Hi Everyone,
Can anyone point me to some good tutorials (I don't mind paying) that show how to create good quality body textures for 3D characters?
I don't have any trouble doing faces (after watching the epic noesis face texturing tute) but whenever I try to patch together body parts (from 3d.sk) I always end up with a really crappy blotchy looking textures with light bits and dark bits everywhere that don't match
When I search online for body texturing tutes, google just comes up with a bunch of tutes that say character texturing but then only talk about the face...
Replies
There ARE colour zones on the body, it varies a LOT more than the face.
Some tips to get you started:
- Backs of elbows, and fronts of knees are generally darker. They tend to bunch up a lot of dead skin.
It is usually 'darker' where hair 'would' grow. The outsides of arms, outsides of legs, back/shoulders chest etc. The sun tends to hit those areas more. Freckles are more visible
Under the arm, inside the elbow, back of the knee are a lot more pale, anre varicose veins are more visible etc.
Paint some moles here n there.
Start with that. Show us some progress plz.
You can see a typical example of the problem I was talking about in the 3D viewport, and if you look in the uv view you can see how I've pretty much given up on the legs entirely and just gone for a simple tiling skin texture. But this just looks stupid so I'm looking for a way to learn how to do it properly.
Fwiw I'm using Blender 2.56 and projection painting to (try and) mix 3d.sk photos of a male subject (arnold) from 3d.sk.
What you have from projected texture is a good base to start from, but if you aren't willing to put in the elbow grease, with the levels, clone, and smudge tools, then nothing anyone will say will work. I don't see anything amazing or hard as far as mesh or texturing needs go.
What painting program are you using? Both GIMP and Photoshop are good options and the toolset is almost the same.
How many hours did you spend cleaning up the textures post projection?
Did you use any of the 'painting' tools in Blender?
Do you use Mudbox or ZBrush?
I would say with the clone tool alone from PS and GP, you could get a pretty basic working and looking texture in under 15 minutes, literally from where you are right now.
After posting I did have the idea to projection paint this texture into a new texture using a second UV map with the seams in different places to make it easier to clone stamp out the terrible contrast?
The one that I really can't seem to get right though is the legs, I spent hours and hours and hours projection painting and repainting from photos at different angles and the more I tried the worse it got
I guess I was just hoping that someone out there had done a tute similar to the one ancient pig did for face painting but for bodies...
http://www.immortal-doll.com/imvu-resources.html
I'm going to see what I can come up with using a simple tiling skin texture and painting in some shadows and hightlights using the templates above as a rough guide.
Do you think John Lasseter and other Pixar guys had tutorials to help them build the awesome things they have today?
Sure they now have a workflow down now and younger people coming in knowing how to do things, but the point is to seriously work things out and spend time and don't be afraid to create something horrible.
Use incremental saves and if you mess up, go back. Also, you've got an undo button available to you.
Use reference for examples of how you want to paint the skin tones.
Use reference...use reference...use reference. <- can't stress this enough!
If you've got a biology book with some naked dude in it, use that. Or I'm pretty sure you can "Google" for a naked dude.
I'm a little tired this morning, so excuse the tone of this message, but I've read so many posts of people wanting free information and when you see the work, they are just not trying hard enough. You've got an awesome base here to expand.
Also, if you're having problems with Photoshop, try taking a picture of yourself and painting a self portrait. This will get you more familiar with the tools and colors.
And don't trace via layers. Put the images side by side and try recreating yourself. This is traditional art lessons.
You are attempting to do something artistic that requires skills you don't have, yet. So you totally have a right to be frustrated and of course its going to look like ass, there are some inherent problems with using photos as skin materials. But more on that in a min I think you have a bigger problem...
Just like a 1st grader doesn't go from stick figures to "Young Lady in a boat" with one drawing.
Ctrl-C Ctrl-V from 3d.sk doesn't make outstanding skin materials without some hard work.
So how did James Tissot go from 1st grade drawings to some of the most amazing art on the planet? How is it that he captures every wrinkle and fold of complex cloth while children make little scribbles and sticks for appendages? What did Tissot see that kids aren't paying attention to.
When a kid goes to draw a person they think "I need a body, scribble done!"
When Tissot looked at a person his mind must have gone into overdrive taking in all the details and dissecting what he saw. This was a guy who didn't have the luxury of ctrl-c-v-z so he had to take his time and understand perfectly what he was looking at.
Learn to ride one bike and you find the rest of them aren't that different.
If you learned what the tutorial was teaching instead of just recreating its steps then you would be able to take what it taught and translate to bodies, (like all of the other tutorials that you've encountered that deal with faces). The lessons that are packed into the face translate to the rest of the body which is made up of the same materials and plays by the same rules.
Observation and recreation
These things are critical in being an artist. What makes a knee look like a knee, why does skin hang like it does on old people, why are some areas lighter than others. What roll does blood flow and circulation play in skin tone? What about tissue depth, health and nationality?
You need to work with the tools more and refine your critical eye skills. Find something that is wrong and fix it. I think using photos is hurting you right now because it keeps you from taking the time to learn what you really need to learn. They are too much of a crutch and they are making it hard to see whats wrong and even harder to learn what you need to learn.
Using photos for skin textures has a few weak spots
The first has to do with the shape of the body parts and how they interact with light. Humans are made up of tubes, tubes are hard to capture with a simple single planar snapshot. There is a part that faces the camera and catches light which will be brighter than the outer edges. The band in the middle can't be trusted, its a highlight, the bands on the outer edges can't be trusted because of the perspective and shadows. So there is a narrow band between them that is accurate, that's a lot of waste and very little gain. You can map the whole shot to your object but when viewed at any other angle it won't work.
To accurately photo map a human you would need to take hundreds of photos from almost every angle imaginable and use only the thin accurate strips from each... what a pain in the ass, which is why a lot of people learn what they need to learn and do materials from scratch.
This is kind of an extreme example but it should get the point across.
See how the light hits the "tube" and creates highlights and shadows. You could take this and map it to a leg, but then when you view it from another angle that doesn't match this, it will look off.
Photo's are the culmination of a few different effects.
You have diffuse color, specular shinny highlights, sub surface scattering and few other minor effects. In order for the lighting and render engines to work with these you have to break these effects apart into separate maps. Your diffuse shouldn't have any specular highlights or shadows in it, your specular map probably shouldn't have too much diffuse coloring. It's hard to separate these effects from a photo, especially given how dynamic they are and how static photos are.
Even with a fairly robust ambient light setup that can wash out shadows and even out highlights, you still run into another problem.Skin scatters light around inside of itself (subsurface scattering). When you look at a tube that is semi-transparent and scatters light inside of itself you the edges of the cylinder will appear different than those that are straight on.
So photos are a tiny band-aid over a much larger issue. I think you should be using photos to observe but not directly use them. It's a bit like doing a jigsaw puzzle but someone comes along and dumps a bunch of other puzzle pieces on your pile. It's just a bunch of noise and info that clouds your picture.
anyway... my 2 cents take it or leave it
Thanks for the awesome feedback, I don't think either of you were harsh at all, if anything it's some of the best advice I've ever had for stuff like this. Apologies for being the typical "show me the shortcut" poster, I wasn't sure if I was over complicating the process unnecessarily.
That sums up my experience so far with an accuracy that borders on x-men style psychic super powers. So I'm ditching the paper mache photo approach and going to start working on learning some digital painting skills instead.
Or you could do something sort of like this http://udn.epicgames.com/Three/TakingBetterPhotosForTextures.html