Texture workflow

AlexLegg

I was reading up on curavture maps the other day, and it got me thinking are there any other maps like this that are commenly used that I am unaware of.


So what texture workflow does everone use?

And what maps do you bake out/make in Crazy bump/xnormal etc.., and for what reasons (masking etc)?



I guess mine would be:

Bake coloured diffuse from Max (so I can use the differnt colous for quick selections in Photohop)

Bake AO, height, normals, Curvature/Cavity (controversial, I know!) in X normal


Baking note:

When baking maps, bake them with every object together as it should be, then another sets of the same maps, but with the objects indivdually seperated/exploded.

Then in Photohsop I can combine/mask the best bits fropm each render.


Then in Photohop:

Tidy up normals, ugly seems potruding objects etc


Create colour map, mutliplying AO over the top. Maybe A cavity map too.

Use the Curvature map as a mask for grunge.


For the spec, I guess Id re use the curvature map to make shiny, worn edges. Maybe a tweaked occuluse map in crazy bump.


Thats all I can think of for now.

I'd be interested in what everyone else has to say.

AlexLegg

A while back there was a lot of heated discussion about using curvature maps, whether they were useful and what the difference was between them and cavity maps. Not looking to start that debate again!

AlexLegg

NP.

So what does anyone else think?

So what texture workflow does everone use?

And what maps do you bake out/make in Crazy bump/xnormal etc.., and for what reasons (masking etc)?

tristamus

I bake everything from xNormal....

Normal Maps...AO...and Curvature Maps (I use the red channel of it on top of the AO).

And after I'm done creating my diffuse maps, I pass them lightly through crazybump to get a detail normal map, and bake that on top of the normal map obtained from my high poly to subtly enhance the details..

AlexLegg

Thanks tristanus, that very useful to know. Esp. using just the red chanel of the curvature, and creating the detailed normal.

leechdemon

AlexLegg wrote: »

Bake coloured diffuse from Max (so I can use the differnt colous for quick selections in Photohop)

Bake AO, height, normals, Curvature/Cavity (controversial, I know!) in X normal

That's roughly what I do. I assign MatID's to different colors for easy of reference, duplicate all parts of my model (I work with rigged vehicles w/ moving parts) and collapse to a single mesh, bake diffuse. Bake AO, Normals if I go that route (versus generated in PS). After building up Diffuse I'll convert that to a reasonable Spec and Gloss map.

The only other significant thing I do (and it's specific to vehicles) is projection my reference textures onto my mesh using a second UV channel, then baking to my original channel. This makes drawing (tracing) body panels and other details much, MUCH easier. I'll bake from top and side, composite the two together, then trace them for a body panel pass. Once they're baked you can dump UV2 and the duplicated geometry.

Ruz

Yhe problem with baking multiple maps to make your diffuse is if you have a project with a lot of bakes to do, then you get asked to make edits, you have to redo all of it.
I try to stick to ao +ao detail from xnormal plugin for pshop.

leechdemon

If I understood Alex correctly, he's saying that he's painting pieces specific colors so that he can easily grab them in PS. For me, I assign each section of my object a different Mat ID, put a MultiSub material on it and make each ID a drastically different color. That way when I bake it, I can look at the blue chunks in PS and go "Oh, those are the windows", look at the Black and go "Oh, that's the tires", etc. It takes a couple of minutes to paint them all up, but then they're in Photoshop, you can select each chunk and split it to it's own layer, and now you've got an easy layer mask for each section of the map.

Is that what you were talking about, Ruz? Or something different?

bgoodsell

I haven't herd of using the curvature map (same as the cavity map in xNormal i assume) for use in the diffuse map. What exactly does that help do?

For me, I use Photoshop CS4's 3d feature to do most of my texturing. I bake out all my maps and save out a low poly .obj file and create a 3d document with no Photoshop lighting so i can quickly tab between my 2d texture and how it goes on the 3d model. This saves me time in trying to A) generate a UV snapshot as Photoshop can generate them for me, and trying to figure out what UV chunk goes to what on a model. I can just create a 'paintwhat' layer and color some bright color on it.

Specifically to texturing workflows, I work form big to small details. The flow usually goes something like this:

-Import and combine normals, get AO and UV overlays setup
-Create a base color layer and get major color regions blocked out (without AO on)
-Turn AO on, get detailed areas colored blockedout
-paint in major wear or scratches
-overlay textures or begin pushing highlights / shadows if i modeled the texture into my highres bake
-work on texture until satisfactory results are achieved (or the suck is eliminated)
-export normal and difuse into engine
-work on spec which usually involves retooling areas to achieve correct material feel
-post on polycount and get critiques
-refine and polish

Concerning bakes, is a height-map really necessary? I normally just use a diffuse and normal and paint in everything else.

r_fletch_r

bgoodsell wrote: »

I haven't herd of using the curvature map (same as the cavity map in xNormal i assume) for use in the diffuse map. What exactly does that help do?

