Hello, I’m trying to figure out how many textures should be applied to map some high resolution character models for film. I’m not interest in making it look photoreal, I just want some great detail.
If someone could give me some pointers I’d really would appreciate it! 
If I were to use two UV maps on a character with two 4096x4096 pixel textures, would that render faster or slower in an animation package, opposed to mapping with four 1024x1024 pixel textures or eight 512x512 textures?
Actually, one 4096x4096 = 8 1024x1024 or 64 512x512 textures. Surface area of a texture is squared, not linear.
To answer your main question, however, there is no hard and fast rule. The size of the texture that you should use is the smallest size that will support the amount of detail that you want to offer at the distance that any given part of the model will actually be viewed from.
In other words, a model that is 100 yards high in world terms can get away with a texture that’s only 64x64 if the model is far enough away from the camera.
In terms of rendering time, a 4096x4096 will require as much in the way of time and system resources as four 1024x1024 textures. But that’s not usually how you’d break the model up. A workable practice, for example, would be to use a 2048x2048 texture for the head, 1024x1024 for the hands, and 1024x1024 for the body. This would work, if your animation is going to have close-ups of the head and hands, but medium shots of the body. Also, ask yourself if any parts of the model aren’t going to be visible to the camera at all. If not, they don’t need texture. In fact, they don’t even need polygons!
Also, here’s another factor to bear in mind since you brought up photorealism. 4096x4096 for a head or body texture generally is considered to be photoreal. For the head, it offers enough pixels per polygon to include details such as skin pores and really fine wrinkles. But such details are only necessary if you’re zooming in close enough to see those pores. If not, then you’d want to go with something like 2048x2048. Or even lower.
Super Answer! Thanks Aurick!
Ive used Zbrush for displacement detail on 3 films. For mid ground characters you could maybe get away with:
2 2k maps for the head
2 2k maps for the torso front and back
2 2k maps for the left and right arms
2 2k maps for the hands.
2 2k maps for the legs left and right
2 2k maps for the left and right feet.
12 2k ( 2048x2048 ) maps at least.
You will also use this space for you color maps as well.
Again this would be for a character that could hold up say three quarters of his body in frame. If you are going closer you will need more texture space. For a hero model that number can go as high as 150-200 maps 1k-2k. For background characters you could use 1 texture map
This is why Zbrush needs to support painting across multiple maps. The average amount of 2k maps for characters you would use will be at leats 25 or so. Zbrush should support the needs of Film, TV, Games and home hobby users.
(skycastle) I would really like to hear more about your process for creating “hero” character textures, like what you use to as far as software, texture management etc… This stuff really interests me I could talk (or ask questions I should say ) about it for days.
Thanks for the informative answer SkyCastle!
I have a few more questions if you don’t mind.
A 2K map = 2000 pixels squared right?
When you say an average of 25 maps for mid range characters, do you mean 25, 2K textures or do you mean 25 map projections (hands, fingers, feet, toes, legs, torso ,head, etc) on 1, 2k UV Texture?
Thanks!
And I’m pretty positive that skycastle means 25 seperate 2K maps and a 2K map is 2048 x 2048 square map. Most textures are made at a power of 2.