[gtalarico]

I am in the early stages of a rendering a chair that has a perforated metal surface.

I have played with displacement before, and I have been able to achieve decent results...
http://i.imgur.com/ZFARE.png


As I was playing with the material, and switching back and forth between PreTess and On-the-Fly, I noticed that not only these options were affecting the quality and rendertime, they were also affecting the color of my metal...

Pretess: http://i.imgur.com/dR3Ra.jpg
On-The-Fly:http://i.imgur.com/JxZis.jpg

Is this by-design?
Has anyone seen this before?

Tks!

[JDHill]

Are those rendered in Maxwell Fire, or Maxwell Render? What is the difference in SL between the two? The pretess one is clearly much further along -- what does it look like with less render time? Is the metal using complex IOR? Just going out on a limb, but it doesn't seem unreasonable to say that more render time might bring the on-the-fly one in line with the pretess one. The apparent color difference is curious, but I can't rule out the possibility that you just have not rendered the on-the-fly image enough. Other than that, I would just check in Studio to make absolutely sure that the intended material is actually applied, and has been written correctly by the plugin.

[polynurb]

are you using any maps in ref0/90?

from working with displacement i know that textures can lead to color-shifts when tessellation/resolution is different.
especially when you have very steep edges like your grid has.

[zdeno]

or mayby it just reflect other parts of enviroment ? , one of them looks more hard-edge and another with more smooth surfaces

but if your studio is grey only then it is something very weird

[gtalarico]

Thanks guys.

I am letting some more tests render over night. I will post back the results and the scenes.

[gtalarico]

Here it is...

Same Material, one using PreTess, and one on the fly:
0.1cm:
http://i.imgur.com/5q2Lr.jpg
1cm:
http://i.imgur.com/P0J7v.jpg?1
5cm:
http://i.imgur.com/1sCut.jpg

There is a subtle difference, but I don't know it it is related to the precision of the displacement mesh.
The difference seems to increase with the displacement distance.
(PT seems warmer, on-the-fly cooler)

Here are the materials if you want to take a look at it:
https://www.dropbox.com/s/pzz3wbwv6p2v1ok/PerfMetal.zip
(PS: I am suck at making materials, so don't judge me on it...! : )

[JDHill]

I still think it comes down to the fact that on-the-fly needs more computation; the more extreme the displacement, the more computation required, and the more pronounced the effect would be. Here is an example using IBL and 0.1 cm displacement height, where the meshes have been repositioned to sit next to one another:



Is it readily apparent where the on-the-fly shape ends, and the pretesselated one begins? Increase the displacement height, though, and where you are paying a one-time up-front cost for the pretesselation (compute + memory), you will pay an ongoing cost for on-the-fly, which increases with displacement height.

Just a sidenote on your material: with pretesselated subdivision set to 32, as it was in the material, this scene consumed a maximum of around 8GB on my machine; the above was rendered using subdivision of 8, and might be able to get by with less, depending on the scenario.

[gtalarico]

Tks.

I guess I will have to agree with you...

I changed it to 32 when I was troubleshooting the material, just to see if it affected...


PS: how is that scene lit JD? It looks good for testing materials....
(Mine is just default sky dome, and it's not very good...

[JDHill]

That is just carb_04.hdr from the Maxwell/hdri/dosch directory.