Texturing a head from a plane

[pablete]

Hi guys! I think this is my first post (and I say I think because I’m not sure, I actually registered long time ago and don’t remember if posted anything already)…
OK so here’s the thing. I created a head in Maya and I’m trying to texture it from some photos of myself, in Zbrush, with the Projection Master. The thing is I know how to do it and everything goes fine and I get my texture in psd format and apply to my model in maya and it looks relatively fine, but there’s something I’d like to better understand about textures and sizes and projection master, hence the creation of this thread…

The steps I follow:
I change the zbrush doc size to 2123 x 2123. Import my .obj head into zbrush, divide it up to sdiv 6. Assign a fast shader. Isolate just the head, from the chin to the line between forehead and hair. Assign an 8192 x 8192 texture. Launch projection master, just color. Create a new plane with a texure of my photo (size 2123 x 2123). Rotate/Scale/Move to fit the head model. Turn off Zadd. Then Edit-move and start tweaking very slightly the image to fit my head. When I’m done, I pick up the projection. Then export the 8k texture and open it in ps. What I find is that the image is a lot less defined (I’ve lost lot of detail in the skin, which is somewhat blurred) than the original photo. Is it ok? am I doing something wrong? I’ve tried picking it up with and without the fade option selected.
The UVs are correctly laid out, and the size of the head uvs (from chin to hair) in the 8k texture is roughly the size of the face (again, from chin to hair) in my photo -2123 x 2123-. I can’t make larger photos, at least not for now, with my nikon 4300 (4mpx)

If anybody could give me some further insight on this issue, I’be really grateful.
I really appreciate your help
cheers
pablo

[eldee.s]

well, for starters you’re up-rezzing, which will always create artifacts. zbrush doesn’t magically sharpen the quality of your photo-source when you save it as a larger texture zbrush has to resample all of those pixels to fill a larger space, AND fit within your UV coordinates. You should never ever ever need an 8k texture for a human face. a 4k texture is sufficient for a 16-million polygon mesh, you wouldn’t need an 8k texture unless you’re up in the 64 million range. Which, with today’s hardware, is a bit ridiculous. Depending on how big your base mesh was, with a 6-subd face, you could probably even get away with a 2048 texture.

[pablete]

thanks for your reply eldee.s.
But here’s where I get really lost, I mean, when it comes to sizes and rendering.
I though I was doing it right, I mean, the texture in zbrush is 8k, but the uvs I’m painting onto just take roughly 2k of the 8k texture, so it shouldn’t be any problem, right?
If I use a 2048 texture the uvs on the face would take aprox. 500 pixels, so then if I tried to render a first plane of my face at 4000 pixels, for example, the texture would be somewhat blurred, right? I really don’t know if I got it right or not (probably not…)

[eldee.s]

if your head is only using that much space of an 8k texture, you might consider using multiple UV tiles… with a texture that large you’re probably wasting a ton of space between islands, if you separate areas out into different UV spaces, you can use 3 2k textures (or maybe 2 4k textures), which still amounts to less than half of an 8k texture in terms of memory.

however, this is all beside the point of your post (but a helpful tip!). If you’re within the pixel->poly ratio, the image ‘distortion’ you’re seeing is likely just the interpolation between 3D and 2D. Your uvs will always have at least a little bit of stretching or compression on each face, and in some areas there will be more than others. When you use zbrush to paint in 3D, you’re actually applying the ‘pristine’ texture in 3D, therefore causing an equal amount of compression or stretching in the 2D version of the texture map. This is just the way things are, and it’s actually not a huge deal- if you do things the opposite way around and go from 2D to 3D, your compression or stretching happens in 3D, but your texture map looks nice and clean.

So ultimately, if the final product is supposed to be 3D, you want it to look as good as possible in that final environment… and when it comes to working with the texture in photoshop, you’ll just know that your textures are always going to look slightly distorted compared to what you expect to see.

now- if the problem ever becomes that you’re picking up your mesh from projection master, and you lose detail at THAT point, it simply means you don’t have a large enough texture (or in the case of polypainting- geometry) to support the detail you painted from within projection master.

if i’m off target, or if this didn’t make sense let me know

[pablete]

well now it’s starting to make some sense. The truth is the distortion/ blurring is very slight, but as this is the first time I’m using this technique (in fact this is my first time using Zbrush for texturing) I wanted to make sure I was doing it all right.
I guess I should try with another camera and take some 6k or 8k pics just for testing purposes…
thanks again for you your help eldee.s
cheers!