, rather 'Bad Point' ;-)

[dablan]

Guys!
You can totally do this in LightWave!!
It's been in there for a while, although it's not called Edge Weighting.



I'm going to make a small AVI right now to show you how.

Hang tight -

[dablan]

Ok -
here's the deal: In LightWave, you can use SubPatch Weights to acheive the same effect you saw in that Cinema4D movie (posted earlier at the front of this thread).

The difference is that you're not really weighting the edges of polygons, but rather the points of a SubPatch curve.

I made a really really crude mock up of the Cinema4D example, and used some quick SubPatch weights on it. You can see that without any effort, you can change the shape of the edges, without added polygons. I've used this technique to also change how a texture was applied to an object. It's very handy.

Here's a movie I just made as an example -
http://www.3dgarage.com/EdgeWeightsinLW.avi
It's about 8MB - sorry for the file size. I'm working on a conversion right now. It is DIVX however, so both Mac and PC people can play it.

Hope this helps!

[jb_gfx]

I believe we all know what Subpatch weight are but we want them to works on an EDGE basis; not only point.

Dan you missed that in the C4D video.

Look at the pic I attached..

[jb_gfx]

...and that one from the video

[jb_gfx]

or..

[Karl Hansson]

If Dan Ablan don't know about edge weighting I guess it wont be in LW8.:(

[Original1]

OK Guys I see what you are getting at.

In some cases Edge weighting could be useful, but its not a killer application, in some cases you would want to model the detail in.

In others knifing the SubD's very close to the edge would give you the same thing.

I guess the programmers have to look at it and work out whether adding it to the feature set would be a plus or minus or wether the workflow would take a hit.

SubD are still converted to a polygonal mesh at render time, so sometimes you are better off leaving some bits as polys since you can mix and match polys and SubD's anyway

I am not sure you guys have completly made the case yet;)

[spikey]

is this what you're looking for? ;)

[sire]

Originally posted by spikey
is this what you're looking for? ;)

You bandsawed this, right? But it would be cumbersome to change the edge sharpness at a later time then, opposed to the true edge weighting.

However, edges as an own selectable category would have lots of other advantages than being able to apply weights to them. For example, it would also be easier to select parts of an object for beveling or rounding. Then, imagine a on-the-fly rounding mode similar to metanurbs, which draws its rounding parameters from edge weight maps... Non destructive (or actually non constructive so to speak), so that the rounding doesn't create real new geometry and can be changed anytime later. Would be great and kind of unique, I guess.

[TerryFord]

Originally posted by spikey
is this what you're looking for? ;)

Not really (that's using the cross-welded 4-point poly workaround isn't it?). It's an interesting effect but not very practical to set up in a complex model, and there's no control over the edge sharpness.

Dan's video example actually shows the weakness of LW weighs in comparison to true edge weighs (point weighting squares off the geometry).

LightWave needs edge support, for subpatch control and edge sharpness without unwelding/smoothing angle tweaking. These are standard (but not trivial!) features in all the other major apps, not something that we should have to justify or fight for.



Regards,
Terry

[sire]

Originally posted by TerryFord
Not really (that's using the cross-welded 4-point poly workaround isn't it?).

The WHAT? Terry, could you explain this trick to me? Sounds interesting.

[riki]

Does this mean then that Edge Weights could give you a result like this, but without adding extra vertices.

[riki]

Whereas using Point Weights squares things off.

Sorry for the nasty screenshots

[Yog]

Edge weighting is all about putting polygons (and sub-divisions) where they are needed.

You can get a nice curve in a 2x2 polygon grid by just pulling out the centre point, this is a very elegant way of making a nice smooth curve with the minimum of polygons.
If however you want the edge of the same model to have a sharp edge you have to currently add at least one or two more cuts, which is very wasteful.

When it comes to sub-dividing the above model at render time you will have to bring the sub-division up quite high in order for the main smooth section to have enough polygons in order to appear smooth. Unfortunately because it uses the same sub-division level for the whole model, the sharp edge, (where you actually need a lot less polygons), will get the same number of subdivisions as the main curved body, or far more if you used more than one cut.

On a heavily subdivided model edge weighting would allow you to have exactly the same effect, but with just a fraction of the polygons.

