PDA

View Full Version : LighWave and Game Development



VonBon
10-10-2014, 11:34 AM
LW 3D group, you really need to take Game Development more seriously.
And not just for Unity, but for AAA creators like Unreal and Cry.

If you haven't looked at the new Blueprint functionality in Unreal 4 you should. :twak:

You could create a place for LightWave in this Industry if you stop limiting yourselves
to "O, we're great for small studios" or look at all the televisions shows that use LW.
Cause creating "Fast and Cheesy" effects will not keep you in the game. Integration
to other products is a good idea, but you need to do something better than what others
are doing in order to warrant access into an already existing pipeline. I want to say
some other things but I want to talk about Game Development.

A lot of people want to make games and if they are like me, they can't program. I have
been working in Unreal(UE4) for almost 2 months learning Blueprint and I have to say,
its really, really good. I'm picking it up faster than I thought I would. I've been figuring
out more and more complex blueprints to get things to work how I want them to work,
and not just copying work flows from tutorials.

My point is that I don't want to have to jump through hoops to use LW with UE4 when
I'm ready to get the art side of everything in the game. Yea Unity is cool, but I dream BIG.
In my opinion LW 3D group needs to take integration to "AAA Title" game engines seriously.
I'm sure you have game studios who would love to have options for software to use other
than what they are offered from Autodesk.

The way I see it (and Ima keep it real wit u) Autodesk is way out front when it comes to
Film effects. Animation, particles, fluids, simulations, etc..... I know, these companies have
huge amounts of money to blow on R&D and sometimes just outright purchasing new Tech,
so it means Newtek has to be smarter in its approach to competing. Game Development is
that best avenue in my opinion to take when competing with competitors. To compete in the
Game Industry you need to do less things great to compete unlike in film and television. There
isn't as great a need to compute fluids, particles, muscles and simulations in game engines as
you do in film and television. At the price you offer more than any other 3D app (besides Blender)
on the market.

What I think LW needs is a GREAT conversion Tool/Plugin that translates may animations to Unreal.
I need the surfaces that I create converted to materials in Unreal. I need my smoothed areas
to be converted into smoothing groups for Unreal.

Why Unreal? Download it and checkout Blueprint, and that's why Unreal.
Why do I stay with Newtek? Because it is privately owned, and we need
to support privately owned companies so that monopolies can't just go
around doing whatever the hell they want. And lastly, I'm a FanBoy :heart:

(O'yea, to you know who, don't be addn yall "Crappy Two Cents") ;D

MarcusM
10-10-2014, 12:15 PM
I support you in this post. In my opinion there is two possibilities. There is no man power for developing LW in game dev direction or there is noone who know on this market. Or both.

GandB
10-10-2014, 12:40 PM
I also support this; though I have serious doubts they'll follow through with anything. But kudos for keeping up the pressure, you never know. I would certainly change my opinion of the LW3DG, if they started taking game art/dev seriously.

Sensei
10-10-2014, 12:48 PM
VonBon, you wrote so much text, but to be honest, I don't know what exactly do you want.. :) So probably the same other non-artists programmers, developers.. Confirming what MarcusM said.

Features should be straight forward, specific.
Like f.e.
write importer for format XXX,
write exporter for format YYY,
(XXX and YYY must be open)
export data of selected objects using that and that function from library that's given in link...
etc. etc.

If you don't know what do you want from developers, how can they know?



I need the surfaces that I create converted to materials in Unreal.

Isn't Surface Baking Camera tool doing that?


I need my smoothed areas
to be converted into smoothing groups for Unreal.

Isn't baking normal map doing that at much better resolution than per vertex?
I can't understand why people want such obsolete vertex normals that were good in '90, while now should use normal maps for everything..

VonBon
10-10-2014, 01:14 PM
Glad to see you participating Sensei.
I will definitely be more specific and clear when asking for features from this point on.

