this is my attempt in 3.5.1… i am having real issues in version 4.0. same setup but tottaly messed up renderings. Anyone please post those issues in a new thread i made.
Nevermind it works with 4.0 my mistake on the settings
For the renderman tests that i did i first had to convert the tiff from the comand line to keep the image 16bit (txmake -float file.tif file.tx) this made the steps go away. The second part is to use a remap node between the displacement node and the image file node with a new min val of -.5 and a new max val of .5. This will put the 50%grey back on the normal.
As the file node will be a float make sure that you use the red chan as the default is set to alpha. Hope this helps
After a good few hours playing around with Pixolator’s demo file, it suddenly struck me why it seemed to be so difficult to get a good result out of C4D.
The displacement file supplied is set to have 50% grey as a base point, while white is ‘high’ and black is ‘low’; C4D’s displacement works on black as the base point, and the range–going up through grey to white–indicates higher points. So no amount of messing with constrast, gamma, etc. will actually give the correct results.
To test this, I did the following:
I then used the displacement map filter, with constrast, gamma, etc. all at the default settings, but using the clipping values to only use the values of 50% to 100% (mid grey to white). Displacement set to 8.34m (taken from the source thread as the displacement-gain of the map), and at 100% strength, this gave me a render with the ‘high’ displacement areas.
Same again, but clipping the values from 0% to 50% (black to mid grey), gave me the ‘low’ displacement areas.
Then, by combining the two (fairly roughly, I admit) we get the combined, ‘correct’ displacement.
This is, IMHO, a much better match to the intended look.
Until C4D does displacement from a 50% base, displacing up and down in one go, or we get a map out of ZBrush that is from a black basepoint, C4D will never be as accurate as some of the other renderers.
Sounds like a new displacement plug-in needs to be written 
Anyone?
Creed - ZB2 is capable o producing seperate +ve and -ve displacement maps. I would think best results in cinema would be achieved by adding two mats, one to displace outwards, one to displace inwards?
I can’t see Maxon changing the system to work in both directions from a mid grey even though that’s what everybody else does! LOL - Baz
Would it be possibel to get this map in a +ve and -ve version, to see if it is any better?
Now I need to figure out a way to make it easy to combine the two renders so you can’t see the join. It seems like some sort of alpha-channel derived from the displacement map is a good way to go…
But yes, it would be nice if C4D did displacement maps with 50% grey as base.
Pixolator said in another thread that he would post some more models for displacemnt tests. I requested seperate +ve and -ve maps for cinema users so hopefully he’ll find time to produce them for 1 or 2 models - Baz
Turns out you can get a fairly good result with mixing textures and creating a +ve and -ve map by inverting and clipping the supplied map…
Rather than explain what I’ve done, I’ve put the file up here. (If anyone else wants to host this, feel free.)
Thanks for the files Creed! I assume you did this in Photoshop or similar app? I’ll have a go at this myself but I’m not sure how to go about it so if my experiments don’t work out I may get back to you! - Baz
Yeah, I used Photoshop to invert the second version of the displacement file (for the low areas), and re-save both as 16-bit PNGs, which seemed to work better in C4D. Other than the re-save and invert, I did nothing else to the files at all. The render above is straight out of Cinema, with no post-processing at all.
Any image-editing program (e.g. GIMP?) ought to be able to do what you require.
c4d is not the only program used by the people here which uses 0 as “midpoint”. renderman also uses 0 as midpoint. but due to the rsl it’s not such a big problem for renderman-users to adopt and use maps which use 0.5 as midpoint.
I played around with the displacement map some more using Renderman/Rat 6 in maya, and found that using a Combine Displacement shader with a base value set to -.5 would do the trick.
Thanks for the info Creed!
I’m not sure if you guys are aware of this, but you can just reload the supplied displacement map as a ZBrush alpha and apply a positive and negative curve by loading the appropriate .ZCV file in the graph. You can probably alter it to have a black starting point as well. Just make sure to have the ep ( Export Processed Alpha ) button checked when you save out the modified version(s).
Thanks Creed for the quick idea on splitting the displacment map into two parts.
I was working on a parallel effort to get the detail using high freqrency bump map and low frequency displacement map. The fact that I was only using half the information kept me from getting acceptable renders.
What I have done is combine your effort with a split bump map materials. I basically took the original displacement map that pixolator provided, then gave it more contrast- about 20%, resaved the image as a +bump map png image file, then inverted this image to get the -bump map png.
I then created a +, and - bump material, using cinema’s filter shader the respective bump material’s bump channel (no other options were checked) (flip it vertically, then follow creed’s design for the +,- displacement materials.
I gave the displacemnt 6 inches, and ramped up the bump materials to about a strength of 175 postive for +bump and -175 for negative bump- put all 5 materials on the object, used default light - moved it a bit to light the object nicely and did a normal raytrace render no GI.
Continuumx
Interesting approach; should be possible to get a lot of detail and lower the required amount of subdivision in the Hypernurbs. My example uses a setting of 5; using the bump for high frequency detail should allow that to be lowered to 4, or even 3.
I’ll revisit my file and see what I can come up with.
Does the bump channel also work on the ‘black is low, grey is high, white is higher’ scale, or is it ‘grey is base, black is low, white is high’?
Might have to do some tests to check if we aren’t sure.
If it uses 50% grey as the base point, you wouldn’t need to make two bump maps, as you have to for the displacement.
Here’s a comparison of the same model with 4 subdivisions, using the displacement map as a bump map for higher frequency detail, to the model with 5 subdivisions and no bump map.
http://homepage.mac.com/creed_nmd/.Pictures/CGTALK/subdivisons_and_bump
Maya 6, mentalRay
Max 5 and VRAY
Rendered in Maya 5.01 with MR.
I only added a MRSubdivApproximation attribute to the polygonal mesh (Feature Displacement is OFF). The MRDisplaceApproximation
doesn’t seem to do anything at all (like somebody already mentioned).
Alpha Gain: 40.000
Alpha Offset: -20.000
Sampling Filter: Lanczos (instead of Gauss)
MRSubdivApproximation set to Parametric;
with NSubdivs set to 1, 3 and 5 (from left to right). Anything over 5 crashes Maya
(memory exception…even with 2GB RAM !?!)
Anybody know why?
cheers
stefan