[riki]

yeah Sire I think what Terry means is this

You get a flat poly like this and weld two opposite Points

[riki]

Then you weld the two other Points

[riki]

And you end up with this. Which can be used in a mesh to create a tight edge but is a very anal intruder way of working.

[spikey]

Hmm. No crosswelding, just selected the vertices where i wanted it to get sharp and extended new polys from them into the torus where they're nicely hidden. When you have a new polygon extending from points of existing mesh, something in LW gets confused and creates a sharp edge there. But yeah, you can't control the sharpness of it which limits things.. Still, a useful little gimmick. Crosswelding is basically the same "error", i believe, just differently achieved.

[riki]

Nice one Spikey, but I guess both are a bit of a hack.

[spikey]

Yup. I'd love to have the real edge weighting as well..

[sire]

Originally posted by spikey
Hmm. No crosswelding, just selected the vertices where i wanted it to get sharp and extended new polys from them into the torus where they're nicely hidden.

Heh... Wicked :) Although, of course, it still has the disadvantage mentioned above, creating lots of wasteful polygons at render time. Just like knifing/bandsawing.

But wait, no, it's still better for doing really sharp edges. At least the subdiv amount doesn't have to be that high to get the sharp edges right.

[TerryFord]

Originally posted by spikey
Hmm. No crosswelding, just selected the vertices where i wanted it to get sharp and extended new polys from them into the torus where they're nicely hidden...

Cool, that's a new one on me :). I think it might require a LWO format revision to support edges (unless a custom VMAP could do it). Since that might break LWO support in third party apps I can see why Newtek might be relucant to implement it. Still, better to do it sooner and get it over with than leave it out until v9.0.


Regards,
Terry

[hrgiger]

Yes, edgeweights are completely different then the point weighting we have in Lightwave. I'm surprised Dan missed that. His model doesn't maintain the smooth corners the C4D model does, nor is is able to cut a circular hole in the prongs without adding a lot of extra vertices.

Hopefully, it's something that will work it's way in somehow...

[lunarcamel]

I've been waiting for Edge support forever - I guess I'll just have to keep waiting :(

Truly one of the biggest drawbacks to LW's modeler in my opinion.

[Karl Hansson]

I think that edge weights would be one of the most important improvements to modeler that I can think of.

[jb_gfx]

Originally posted by Original1
OK Guys I see what you are getting at.

In some cases Edge weighting could be useful, but its not a killer application, in some cases you would want to model the detail in.

In others knifing the SubD's very close to the edge would give you the same thing.

I guess the programmers have to look at it and work out whether adding it to the feature set would be a plus or minus or wether the workflow would take a hit.

SubD are still converted to a polygonal mesh at render time, so sometimes you are better off leaving some bits as polys since you can mix and match polys and SubD's anyway

I am not sure you guys have completly made the case yet;)

With edge weighting you should have more control without adding extra geometry, plus you can easily tweak the shape at a later time and finally texturing will be easier (cause less polygons).

[jb_gfx]

Originally posted by spikey
is this what you're looking for? ;)

I know how to model, thanks.

[spikey]

But wait, no, it's still better for doing really sharp edges. At least the subdiv amount doesn't have to be that high to get the sharp edges right.

This is interesting.. I went and did the same thing for each vertice rows (4) of the torus and rendered it. It didn't make a supersharp edge, there's a slight round-off. Now I'm curious if the roundness could actually be adjusted somehow..

[sketchyjay]

i'd love to see edge weighting too.

My only question is what do you do when it is time to move the model to another app?

seem like alot of companies use it in a pipeline and no file format can bring the edge weight info across (fbx may be an exception?) does this mean that I have to bake the object before moving it? If so do the frozen edges adjust to keep the same shape or appear in the middle of a face?

Sorry, but i haven't used cinema 4D so don't know how this is achieved.

Personally i'll never have to move the model outside lightwave so have no problem with edge weighting.

J

[Dick Ma]

I don't know how the edge weighting can applied into our daily modelling jobs. (are you doing plugs like Cinema 4D everyday?)

Just have a feeling that Modelers are getting lazy and lazy..

Make use of your bandsaw and knife tools can achieve the same results.

[dablan]

Naw, I know about Edge Weights just fine. These tools are pretty powerful in XSI, which we use here at the studio. But LW is our platform of choice.