I will tackle the Materials request first sense I'm working on spawning
dynamic objects with dynamic materials now in Blueprint. Probably by
Sunday I will have a detailed request ready to post.

mummyman
01-09-2015, 09:40 AM
I'm interested to see more of your progress with Unreal. Any updates?

MSherak
01-09-2015, 10:46 AM
LW 3D group, you really need to take Game Development more seriously.
And not just for Unity, but for AAA creators like Unreal and Cry.

I work in the game industry and while this is all and great for the independent or just starting our 3rd party developer. When you get to serious game development you are not using Unity or Unreal or any third party game engine. You are using a proprietary game engine with tools support in-house. A more robust SDK would be my first asking on the list with more samples geared for developers that are looking for easy ways to data collect from a 3D package. The most one needs from a 3D package for the use in a third party game engine is geo and animation. Everything else is done in the editor. For that, FBX is adapting very well into the format for conversion between applications.

Other notes. Try the special "GAME" versions of 3D software. They fall flat since they are lacking options most artists use. They don't help with development, they hinder it in many ways. You will not see a major studio using these versions in development. Again they are great for the independent and are marketed that way. As for the "smoothing groups" request, this is MAX's lacking and lazy way of not having to store vertex normals and there are lots of reasons why it's not a good thing which I won't get into. Much easier to assign the normals yourself with correct tools.



Isn't baking normal map doing that at much better resolution than per vertex?
I can't understand why people want such obsolete vertex normals that were good in '90, while now should use normal maps for everything..

With vertex normals one gets better control over the normal map generation. Lots of other applications, like Zbrush, rely on the vertex normals to control the brushes flow when scuplting. So they are very relevant in the current day and age. Most powerful thing is 3D is a vertex. Without it you have nothing to attach to. Vertex normals is the one thing lacking in Modelers tools. If we even had a simple tool that assigned smoothing angles to selected poly, edges or verts many artists would be in heaven. Even if there was no undo to it right now. Funny thing is we have a Vmap for them. They are viewable in OpenGL and render. Just can't assign them. Have to go outside the package to assign and bring back in.

-m

lightscape
01-15-2015, 06:03 AM
As for the "smoothing groups" request, this is MAX's lacking and lazy way of not having to store vertex normals

If we even had a simple tool that assigned smoothing angles to selected poly, edges or verts many artists would be in heaven.
-m

Say what? ;D
So we shouldn't ask for the way to assign smoothing groups but we should ask a way to assign smoothing angles?

jasonwestmas
01-15-2015, 09:28 AM
For starters I'll just say that AAA game characters need a very strong vertex weighting set of tools, LW does not have this when compared to competing animation/deformation toolsets. Artists working on such games can't have arms and legs that look like a towel folding over when hold bones are not always an option. Most of this deformation control needs to happen within the weights. This is especially so because everything deforming is most likely driven by bones and weight maps. When we get to assigning gear and such to deforming models we need a lot of control over how rigid or stretchy the models deform and also to prevent inter-penetration of polygons "clipping". I can't count the number of times I refused to use Lightwave for animation because of the difficulty to get good weightmaps quickly.

I tend to agree that normal maps give more control over hard/soft edge control. I definitely use Normal Maps for everything these days. I'm not sure what vertex normal smoothing has to do with zbrush. Zbrush doesn't have vertex normal smoothing yet unless you do a BPR render.

MSherak
01-15-2015, 10:33 AM
Say what? ;D
So we shouldn't ask for the way to assign smoothing groups but we should ask a way to assign smoothing angles?

Assigning to the vertex normals to the vertices gives more control over the edge smoothing due to not being limited to a fixed number of angles. Would be like only have a fixed number of values to type into the smoothing angle value in the surface editor. Ick. Let me select poly, points or edges and assign an value.


