Home Technical Talk

Mental Ray Problem (Smudged Surface?)

Big Rocket
polycounter lvl 18
Offline / Send Message
Big Rocket polycounter lvl 18
Hello everyone,

I ran into a problem with the Mental Ray renderer in 3ds Max 2009. Basically, I followed 3ds Max's official tutorial for rendering (the one you get in the help files), and instead of the military compound the tutorial used, I rendered my own model. As far as I can tell, I followed every applicable step in the tutorial, and I ended up with the strange result you see below. The last of the 3 renders look smudged across the surface, as if someone with dirty fingers touched it and left fingerprints all over it.

mentalrayproblempe3.th.jpg

By way of background information: The model started out as a low-poly model consisting entirely of quads, then TurboSmoothed with 5 iterations for rendering. In the Mental Ray render, I used different antialiasing levels, but they all had the same problem (the screenshot was done with the highest antialiasing level). The problem manifested itself in Mental Ray only; the viewport real-time renderer and the Default Scanline Renderer did not have a problem at all.

1. Has anyone else seen this smudging problem before? Am I missing something obvious?
2. Is there a way to tell Mental Ray to disable shadows for certain objects (like the whiskers?)

Thanks.

- BR

Replies

  • IxenonI
    Offline / Send Message
    IxenonI interpolator
    Looks like a sampling problem with the shadows. Have you tried upping the sample size for the lights that cast shadows, or if you use shadowmaps, increase the filter size and up the resolution?
  • Big Rocket
    Offline / Send Message
    Big Rocket polycounter lvl 18
    Thanks, IxenonI.

    It turns out my model's scale was way too small. When I did the head, I started out with a polygon cube whose length = width = height = 2 units (the default 3ds Max units). The "smudge" problem appeared because I was doing the equivalent of zooming in really close. After I scaled up the model by a factor of 100X, the problem disappeared.

    By the way, once I fixed that initial problem, there was some fuzziness in the shadows, and your suggestion with the sample size took care of that. You preempted my question, which was nice. Thank you.

    For my information, are there any general guidelines as to how big a model should be, in terms of the bounding box's length, width and height?

    Thanks.

    - BR.
Sign In or Register to comment.