What I was going at was a similar look - done differently, but similar results. Actually weighting the edge would be cool, but using the subpatch weights on points is a really good way go.

Let's weight, and see what happens with 8, so to speak.

[dablan]

hgriger -
yeah, I did misread one of those earlier posts -

what can I say - it was the end of a 16 hour monday. :)

[policarpo]

i know you guys want Edge Weighting now, but here's a little animated GIF showing how to achieving edge weighting now.

I just used David Ikeda's PowerTools (Connect, and Select Loop), the Knife Tool and Stretch.

I know it's not the same as the C4D and XSI tool, but it works just as well and it's fast to achieve.

The key is to have quick access to your edge loops so you can customize the weight of your edge to your liking.

Cheers.:D

[hrgiger]

Originally posted by dablan
hgriger -
yeah, I did misread one of those earlier posts -

what can I say - it was the end of a 16 hour monday. :)

Well Dan, you are entirely excused. You can't do much wrong in my book (or in yours). You're definately a great asset to our community here.

By the way, I haven't seen the latest Keyframe yet. I heard you say it was going to be at Barnes and Noble. I used to get it at Borders but I was just in today and they said they don't carry it. What's the latest issue?

[dablan]

Thanks man,

As for Keyframe - it's issue 34. Nice kick *** cover by Ace Miles. Red bkd, army dude kickin' some butt...

Should be in bookstores at the end of the month. I just got my copies today.

Thanks,
More to come!

[Alex Rooth]

This is a major issue for me. Certainly if it is not in 8.0, then I will be considering another modeller. Using the bandsaw or knife is a very inadequate workaround because it adds massively to the geometry of the model. This makes the model unwieldy and adds to the render time. It can also change the profile of the model. If you use the bandsaw too much, eventually the model will become virtually unworkable.

But the main thing is the extra time spent modelling. It is almost like having to model everything twice because you start off with your basic model and then have to use the bandsaw to get the edges looking right. The more time spent, the less productive you are, which is as frustrating for a hobbyist as someone trying to run a business.

Somewhere in the tranportation section of the LW gallery I have a cel-shaded jeep. Proper edge weighting or some other solution would have more than halved the time spent on this model. I used the bandsaw on almost every part and it is not an inuitive way to model. As workflow is the in thing, I think this issue should be addressed by Newtek.

My comments relate primarily to modelling non-organic objects. They also assume that the modeller is aiming for a high degree of realism in the object, i.e. something that looks like a copy of a real object and that will hold up to extreme close up still shots. Maybe this is not the aim or intention of Newtek or the rationale behind Modeler, in which case I am expecting Modeler to do something which it is not intended to do..in which case I should look for another modelling tool..

I will post examples of what I mean if anyone (someone at Newtek maybe) is interested.

Also, less time spent modelling equals a more productive person equals a happier person equals less ranting on mailing lists equals a happier Newtek : )

Best regards

Alex Rooth

[Original1]

Alex

did you see this stream from the show

http://www.3dgarage.com/Sig03/Sig03_JH_Celshade.wmv

I thought jen came up with some interesting tips for cel shaded stuff.

One of the things I find a real pain for Non - organic modelling is Screw and Bolt threads a plug or an L-script that gave you an accurate nut and bolt in US Imperial and metric threads would be good for modelling machinery

[policarpo]

Originally posted by Alex Rooth
This is a major issue for me. Certainly if it is not in 8.0, then I will be considering another modeller. Using the bandsaw or knife is a very inadequate workaround because it adds massively to the geometry of the model. This makes the model unwieldy and adds to the render time. It can also change the profile of the model. If you use the bandsaw too much, eventually the model will become virtually unworkable.

But the main thing is the extra time spent modelling. It is almost like having to model everything twice because you start off with your basic model and then have to use the bandsaw to get the edges looking right. The more time spent, the less productive you are, which is as frustrating for a hobbyist as someone trying to run a business.

Somewhere in the tranportation section of the LW gallery I have a cel-shaded jeep. Proper edge weighting or some other solution would have more than halved the time spent on this model. I used the bandsaw on almost every part and it is not an inuitive way to model. As workflow is the in thing, I think this issue should be addressed by Newtek.