For starters I'll just say that AAA game characters need a very strong vertex weighting set of tools, LW does not have this when compared to competing animation/deformation toolsets. Artists working on such games can't have arms and legs that look like a towel folding over when hold bones are not always an option. Most of this deformation control needs to happen within the weights. This is especially so because everything deforming is most likely driven by bones and weight maps. When we get to assigning gear and such to deforming models we need a lot of control over how rigid or stretchy the models deform and also to prevent inter-penetration of polygons "clipping". I can't count the number of times I refused to use Lightwave for animation because of the difficulty to get good weightmaps quickly.

I tend to agree that normal maps give more control over hard/soft edge control. I definitely use Normal Maps for everything these days. I'm not sure what vertex normal smoothing has to do with zbrush. Zbrush doesn't have vertex normal smoothing yet unless you do a BPR render.

If you just rely on bones auto assigning weights you would get those results of bad folding. If you want great refined vertex weights in Lightwave use VertexPaint. It's a great tool and has been there since 6.0. Here is the old DStorm docs for it back in 2001. ftp://ftp.newtek.com/pub/LightWave/Tutorials/modeling/vertexpaint/VertexPaint.pdf So Lightwave was doing weight maps way before most packages were. I'm not saying we could not use better tools. It can be launched from Layout or Modeler but most overlook it. It was originally made for game development. I would like to see and update to it so it could read joints and the new scene format for animation. Try it.

As for normal maps. If you are using them for all your edge looks then you are missing the power when used with vertex normals. Here is a quick read and video showing that more control over the model when using both. http://www.polycount.com/forum/showthread.php?t=107196 Now don't mind the vertex count since MAX does not create Vmaps like other packages so it just adds extra verts. This means the verts are always discontinuous along UV seams and hard edge vertex normals at the point construction level. Hence why they having smoothing groups to get around this. Having a construction vertex with stored vmap sets keeps your model as efficient as possible by not increasing the the construction points and storing the delta offsets for UV's and vertex normals. Hence why in Lightwave's UV panel you can see the red dots of this delta on discontinuous points. The red dots are the offset seam within the vmap and not extra construct vertices. This is one area that Lightwave needs is the ability to assign vertex normal values. It already can assign, display and render them if they come in as a FBX or OBJ model. All we are missing is the ability to assign them in the VMap.

-M

jasonwestmas
01-15-2015, 11:29 AM
If you just rely on bones auto assigning weights you would get those results of bad folding. If you want great refined vertex weights in Lightwave use VertexPaint. It's a great tool and has been there since 6.0. Here is the old DStorm docs for it back in 2001. ftp://ftp.newtek.com/pub/LightWave/Tutorials/modeling/vertexpaint/VertexPaint.pdf So Lightwave was doing weight maps way before most packages were. I'm not saying we could not use better tools. It can be launched from Layout or Modeler but most overlook it. It was originally made for game development. I would like to see and update to it so it could read joints and the new scene format for animation. Try it. -M

I hope Vpaint isn't reason Newtek hasn't brought these kinds of tools to a better standard. I've tried that approach a while back where you can import your bones from layout (I'm not a fan of skelegons) using a very old script from DStorm. Inside VPaint, the skelegons have no IK or key framing ability so I can't quickly test my poses. The layout rigs that I was using required certain bones to be unparented, so that made deformation testing pretty much nonsense in Vpaint. Plus you're not really painting in vpaint as much as just clicking on vertex points and assigning a specific value. Honestly I can do a better job just using the old method of assigning weights to selected polys and then doing a weight blur. There are a few other reasons why I didn't find VPaint useful at all but I can't remember exactly what they were but I know that Vpaint deleted any vmap that I didn't edit in Vpaint. So in order to check my work in layout I would have to first edit each and every Vmap and exit VPaint completely, which is not a useful workflow of course.

I understand there was at one time an ability to use Vpaint in Layout but I never was able to get that to run over there.




