ZBrushCentral

Zbrush to Maya, rendering displacements with Mental Ray.

Hi, please bear with me as this is my first post on zbrushcentral.

I’ve always had problems when attempting to render displacements in Maya. I think it’s only ever worked for me once and I’ve searched around many places, looked through many different sources and have never come across one solution that works.

Now to the problem, I’m currently working on a human head sculpt and if you look at the images, the single head is what it looks like in Zbrush. In the image with the two heads, they were rendered in Maya with Mental Ray, the one on the left with a 32-bit displacement map the one on the right with a 16-bit one both of which were exported with MD3 within Zbrush.

If you compare the images you can see that there are a lot of differences. The head rendered in Maya with the detail has an alpha gain value of 16 with an offset of -8 (the 16 came from the alpha depth factor in Zbrush), but you can see that it doesn’t quite look like the Zbrush one.

The image of the ear shows problems I have over random places across the mesh, I think it has something to do with the AUV tiles, UV method I used to create the UV’s in Zbrush (because I started in Zbrush there were no UV’s and the lowest Subd level of the head I have is 25K polys which is a little impractical to UV in Maya).

I know that using certain brushes can affect rendering in Mental Ray, and as I’ve found Maya 2008 (which I’ve used here) doesn’t render displacements as well as version 7.

I know I haven’t provided much information on the whole workflow but if anyone has any suggestions then I’ll welcome such advice.

did you look into this thread

http://www.zbrushcentral.com/zbc/showthread.php?t=56061

also make sure you have your filters off on your displacement file.

when using a 32BIT map you should use 2.2 in Alpha Gain -1.1 in Alpha Offset.

for 16BIT play with those settings. It looks like you may have your settings too high on your 16BIT Alpha Gain and Alpha Offset

I’ve been following the method outlined in Scott Spencer’s “Zbrush Character Creation” book (as well as having knowledge from other sources like Gnomon and Digital Tutors).

No filters are on and the 32-bit head does indeed have an alpha gain of 2.2 with an offset of -1.1. That’s the reason I then tried the 16-bit map, it gives me more of the detail but looks quite bubbly at the same time, I can’t seem to get the fine rake details, even applying the displacement to the bump channel doesn’t bring them out.

I think the seam problem is fixed though.

hello tesh

alright then. what are your settings in alpha gain and offset for the 16BIT. From the image it looks like those settings are too high and that is sometimes what can cause the bubble looks.

What is your approximation editor settings?

What were your settings for the displacement in the exporter?

Can you put an image of your displacement up as well?

talk to you later

paul

Okay, for the 16bit, the alpha gain is 16 and the offset is -8 (the 16 comes from the alpha depth factor that is stated in the Zbrush alpha palette).

Subd approximation editor is parametric with 3 for N divisions (which I think subdivides the model to 3.5 million tris as mental ray states when rendering). The Zbrush model at the highest setting is around 1.5 million quads.

Multi Displacement 3 settings in Zbrush:

MaxMapSize = 4096, MapSizeAdjust = 0, DpSubPix = 1.

D32 profile which also exports the R16 16bit map.

Channels = 3, Bits = 32 float, Vertical Flip = Yes, Scale = A. D. Factor, Smooth = No, Seamless = No, Channels 1, 2 & 3 Range = Full Range, Channel 1, 2 & 3 Res = Full.

As for the new images, they’re of the 32bit head with increases in alpha gain (and the negative half offset) from 16, 32 and 64 (I get the same results with the 16bit map). The side image shows that the displacement in areas resembles a bit of a bump map effect where the detail has just been raised one layer as such. I know that you won’t get the detail that you get in Zbrush but the displacement can be fine tuned better.

The image of the 16bit tiff that I’ve uploaded (doesn’t really make much difference if I convert it to .map format) is a scaled down, compressed jpg. THIS image is after Auto Levels have been applied in Photoshop (a TD once gave me that advice). The original tiff that I’m using in Maya just looks an even grey all over, both 32bit and 16bit versions.

Hello Tesh

I would have your DPSubPix higher then 1. I always use at least 2. Try turning the DPSubPix off and use adaptive and see if you get better results.

Let me know what results that get you.

Adaptive is found in the Displacment Palette of your Tool Palette

Sorry, originally I used a DPSubPix level of 3 (because I assume the value is linked to how many times you subdivide a mesh). After your reply I exported various displacements using DPSubpix 2, 3 and 4 and even exported displacements the old way. via the menu that exists with all the tool stuff. I tried using DPSubpix and also adaptive, it doesn’t matter how I’ve exported the map I just can’t get the detail out of the renders. With the 32bit map, increasing the alpha gain has no effect whatsoever (because with 32bit maps I think the object scale is baked into it).

I’ve uploaded a few images of something strange, the UV’s in Maya are all separate squares whereas I thought that using AUV Tiles in Zbrush creates UV’s that are next to each other. The exported map certainly supports this notion (this zoomed image has auto levels applied so you can see the detail). The thing is when rendering, the mesh doesn’t show that it’s missing any tiles so I suppose there’s nothing wrong here. I’ve also included an image of the material I’ve applied from the hypergraph.

Just to make sure, this is the workflow I do in Maya.

  1. Import the OBJ exported from Zbrush Subd level 1, make sure in the import options “create multiple objects” is off.

  2. Rotate the mesh so it’s facing +Z.

  3. Turn off feature displacement in the shape node.

  4. Create a blinn shader and right click “assign material to selected”.

  5. MMB drag a displacement node onto the blinn, and map it to displacement map.

  6. Load the .map (or tif) displacement in the file node. Turn off filter and input an alpha gain of 2.2 and alpha offset of -1.1. “Alpha is luminance” is off.

  7. Open the mental ray approximation editor and create a subd approx, put it on parametric with N = 3 (sometimes I use Spatial with min n = 1, max n = 5, length = 0).

  8. Hit render, with mental ray as the default renderer.

Just to make sure, in the shape node I entered 3500000 into the max poly count (if that makes any difference?).

Anything I may be doing wrong?

tesh

how many subdivision levels do you have on your mesh? It looks like everything is good. Did you save a morph target? Do you have the original mesh before you subdivided. can you give me your tool so I can do some tests? Let me know.

Paul

Subdivided 3 times I think, 25k is the start, then 98k, 400k, 1.5 million. I made this sculpt from a polymesh sphere in Zbrush so there is no original mesh that I exported out of Maya. I could give you the tool but as far as I’m aware I can only post 300kb files here. I might just leave it and move onto another project and see if I can get a render out of that.

tesh

try another model and let me know, but if you did want to get the other one to me I do have a FTP site

Paul

Hay,
i m also having the same problem i do not get the finner details and when in may do i have to subdiv the mesh and to how many times with me i have to div it for all the four time to get some bubbly effect , as in the tuts i see that no div or only one is enough for to get a good fine result like in the Rhino tut Can i get a proper step by step way of generating a dismap and rendering it in maya pls
that would be a great help it driving me crazy.
Total Respect

Ronybuster

hey ronybuster

check out the videos in thread

http://www.zbrushcentral.com/zbc/showthread.php?t=61676

Let me know

Paul

thanks a million dude u r doing a grt contribution…

Total Respect,

Ronybuster