My comments relate primarily to modelling non-organic objects. They also assume that the modeller is aiming for a high degree of realism in the object, i.e. something that looks like a copy of a real object and that will hold up to extreme close up still shots. Maybe this is not the aim or intention of Newtek or the rationale behind Modeler, in which case I am expecting Modeler to do something which it is not intended to do..in which case I should look for another modelling tool..

I will post examples of what I mean if anyone (someone at Newtek maybe) is interested.

Also, less time spent modelling equals a more productive person equals a happier person equals less ranting on mailing lists equals a happier Newtek : )

Best regards

Alex Rooth


if you are truly serious about non-organic modeling, you'll want something like rhino or SolidThinking or Alias Studio, since they support NURBS. Nurbs are the most efficient way to model non-organic models.

But I've seen enough bad *** work done in polygonal mode in LightWave that holds up to close up scrutiny, so I'm not really sure how "Edge Weighting" will help you in this regard. Geometry must be present if it is going to hold up to a close-up shot. Edge Weighting won't do this efficiently for you. Not too mention that the object will need to be frozen in the end if it is going to exist in a multi-app pipeline, which in the end it just might if it is client specific work.

Edge Weighting is cool, but it will never replace efficient and concentrated modeling.

Cheers

[sailor]

Hi policarpo :)

i dont get how you are changing your subweights in your gif without chanfing the tensionon the other edges? i just dont get IT? are you subweighting the points selected in your gif?

thanks for sharing...

[sailor]

BTW my 2 cents...

actually to be honest there are several problems to solve about edge weighting, lets brainstorm a little bit...

first of all the Import/export problem: for this reason certainly i have seen no modeler in a Maya/LW/Max pipeline using them...
i have a clue for this...considering that Catmull Clark algorythm has some "Nurbs roots" there is also in the Nurbs workflow some stuff like cv weights that are not often used because of IMPORT/EXPORT problems and things like that...it also seems to screw NUrbs UV parameterization...chances are that Catmull Clark smoothing will inherit those as well...in the end CAD modelers end up not using CV weighting in Rhino for instance...just to be sure that the model will be easy to transfer and not to deal with other probs later.

Also about WOrkflow: personally i dont find the bandsaw/split routine that hard to deal with...but this is personal of course...the split poly routine is universal and can be applied in any soft... in the other hand ...instead of making bevels with weights or dealing with the smooth aproximation i would like to have more rounding tools ala rounder or stuff like that...in Maya instead of making a split routine for sharpening edges with a smooth on top i often make a bevel...result the model has an additional row but it also has a bevel wich means that the bevel exists WITH or WITHOUT smoothing on top...

another problem could be UV smoothing have to check but this could screw the UV's or deal with new UV problems to fix (lets already fix the actual UV smoothing first :D)

another one: resolution:
this is important: using weights can end with less polys right...actually this is the only real good point of edge weighting versus extra control points...because this in the end is the real question: EDITABLE influence CV VS uniform weight with extra CV's for control...ask a Nurbs modeler and you will see that all the rounding routines are done using EXTRA CV's...i've never heard of CAD modelers using weights...this means that between 2 planes for instance in order to achieve a rounded fillet they use an arc based (extra c'vs) tangent surface instead of editing the CV weights must be a good reason for that (maybe tangency being screwed?)...i'm not saying we should not have edge weights but i'm saying that we should chechk first what could be wrong with this method :D

if the main problem is resolution with less polys i would rather use other Nurbs advantages wich are tangency between low and high density surfaces (no need extra rows at the edge) and non aligned surfaces (wich means no connecting points= less points= less polys) or accurate trimming and booleans...this kind of stuff will make a much greater economy of polys in the end

this in conjunction with a optimized MR tesselation control will be a resolution free superfast modeling render solution

:)

as a conclusion:

You NT guys listen this one: in about a month i'm starting another big production (this will be a Mayatollah one ;)

having some great poly tools could help me sell LW (wont be easy but well you never know)for its modeling strengths
i know you wont have time to change the modeler architecture but if you buy and implement stuff like vertibevel and rounder in the actual toolset believe me this are stuff that impresses A LOT Maya or Max poly modelers...so dont need this time to go for a very complicated stuff but at least this please? ;)
personally i would love to have history and a nurbs toolset but being honest LW poly toolset is rock solid...make it even better !!!!