As for normal maps. If you are using them for all your edge looks then you are missing the power when used with vertex normals. Here is a quick read and video showing that more control over the model when using both. http://www.polycount.com/forum/showthread.php?t=107196 Now don't mind the vertex count since MAX does not create Vmaps like other packages so it just adds extra verts. This means the verts are always discontinuous along UV seams and hard edge vertex normals at the point construction level. Hence why they having smoothing groups to get around this. Having a construction vertex with stored vmap sets keeps your model as efficient as possible by not increasing the the construction points and storing the delta offsets for UV's and vertex normals. Hence why in Lightwave's UV panel you can see the red dots of this delta on discontinuous points. The red dots are the offset seam within the vmap and not extra construct vertices. This is one area that Lightwave needs is the ability to assign vertex normal values. It already can assign, display and render them if they come in as a FBX or OBJ model. All we are missing is the ability to assign them in the VMap.

-M

I guess if you need razor sharp creases or completely flat polygons then vertex maps would be nice thing to use. I think I would have to retry that technique in maya or something to fully appreciate it. I just never saw the point because I usually have a slight bevel on my edges that comes through from the normal map.

MSherak
01-15-2015, 06:11 PM
I hope Vpaint isn't reason Newtek hasn't brought these kinds of tools to a better standard. I've tried that approach a while back where you can import your bones from layout (I'm not a fan of skelegons) using a very old script from DStorm. Inside VPaint, the skelegons have no IK or key framing ability so I can't quickly test my poses. The layout rigs that I was using required certain bones to be unparented, so that made deformation testing pretty much nonsense in Vpaint. Plus you're not really painting in vpaint as much as just clicking on vertex points and assigning a specific value. Honestly I can do a better job just using the old method of assigning weights to selected polys and then doing a weight blur. There are a few other reasons why I didn't find VPaint useful at all but I can't remember exactly what they were but I know that Vpaint deleted any vmap that I didn't edit in Vpaint. So in order to check my work in layout I would have to first edit each and every Vmap and exit VPaint completely, which is not a useful workflow of course.

Not a fan of skelegons?? So you model all the bones in Layout??

For game modeling there is a fixed number of vert influences that an engine can do before you start bogging it down. So weight blurring without that control to only the fixed number is kind moot. Vpaint will do that for you.. Also most game models do not use TONS of joints since those transforms are again expensive at run time. So most of the time if you are making a model the skeleton and geometry is done before animation is ever added. Which means you can load up the model and start weighting it way before hand and you don't need a rig to check your deformations. There is a large difference of what is handed to game engine vs rendering. Sounds like you are talking about weighting in general.

Even the current AAA (ps4/ps3) title game I am working on does not go over 96 joints for everything. That is whole body and face and we don't do over 4 point influence. So no real special tools needed so long as you can assign weights. It takes a keen eye and a long time to get those weights right for 99% of animation going to be thrown at it. That is why the character team takes months on planning models.


I understand there was at one time an ability to use Vpaint in Layout but I never was able to get that to run over there.

Still can in layout. Add to hotkey or menu. Select and object and open vpaint.

-m

sami
01-16-2015, 01:18 AM
Not a fan of skelegons?? So you model all the bones in Layout??

For game modeling there is a fixed number of vert influences that an engine can do before you start bogging it down. So weight blurring without that control to only the fixed number is kind moot. Vpaint will do that for you.. Also most game models do not use TONS of joints since those transforms are again expensive at run time. So most of the time if you are making a model the skeleton and geometry is done before animation is ever added. Which means you can load up the model and start weighting it way before hand and you don't need a rig to check your deformations. There is a large difference of what is handed to game engine vs rendering. Sounds like you are talking about weighting in general.

Even the current AAA (ps4/ps3) title game I am working on does not go over 96 joints for everything. That is whole body and face and we don't do over 4 point influence. So no real special tools needed so long as you can assign weights. It takes a keen eye and a long time to get those weights right for 99% of animation going to be thrown at it. That is why the character team takes months on planning models.



Still can in layout. Add to hotkey or menu. Select and object and open vpaint.