Red areas are concave areas, green areas are convex. soooo. you can use the red channel to easily mask crevices for things like dirt and AO.
The green channel gives you a mask of the models hard edges, this can be used as a base for wear.

Ruz

I do that too leechdemon, but I was just generalising about the whole proces ie make it too complicated and you will sufffer:)

AlexLegg

I assign each section of my object a different Mat ID, put a MultiSub material on it and make each ID a drastically different color

Yep that's what I meant really.

bgoodsell, that's interesting about using PS for the 3D side of things. I've only played with it briefly but found it a bit clunky! Maybe I'll need to play with it a bit more.

Concerning bakes, is a height-map really necessary? I normally just use a diffuse and normal and paint in everything else.

Well the height map is mostly if I was thinking of doing a parallax map, but could also be used to some effect in the colour map.

And i'll get in there before Eric and provide a link to the wiki for curvature map:

http://wiki.polycount.com/CurvatureMap?highlight=%28%5CbCategoryTexturing%5Cb%29

bgoodsell

AlexLegg wrote: »

I've only played with it briefly but found it a bit clunky! Maybe I'll need to play with it a bit more.

Its worth noting that the 3d paint feature is in nowhere near as good a shape as its zbrush equivalent, but for previewing how a texture looks on the model, it does the job. I have a pc from 2008 and it seems to update with no lag once you disable the lighting photoshop has (which is useless for this anyway).

r_fletch_r

I like to do the following
Bake: Normals, Mental Ray FG, Cavity AO, Curvature, diffuse label.

Normals: Obvious
Mental Ray FG: Gives light tracer style 'AO' but is much faster. down side is it lacks the fine detail you get from light tracer.

Cavity AO: low distance AO, captures all the fine details and can be solved at very low sampling: 8 samples often is enough.

Curvature: Nice base for a wear map.

Diffuse label: what you guys have mentioned already, a load of masks.

A well structured PSD broken into masked layers can make iterating much easier.

Say:
base layer
paint layer
dirt layer

changing, paint coverage, paint colour, dirt density/colour is a quick easy process. also using the same masks its pretty simple to propagate chanages into gloss and spec maps.

[HP]

Lemme just drop a tip in here. I was chatting with Hanno in the Polycount dinner the other day and he mentioned something I never really thought about it before.

Multiplying the AO on top of the difuse might be a bad thing to do, cos it doesnt really occlude anything, it tends to burn the colors (making them darker) instead of giving the texture a convincing darker crevice.
Using the blend-if in the layer style and tweak the opacity of the AO layer will generate better results cos it doesn't mess with the colors of the underlaying difuse map.

r_fletch_r

[HP] wrote: »

Lemme just drop a tip in here. I was chatting with Hanno in the Polycount dinner the other day and he mentioned something I never really thought about it before.

Multiplying the AO on top of the difuse might be a bad thing to do, cos it doesnt really occlude anything, it tends to burn the colors (making them darker) instead of giving the texture a convincing darker crevice.
Using the blend-if in the layer style and tweak the opacity of the AO layer will generate better results cos it doesn't mess with the colors of the underlaying difuse map.

Or if your shader supports it, use it to mask the ambient light, and get it out of the the diffuse.
Mental Images MasterZap did a write up on correct use of ambient occlusion for realistic rendering
http://mentalraytips.blogspot.com/2008/11/joy-of-little-ambience.html.
I know were not using a raytracer but knowing how it should work is valuable

AlexLegg

[HP] wrote: »

Multiplying the AO on top of the difuse might be a bad thing to do, cos it doesnt really occlude anything, it tends to burn the colors (making them darker) instead of giving the texture a convincing darker crevice.
Using the blend-if in the layer style and tweak the opacity of the AO layer will generate better results cos it doesn't mess with the colors of the underlaying difuse map.

Oh that's interesting, I'll have to try that out and see if it does yeild better results.

RedRaven

Hmmm, that IS interesting, as multiplying the AO over the diffuse has pretty much been how I've gone about my texture workflow since the beginning.
I might have to play around with it now, and I need to have a read of that blog post by Zap. I don't use mental ray for my AO bakes, usually just go with xnorms for the quick and easy, but, eh.

tristamus

RedRaven wrote: »

Hmmm, that IS interesting, as multiplying the AO over the diffuse has pretty much been how I've gone about my texture workflow since the beginning.
I might have to play around with it now, and I need to have a read of that blog post by Zap. I don't use mental ray for my AO bakes, usually just go with xnorms for the quick and easy, but, eh.

Same here....always have gone about it that way, but I definitely won't oppose a new workflow if it produces better results!!