[sketchyjay]

Originally posted by sailor
Hi policarpo :)

i dont get how you are changing your subweights in your gif without chanfing the tensionon the other edges? i just dont get IT? are you subweighting the points selected in your gif?

thanks for sharing...

there is no weighting. I made the same thing by making a box. bevel in the top a little. Then use Di_Connect the centers

to get the sharp top just use the knife.

rest is just selecting the points and scaling them to what you want.

See the attached picture.

I like the idea of weighting but if it can't freeze to an optimzed mesh when you export it then it will cause more geometry than just using bandsaw and knife selectively. (this is for exporting) internally it is very useful although what happens when you want to explode it or otherwise deform it? Does it hold up or does it act like a weightmap/endomorph?

I'll be happy when they get edges in.

J

[sketchyjay]

here is another view, smoothed and unsmoothed.

not that difficult or slow.

J

[policarpo]

thanks sketchyjay!

that's all i did. knife and select the edge loop and adjust to define the hardness/smoothness of the edge.

why fake it when you can build it. :D

[sailor]

"there is no weighting. I made the same thing by making a box. bevel in the top a little. Then use Di_Connect the centers

to get the sharp top just use the knife.

rest is just selecting the points and scaling them to what you want. "


well if i follow you you are just then knifing and stretching to make a sharper edge? hahaha so well the normal stuff u mean? if this is what policarpo ment then i dont see what is the deal ...come on you Poli always pointing OBVIOUS stuff so obvious that i always end up thinking it is revolutionary !!! :p

teh rpurpose of the therad poli was using weighting to avoid knifing/bandsawing if u answer with a knifing technique then you are facing the wrong way


The DI Connect what it does is what you can already do in native LW using a longer routine of course...select polys then multiply/moph or multiply/skin if you make this one though you can run a merge trigon x to convert the tris in quads...in a box example just select the top and bottom cap and run multiply/morph....no need DI connect actually

[policarpo]

Originally posted by sailor
"there is no weighting. I made the same thing by making a box. bevel in the top a little. Then use Di_Connect the centers

to get the sharp top just use the knife.

rest is just selecting the points and scaling them to what you want. "


well if i follow you you are just then knifing and stretching to make a sharper edge? hahaha so well the normal stuff u mean? if this is what policarpo ment then i dont see what is the deal ...come on you Poli always pointing OBVIOUS stuff so obvious that i always end up thinking it is revolutionary !!! :p

teh rpurpose of the therad poli was using weighting to avoid knifing/bandsawing if u answer with a knifing technique then you are facing the wrong way


The DI Connect what it does is what you can already do in native LW using a longer routine of course...select polys then multiply/moph or multiply/skin if you make this one though you can run a merge trigon x to convert the tris in quads...in a box example just select the top and bottom cap and run multiply/morph....no need DI connect actually

heheh...i know the subject of the thread. i was just illustrating that there are efficient and quick ways around limitations we all face.

since we don't have the feature (and i am sure newtek is aware of it) i feel it's important that anyone new to LightWave reading these threads who think it's a critical feature might miss the point without seeing a simple solution.

how often do people yell for having access to edges for modeling and never use Jigsaw which allows you to cut edges via defined loops and rings (while retaining the sub-d nature of your object)? We have most of the tools we need right now (and yes i know some of it is work around, but it is quick once you understand it).

David Ikeda's tools really help to optimize the workflow in modeler. Edge Loops and Edge Rings are now there and easy to access to the user and this is great, because it makes modeling more of a joy than a chore. SuperShift and Connect are essential tools for anyone serious about modeling quickly.

Well, anyway...i try to bring a little light into the fog ya know.

Cheers!:D

[glassefx]

You know I love all of this information going back and forth about the great LW modeler. I really have nothing to complain about when it comes to Lw's Polygon toolset (modeler wise). I just would love to see "Points, Polys, Edges, Volume" at the bottom left hand side of the modeler screen. Just selecting them and using them to manipulate one's model would be great! Not unlike (don't laugh) Nichman's Nendo.

<C!>

[Original1]

Originally posted by policarpo
heheh...i know the subject of the thread. i was just illustrating that there are efficient and quick ways around limitations we all face.


