Crit My Painting Workflow Please!

BradMyers82

Hey guys, this is sort of an unusual thread I guess; but I think I could get some good information here that could really help me if I post this.

Anyways, I have some freelance work lined up, where I will be doing handheld type characters. So of course, that means good old fashioned hand painted textures.

I really want to be sure I am doing things efficiently texturing wise for obvious reasons. So I'm trying to really analyze my own current workflow.
Here is what I do in a nutshell:

-Bake a general AO (from the final low Poly) to get some standard shading, and blend it in photoshop/bodypaint so the forms look right.

-Once I get the AO setup in a .PSD along with the uvs, I make several layers.

+ Color layer (just flat colors on the uv Islands (Normal Blend))
+ Overlay Layer (Grayscale for midtones (Overlay Blend))
+ Multiply Layer (Grayscale for shadows (Multiply Blend))
+ Color Dodge Layer (Grayscale or slight color for highlights (Color Dodge Blend))

When I feel that I have the values down, I throw in some color variation via Overlays or Softlight just above my Original Color Layer.

Finally, when it looks decent, I flatten the layers; do a Unsharp Mask, and adjust the levels and Hue/adjustment if necessary.

Then I pretty much just Brute force the rest until I have the texture looking right (with any type of layering necessary).

I used to use Brute force way too much but now that I sort of have a plan; I am definitely more efficient.


So...

If anyone has actually read this far and can give me any tips / or tricks, chime in on your own workflow, or even general crits I would really appreciate it.

I have a timelapse video here too of me making PART of a Quake 3 texture from scratch (Ranger). I made a bunch of mistakes in it, so its not the most flattering clip; but I think that's good since I'm looking for help. Its about an hour long but sped up to last about 9 min.

http://www.bradm3d.com/BradM3D_QuakePainting01/BradM3D_QuakePainting01.html

renderhjs

I actually never really use black at the beginning - only at the end for the finishing touches. I prefer often the additive painting mechanic in which you start with the brighter colors first especially for the base or grid constructs. The the more and more things evolve I add step by step more contrast.

[Deleted User]

I start with solid colors, midtones first, then do highlights and shadows on top of those (same layer usually, with the paintbrush, soft edges and pressure-opacity) to get the shapes in, then move to texturey brushes for things that need them and do linear details with tiny pressure-opacity/flow brushes. When it looks mostly passable, I start on adjustment layers, overlays, burning/dodging, etc. Then I take it through ZAppLink to fix the seams. If there's a lot of stuff that needs to be lined up and it's hard to tell what's what in the UVs, I start in ZAppLink first and draw lines as guides for the base colors.

BradMyers82

Hey, thanks for the info guys. I guess my workflow is okay then, and I just need to get better by painting as much as possible.

renderhjs: I'll have to give your method a try. I know I do what your saying when doing skin/faces, but it would be interesting to try it on other stuff.

fly_soup: Sounds like a pretty solid workflow. That reminds me, I need to get some more of those linear style brushes you mentioned. It will definitely speed things up versus drawing with a simple soft/hard brush.

dolemite

paint AO
paint mass (white for higher)
paint shadows
paint highlights.

adjust all opacities and then Copy all layers and merge into new layer.

Gradient map overlays for color
pant any color that wasn't captured in the gradient map overlay
paint darkest darks
paint lightest lights
seam fixing (optional)

reiro

Interesting Thread. Iam wondering why you would paint a face from scratch like in the old days, if you can use photos as a base and then just paint in details and customizations.

I guess it depends what you are trying to achieve but isnt painting everything less and less used in the industry? I just dont get this workflow unless you are working for a game with a cartoony, stylized look.

Any opinion welcome

leslievdb

reiro: If you can paint a face without having to use photos as a base you can really make it look how you want it to look. I noticed that when you use photo's you lose a lot of control on what you really want to create because the photooverlays will make there mark in your texture in one way or another. Especially when it comes to characters you have to try and paint everything by yourself and use the photographs only as pure reference when needed.

BradMyers82: your workflow seems to be pretty good, i only tend not to use blending modes to much on my layers themselves, try to blend you colors themselves the way you want them to look and try not to depend on the blending modes too much.
The only thing i use blending modes for is to make those really hard speculars (set your brush's blendingmode to linear dodge (add).

anyway good luck and keep us updated ^^

Neox

I think it totally depends on what you want to do, there is nothing wrong with using photos for photorealistic characters, and if you want to go to a specific direction take photos of people that look different then your model and make them fit...
taking a 3d.sk photo and work after that and then using the initial photos for textures are somewhat obvious and i see it some times in games. But if you take some ref for the forms and others for the textures its way harder to tell the impact and its not like "hey thats that dude from 3d.sk" crysis, ut3 and the latest resident evil where sometimes pretty obvious, crysis did the best job on hiding the ref, but well if you have like 5 black guys with good enough refphotos you see where the artist took his refs. they also didnt use the free sample which resident evil and ut did ^^

leslievdb

no of course there is nothing wrong with using photos, i was actually referring more to the quake 3 example where most of the time your textures will be a lot smaller to paint on.
But when you do use them i think it should be used more in a way of taking a part of skin for example and using that to give you painted skin parts that final push towards the realism it needs

BradMyers82

Hey guys, thanks for keeping the thread alive. I thought it was dead, and haven't really looked at it till now.

The main reason for my workflow here, and not using photo references is because on such small texture sheets (256 and lower) if you photo reference too much it looks like a blurry mess. You have way more control in my opinion when building up by hand painting and to make characters read well that will be seen on a tiny screen, you really have to exaggerate things like the overall contrast and such.

I try to use photo ref. as much as possible when doing current gen / next gen stuff, so yeah; it makes perfect sense for that.

Ravenslayer: I agree with everything you said, but I am still curious why you wouldn't want to use the blending modes too much. I actually try to use them as much as possible, then in the end flatten the layers and tweak as needed. I do this because I feel it give me more control, like I can easily make changes because everything is on different layers.

On second thought, maybe as I get better and make less mistakes your workflow will work better for me though.