-m

This. +1. I concur. Etc

To elaborate: The 4 pt influence thing is a gpu performance constraint in most cases, where if you have a skeleton/rig where some point(s) (usually in the armpit or hip areas) have more than 4 maps influencing them, say for example 6 maps, then the game engine (or gpu code beneath it) will likely throw out 2 of the 6 bone influences for that specific point, and then average the. 100% weighting among those remaining 4 bones - leading to crappy/inaccurate deformations. This is precisely why your rig for game export has to be as lightweight as possible and that you ensure no additional LW Layout features like joint compensation or muscles are used. This is also why we created that MapMonster/WeightWatcher script - to find and fix these weighting issues in errant points easily. But that happens in the weight tweaking phase,but yes, Vpaint makes a great first stab at creating all the necessary maps needed to do all this. And RebelHill's game rigs are an awesome for this and can easily be modified.

So while I agree that newer, smarter,more user friendly weighting tools are needed, everything basic you need is there - you just need the arcane know-how.

Unfortunately, our attempts to commercialize such smart automatic and user-friendly weighting and vertex map tools like MapMonster that we use internally have all but been ignored, and seemed invisible to any responses from the forum or even recently when I asked Lino directly for feedback. Beginning to think it's my cologne. ;-) but yeah people see the need for something better than Vpaint, but for the life of me, on repeated tries we just can't drum up any traction at all from the community to guide our release of such tools, so we just assume we're the only people who have such game needs with LW and just use our weighting tools internally since they are not yet polished enough for release and no one seems to be interested in them....


http://forums.newtek.com/attachment.php?attachmentid=120152&d=1392247442

lightscape
01-16-2015, 01:24 AM
MapMonster that we use internally have all but been ignored,.

Haven't seen it. Did you post it where it could be easily seen in the forums?

jasonwestmas
01-16-2015, 07:18 AM
Not a fan of skelegons?? So you model all the bones in Layout??

I draw out, rotate and then use a script to orient the bone axis, similar to what one would do in modo, max or maya etc. You seem surprised.




Also most game models do not use TONS of joints since those transforms are again expensive at run time. So most of the time if you are making a model the skeleton and geometry is done before animation is ever added.
Which means you can load up the model and start weighting it way before hand and you don't need a rig to check your deformations. There is a large difference of what is handed to game engine vs rendering. Sounds like you are talking about weighting in general.


I'm thinking more about characters that have to carry gear and have complex topology. So formulas and foresight is not always possible for me. I can see how you can create a weight mapping formula for some topology between characters that are similar but I'm thinking of situations where each design is pretty unique. I am also aware of the bone and weight limitations which makes it even harder to weight map a believable skin imo.

-m[/QUOTE]

You remind me of the people that used to tell me that interactive feedback wasn't very useful or necessary for rendering. ;)

jasonwestmas
01-16-2015, 07:59 AM
So while I agree that newer, smarter,more user friendly weighting tools are needed, everything basic you need is there - you just need the arcane know-how.

Agreed, and also these days it's not all about possible or impossible but what is more efficient and interactive.


Unfortunately, our attempts to commercialize such smart automatic and user-friendly weighting and vertex map tools like MapMonster that we use internally have all but been ignored, and seemed invisible to any responses from the forum or even recently when I asked Lino directly for feedback. Beginning to think it's my cologne. ;-) but yeah people see the need for something better than Vpaint, but for the life of me, on repeated tries we just can't drum up any traction at all from the community to guide our release of such tools, so we just assume we're the only people who have such game needs with LW and just use our weighting tools internally since they are not yet polished enough for release and no one seems to be interested in them....


http://forums.newtek.com/attachment.php?attachmentid=120152&d=1392247442

I would be very interested to give something like that a try and not just for game models. I actually prefer to use weight maps only. Far more accurate.

meshpig
01-17-2015, 03:40 AM
Well this is the thing, indie game engines are %99 for hobbyists and auto-didacts. A bit like the Fender Stratocaster or the .303 British in WW1 :)

lightscape
01-17-2015, 04:30 AM
http://forums.newtek.com/attachment.php?attachmentid=120152&d=1392247442{

map monster looks a lot like the skin weight table for max.
What are its features?

MSherak
01-17-2015, 12:12 PM
I draw out, rotate and then use a script to orient the bone axis, similar to what one would do in modo, max or maya etc. You seem surprised.

You remind me of the people that used to tell me that interactive feedback wasn't very useful or necessary for rendering. ;)


Interesting since in Lightwave, unlike other 3d apps, adding bones or joints in layout does not imbed them into the model which in turn loses ones ability to use them as tools to help in weighting. Until LWDev changes this it just surprises me to not take advantage of them, but again to each their own.

Since trying to keep on topic, Lightwave and game development, I was merely pointing out the limiting rules that one must adhere to when designing a character for a game. The rules apply the same no matter the 3D application. For example in the game I work on. There are 1600 unique players that use one game mesh. Deltas are stored for each unique player along with their specialized items and textures. The characters needed on screen out of those 1600 are generated at load time from the single mesh to the on-screen models. Think of the details, eyes, nose, arm length, ears, body types, hand size, etc. along with additional models of hair types, eyebrows, hats, shoes, etc. fitting to all those details. So in a sense the limiting factors of making a game character have resulted in a complexity that even the best 3D character applications can't handle due to specialization.

All 3D applications are limited and they have to be to cover the majority of what paid consumers need. Specialize in areas narrows your paid consumers. Knowing the limitations of a 3D application and how to take advantage of the them is key in game development.




http://forums.newtek.com/attachment.php?attachmentid=120152&d=1392247442

Wow this looks great. Like having the component editor in Maya. Oh and I love the name to the right of the panel.. LOL!!!

-m

Sensei
01-17-2015, 12:34 PM
Wow this looks great. Like having the component editor in Maya. Oh and I love the name to the right of the panel.. LOL!!!


Modeler, select points, press i, and you can edit point positions, weight maps, uv textures etc. any way you want, one by one.

shrox
01-17-2015, 12:53 PM
I'd say it's up to the artist. When I was doing the AR book stuff, I had to hunt around for obscure plugins to get things exported. I'd guess some wouldn't look as hard as I did. Sure it should be an out of the box thing all ready to go, but it wasn't. Depends upon what you are willing to do, as always.

jasonwestmas
01-17-2015, 02:02 PM
Interesting since in Lightwave, unlike other 3d apps, adding bones or joints in layout does not imbed them into the model which in turn loses ones ability to use them as tools to help in weighting. Until LWDev changes this it just surprises me to not take advantage of them, but again to each their own.

Since trying to keep on topic, Lightwave and game development, I was merely pointing out the limiting rules that one must adhere to when designing a character for a game. The rules apply the same no matter the 3D application. For example in the game I work on. There are 1600 unique players that use one game mesh. Deltas are stored for each unique player along with their specialized items and textures. The characters needed on screen out of those 1600 are generated at load time from the single mesh to the on-screen models. Think of the details, eyes, nose, arm length, ears, body types, hand size, etc. along with additional models of hair types, eyebrows, hats, shoes, etc. fitting to all those details. So in a sense the limiting factors of making a game character have resulted in a complexity that even the best 3D character applications can't handle due to specialization.

All 3D applications are limited and they have to be to cover the majority of what paid consumers need. Specialize in areas narrows your paid consumers. Knowing the limitations of a 3D application and how to take advantage of the them is key in game development.
-m

Yep, I get that. . . except not all game designs require 1600 unique players at once. Some games can have more flexible boundaries than other games because of certain scopes and how much needs to be visible at once. . . and then there is the target hardware to consider. But again, point taken, reusing the same topology and maps whenever possible is a wise decision no matter what.