People build most nobly when limitations are at their greatest. -- Frank Lloyd Wright


since we don't have the feature (and i am sure newtek is aware of it) i feel it's important that anyone new to LightWave reading these threads who think it's a critical feature might miss the point without seeing a simple solution.

how often do people yell for having access to edges for modeling and never use Jigsaw which allows you to cut edges via defined loops and rings (while retaining the sub-d nature of your object)? We have most of the tools we need right now (and yes i know some of it is work around, but it is quick once you understand it).

David Ikeda's tools really help to optimize the workflow in modeler. Edge Loops and Edge Rings are now there and easy to access to the user and this is great, because it makes modeling more of a joy than a chore. SuperShift and Connect are essential tools for anyone serious about modeling quickly.

Well, anyway...i try to bring a little light into the fog ya know.

Cheers!:D

policarpo

Thats exactly the type of debate this thread was designed to start at the very least I just found some new modelling tools to try. Whats more their free:D

There are days when I can't beleive how talented some of the people are that build these freebies, Its all part of what I really love about Lightwave

[Meaty]

I agree with Alex. I do tons of technical modeling at my job, and the routine is bandsaw, knife, sub-d, rinse, repeat. Some of the new edge bevel tools have helped out, but edge weights would definatly be great.

policarpo, i am gonna have to disagree with your suggestion to switch to a different package:D the rest of lightwave is too cool to give up :D. C4D is a polygon modeler right? they managed to get edge weights in. Yes it could break the file format, but if you need to preserve data, cant the format only store edge weight info if you chose to use edge weight info?

[policarpo]

Originally posted by Meaty
I agree with Alex. I do tons of technical modeling at my job, and the routine is bandsaw, knife, sub-d, rinse, repeat. Some of the new edge bevel tools have helped out, but edge weights would definatly be great.

policarpo, i am gonna have to disagree with your suggestion to switch to a different package:D the rest of lightwave is too cool to give up :D. C4D is a polygon modeler right? they managed to get edge weights in. Yes it could break the file format, but if you need to preserve data, cant the format only store edge weight info if you chose to use edge weight info?

Each user is different so the situation is different.

Edge Weights are a good thing...don't get me wrong...but there are ways around it currently that work well and are efficient. This is all i was bringing up with my points.

If you are serious about doing inorganic work, NURBS is definitely the way to go...but like I said, i've seen amazing things created in LW with straight polygons...so in the end it is the artist that defines what is possible.

Cheers.

[DaveW]

Originally posted by policarpo
Edge Weighting is cool, but it will never replace efficient and concentrated modeling.


Edge Weighting is a part of efficient and concentrated modeling. You really think bandsawing 50 edges and doubling your geometry is efficient? And for games, being able to set hard/soft edges for smoothing is very useful. The closest thing to that in LW is cutting/pasting polys where you want a hard edge, but that can often cause rendering errors and it adds a bunch of extra points.

[policarpo]

:rolleyes:

i love things taken out of context. :)

[Dodgy]

It wouldn't neccesarily have to break the format to get edges in. The LWO2 format is an IFF format, admittedly a custom job, but shouldn't they be able to just slot an edge chunk in there, if only for edge weighting? Any third person loader should then just be able to skip the chunk (losing the edge data, but as we've noted, subd's are usually custom to a package anyway aren't they??)

Also, the example by spike of the rounded sharp torus seems to be an effect of smoothing in lightwave. If the 'sharp edge' polys (the ones he created by point extending) have a different non-smoothed surface, while the rest of the torus has smoothing on (of an angle greater than the sharp edge) you get a very neat shading across the sharp edge....Just like in the pic.... Very bizarre!

[Alex Rooth]

Original1 - yes I did see Jennifer Hachigian's Siggraph presentation and enjoyed it very much. Her approach to modelling for cel shading is very interesting and well thought out.

Policarpo - please post links to the bad*** models you are referring to. I would be interested to see them. If I felt it possible to dispense with sub-d modelling for non-organic objects, it would save me a lot of hassle.

As a general comment, I am not sure how well the simple examples that have been given would transfer to complex objects. Can anyone provide screen shots of complex objects using these techniques?

Best regards

Alex

[policarpo]

Originally posted by Alex Rooth

Policarpo - please post links to the bad*** models you are referring to. I would be interested to see them. If I felt it possible to dispense with sub-d modelling for non-organic objects, it would save me a lot of hassle.
Alex

ask and ye shall receive:

http://www.newtek.com/products/lightwave/profiles/GeraldAbraham/index.html

be sure to check out the Mustang Wireframe and his site as well.

All poly modeled. :)

these are particularly great examples of Gerald's awesome LW poly modeled work:
http://www.room201.net/docomowf.jpg

http://www.room201.net/docomo.jpg

[DaveW]

Originally posted by policarpo
:rolleyes:

i love things taken out of context. :)

I don't see exactly how that was out of context. All your posts in this thread have been about workarounds and saying edge weights and edges are not necessary. That quote pretty much summed up your posts I thought.

[policarpo]

i never said edge weights weren't necessary.

i said they were a good thing.

we don't have them now.

newtek should be aware of this.

nurbs are a great solution for non-organic modeling.

great things can be done with polygons.

here are some examples:
http://www.newtek.com/products/ligh...aham/index.html
http://www.room201.net/docomowf.jpg
http://www.room201.net/docomo.jpg

and in the mean time this is what you can do to get around the lack of edge weights.

not everyone knows about Ikeda's PowerTools, and Jigsaw.

that is what i said, to paraphrase what i said...if you want to quote me.

cheers!:D

[Elmar Moelzer]

Actually I dont think that edges in Modeler would be necessary for most of the things discussed here. So far I have seen mostly things that should be doable with weighted Polygons, not weighted edges.
In case of that C4d- video polygon- weights were actually used more often than edge- weights (which I saw him use only once, if I remember correctly). So to me this would mean that there was still a need to add something new to the object- format, but this could be as easy as adding a new type of selection- set or the "parts" we have right now. Which are all based upon points and Polys.
The only thing I am not sure about is whether NTs SDS- algorythm would be able to handle this, otherwise it should be pretty easy to implement.
CU
Elmar

[DaveW]

Elmar, I think polygon weighting just weights the edges of the selected polygons. Which would mean that anything doable with poly weights is doable with edges, but not necessarily the other way around. I could be wrong, but that's what it looks like. There are other things edges are good for besides weights too. There are workarounds, but for every workaround I've come across a situation where the workaround doesn't work. And when they do work they often very time consuming.

policarpo, I guess I just took your posts the wrong way, it seemed like you didn't think there was any need for edge weights.

[Elmar Moelzer]

True and I perfectly understand that. But I was thinking of something that might be doable more easily and therefore might happen earlier. I think that polygons- weights would be better than nothing, and nearly as good as edge- weighting, right?
CU
Elmar

[Alex Rooth]

Policarpo - thank you for the links to Gerald's work. His models and renders are beautiful - no doubt about that.

But, it appears to me that these are mainly subdivision objects, and the way he has got his nice curved edges is through use of the bandsaw tool (in the Newtek interview he even says how much he likes the subdivision LW Modeler). Which brings me back to my first post and my point that using the bandsaw to add detail in order to get the right sort of curve is a cumbersome way of modelling.

I am sure that Mustang is a subdivision object - look at the curve on the rear bumper - that has come through hitting the tab tool right? If not, I really am missing something. Also, scroll down his page to look at the wires for the T-Bird interior. This has to be a subpatch object and it has all the bandsaw cuts where you would expect them.

The point I am trying to make is that I for one want a way of modelling subpatch objects without having to use the bandsaw to control my edges. Assuming I am right, and Geralds objects are subpatch, then it seems to back up my argument that subpatches are the solution for detailed realistic shots. In which case, we need a better method for controlling their edges...

Best regards

Alex

[sailor]

Alex,

yes it is a subdiv model for sure and we are here at the same point of discussion (read my last post)

the edge weighting solution is a cool one but that can bring new probs...this said better have them than not...

the real thing even bettre than that is having the possibility of modeling using NURBS and now this is for Policarpo....can you remember how many times Sailor man has posted about Nurbs in LW ? i'm glad to see you agrre with me then :D

"If you are serious about doing inorganic work, NURBS is definitely the way to go..."

