Page 2 of 2 FirstFirst 12
Results 11 to 15 of 15

Thread: user_ibl_rect with parti_volume

  1. #11
    Join Date
    Nov 2011
    Location
    Paris
    Posts
    115

    Default

    Ah I didn"t see you post !

    Thank you for your reply !
    I'm using maya 2014 so I'm working with 3.11.1.9, I don't have access to the beta.

    So you mean that we shouldn't use user_ibl anymore ?
    I guess the problem is the same if using user_ibl_env.

    So what kind of light should we use to benefit from IS ? And how can we combine an ibl and a mapped rectangular light ?

    And about my issue with parti_volume, is there a solution to use it with a mapped arealight ?

  2. #12
    Join Date
    Aug 2010
    Location
    New York, NY
    Posts
    123

    Default

    Definitly get better results with 3.12, but not quite as clean as user_ibl_rect. Perhaps this can be improved? Also, Global LIS might scale better than user_ibl_rect, it might be usefult to test a scene with more complexity and high rendering quality.

  3. #13

    Default

    Hello MiniZun,

    in theory user_ibl_rect should be superseded by light importance sampling, which works on all lights except directional and user area lights.
    The user_ibl_env shader is not considered by the light IS mechanism, being a user area light.

    In general, the more lights you have in a scene, the better the light IS mechanism works (talking about mr 3.12 here). Light IS works best
    when it is used with dimensional lights (area lights), but can be used effectively also when you have a lot of non-area lights, like point
    and spots. It should not be used if you have just a few lights, it would not pay off.

    Regarding your original issue with parti_volume, I would need to take a better look at that.

    Best regards,

    Enzo Catalano

  4. #14
    Join Date
    Nov 2011
    Location
    Paris
    Posts
    115

    Default

    Hi I'm back !

    new test with the same setup but with the latest mental ray and the new official light, mila_light and object light

    area_light_texture_comparison.jpg

    The usr_ibl_rect is still a lot better, it's very fast (near instant) and noise free.
    The too other technique are similar in term of lighting and performance, but tue mila_light doesn't show the map correct
    The Light Importance Sampling doesn't seems to change anything, the change in render time is not relevant.

    I'm going to keep using usr_ibl_rect, there is such a big gap in quality and performance. We really need a strong working light solution in mental ray. the usr_ibl_rect is not supported and not even accessible by default.

  5. #15

    Default

    Hello MiniZun,

    the user_ibl_rect light shader has its own importance sampling mechanism, so there is pretty much no difference between
    using light importance sampling with it or not. Also, your test has only one light, and that is definitely not the best scenario
    for the light importance sampling mechanism to shine - quite the opposite, actually, you risk to have higher render times
    because of LIS overhead. In fact, the LIS mechanism has some heuristics built-in that decide when NOT to have importance
    sampling, typically when the point to be shaded is too far away from a light to use IS on it.

    The best scenario for LIS is with many lights - in those cases, you should see an acceleration when using it.

    Thanks for testing - best regards,

    Enzo Catalano

Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •