I've looked up tuts and followed them, but seem to still be having issues. I'm just learning about creating maps since its always been a bit of a confusing thing for me to do.
I created it in zbrush. I have the map size at 4096. The normal map options are set to tangent/smoothUV/Snormals. Once i create it i thought i would get the blue texture map with what seems like indents set onto the image. Instead i get a plain blueish purple image. A tut i saw seemed to have the same image so i figured this wasnt an issue.
Heres an example of that image, and the character in zbrush detailed at second to last highest setting, and the character in maya after setting up the map as i followed in the tut. As well as the bump settings.
Is there something i'm doing wrong? The uvs seem fine. I checked them several times, and doesnt seem to be the issue. If theres anymore info needed to decipher the issue ill be glad to post it. Thanks in advance
Replies
My only explanation is that you're maybe baking from your high-poly to your high-poly, so there is no data to be put into the map. I don't know if that's true, but it's all I can think of.
Looks like this to me too. When you get empty purple normalmap, then that means you baked nothing.
@Bartalon: I apologize for my ignorance, but this is one of the things about UVs/map creation is the whole concept of the UV 0 to 1 areas. Heres a screen cap of my current UV layout. This includes the full bodu, and accessories.
Your UVs are not touching nor overlapping the 0-1 UV edges, which is good.
But you have a lot of wasted UV space. You should pack all these into the lower-left quadrant, the square from U0/V0 to U1/V1.
Try to keep the spacing consistent between the UV islands, at least 4 pixels, to avoid seams.
We have some info here about good UV workflow.
http://wiki.polycount.com/TextureCoordinates
But they do extend beyond 0-1, don't they? It looks like he has the face in the 1-2 range. That would definitely cause an issue in zbrush at least, and have xnormal ignore them.
(Edit: Nvm, just read the next sentence again. I thought it was initially just about packing).
also the lower left quadrant and lower right are both separate objects. If all the UVs were packed into one quadrant it wouldnt matter if they were all separate geometry? would it still work properly? Again i know this must seem like very noobish questions, but ive been struggling to understand this for a while. Thanks again for taking the time to help
Otherwise, if you want them on the same mesh using the same texture, scaling things down proportionately should do the trick.
What can be the problem? Is it a shader setting?
Joopson may have a point here.
ZBrush utilizes your lowest subdivision (specifically it's UVs) as a canvas onto which it bakes your high-res sculpted details. ZBrush forces you to be in your lowest subdivision (sDiv 1) in order to create a normal map, but if you have little to no difference in topology between your high and low, the normal map also has little to no information to transfer.
Your lowest and highest subdivisions should have a drastically different point count. If they are close to the same number or exactly the same, you may be doing what Joopson said.
I have however been creating the map at subdiv 5 since i want the higher detail which isnt a while lot, but around the eyes/cheeks/nose/throat there is detail i want to retain.
I assume it will... If that's the case, then the curvature you want to retain from subdiv 5 will have to be modeled into the low poly. There's actually a thread floating around here for facial topology:
http://www.polycount.com/forum/showthread.php?t=80005
Generally, yes: sculpting is all it takes. However, you said you were baking from sDiv 5 in order to keep certain curvature; I assumed you meant you wanted to keep this curvature for your low poly.
In ZBrush, that's pretty much all there is to it. It makes a normal map from your lowest and highest sDivs with the click of a button. The low poly you have in Maya should be the exact same model that you are using in ZBrush to bake normals (which is often sDiv 1).
When the slopes are similar, the normal map will be very plain-looking, and when it is applied to the lowpoly model it will look like it did before being normal mapped.
You use a normal map to grab all those detailed slopes from the highpoly model.
You use subdiv 1 or 2 because game models generally need a low triangle count, to keep a fast framerate in-game.
However, subdiv 2 is four times the triangle count of subdiv 1!
If you don't want to create a re-topo from scratch, you can export subdiv 2 and selectively choose which edge loops to collapse, which will give you a better triangle count. Then you can use this as your low-poly model, and bake it in Maya or Xnormal or 3ds Max or whatever (because Zbrush won't bake from your selectively-collapsed model, unless you do some messy re-projection jiggery).
Also, Zbrush won't give you a flawless normal map if you are eventually viewing it in Maya. This is due to something called the "tangent basis", which is generally different for each normal map baking tool. If you want the map to look great in Maya, you need to bake in Maya to get that, not in Zbrush.
here is the zbrush bust