Thx Poli i would love to see a Nurbs toolset in LW ...this said from a developpement point of view and $$$ POV having NUrbs will be even more complex than adding edges...but for me it will be much more interesting in the long term...would be great :D

[policarpo]

Originally posted by sailor
Alex,

yes it is a subdiv model for sure and we are here at the same point of discussion (read my last post)

the edge weighting solution is a cool one but that can bring new probs...this said better have them than not...

the real thing even bettre than that is having the possibility of modeling using NURBS and now this is for Policarpo....can you remember how many times Sailor man has posted about Nurbs in LW ? i'm glad to see you agrre with me then :D

"If you are serious about doing inorganic work, NURBS is definitely the way to go..."

Thx Poli i would love to see a Nurbs toolset in LW ...this said from a developpement point of view and $$$ POV having NUrbs will be even more complex than adding edges...but for me it will be much more interesting in the long term...would be great :D

well, i'm not sure that i'd like to see NURBS in LightWave. I like the fact that LW is a straight polygonal modeler. It does this job very well. I'd like to see more polygon workflow enhancements in Modeler, before introducing a completely new modeling paradigm in an already solid toolset.

If i were serious about NURBS, i'd look into other packages, like rhino, SolidThinking, and Alias Studio. These are pretty much standards for Industrial Designers and Product designers. (i leave out Pro-E because this is more for hardcore engineering work).

[sailor]

Well i have just to clarify the fact that the Maya Nurbs toolset is very good and apart from some specific surface analysis tools that you find in Studio tools or Rhino it is complete and doesnt lacks any major nurbs tool (but of course a LW hardcore fan will never say "buy Maya then, but buy Studio tools then ;) this way he is not "betraying" the LW cause huh? i only remind you that Studio tools is an Alias product :D)..
if LW wants somehow to be a Maya competitor this prooves that having both POlygonal and Nurbs toolset is not only possible but certainly an extra toolset for the modeler...and an extra argument...saying that if you want to go Nurbs you should try Studio tools is crazy firts because Maya has very good NUrbs (i use them) wich then means that price/features LW is too expensive (this is not true i'm teasing ya) then and also Studio tools is one of the most expensive solutions and it is really "overkill" for the entertainement industry needs..
.i think Poli that your views as usual are pretty extremist...in one side we have the "keep LW without Nurbs" argument ...in the other the "buy a superexpensive only Nurbs package (Studio tools)"...why not a contructive, realistic an feasible integration of Nurbs? why allways a 0/1 logic in here? i would like to have both polys and Nurbs what is wrong with that?

adding NUrbs wont cause any harm to LW modler but rather the opposite i believe....

whatever...personally i would like to see Nurbs and that is my personal opinion of course

peace

PS:
an extra point when you say that Modelr should remain a "straight" poly modeler it is not the case anyways...someone using cardinal splines to control a poly output are basically not using a full poly toolset as a cardinal spline is not a polygon :) result: LW modeler is not a straight poly modeler
if you add to that the fact that we are using a Catmull Clark algorithm in top of the polys then again...Modelr is not a "straight poly modeler" this is just a lack of knwoledge of the fact that all this technics are just differnet ways of achieving the same result......replacing Cardinal by Nurbs splines (with the specific editing Nurbs toolset) could simply enhance the current splines possibility...this alone could be a great improvement....

[policarpo]

true...Maya would be the right solution if you needed a good polygon and NURBS modeler. now with the inclusion of mental ray on OSX and PC and Linux...it's almost a no brainer decision.

we use Alias Studio and Pro-E here...so that's what I am familiar with from an Industrial Design perspective. LW does not work well in the pipeline since everything we do is NURBS based...

I am trying to get the ID guys to switch to Maya 5 so we can have the best renders possible...but they are stuck in their Alias Studio workflow...but if i keep at it...maybe i'll wear them down. :)

[sailor]

:)

well Pro-E is a solid modelr i think? i haevnt used it...BTW tehre are even some solid modeling stuff that maybe could be fun to have in one if the main 3d apps...i wonder if something like Electric Image had a Solid modeling toolset...the improtant think is to use the strength of every toolset as no toolset is perfect...
but i personally of course still find the poly workflow the fastest and more solid of all of them...

[glassefx]

Use Rhino NURBS modeler for all things that make it neccessary.

If one could not achieve it with Wave or Rhino something is not "right."

It exports some pretty nice meshes after one gets used to the converting dish...


:)