Page 1 of 2 12 LastLast
Results 1 to 15 of 18

Thread: Flickering in animation

  1. #1

    Flickering in animation

    I have an annoying flickering going on in an animation. Any individual frame renders out just fine, but during animation there are very noticeable color shifts from 1 frame to the next, creating the flickering effect.

    Here's a short clip of the animation
    https://vimeo.com/387481883

    The color shift is most noticeable in the 2x4 wedges on the left.

    I'm still getting used to the 2019 setup & render process. Is this 'noise' from too few samples somewhere? Or a different issue?
    My current render settings...
    Click image for larger version. 

Name:	Render Properties.jpg 
Views:	79 
Size:	276.3 KB 
ID:	146736Click image for larger version. 

Name:	Render Properties 2.jpg 
Views:	81 
Size:	320.1 KB 
ID:	146737

    What am I missing?

  2. #2
    Registered User squarewulf's Avatar
    Join Date
    Aug 2014
    Location
    South Bend, Indiana
    Posts
    437
    Looks to me like it could be overlapping geometry. In modeler try merging points (m) and then unifying polygons (shift+m) and see if it does anything.

    edit: may want to save out incrementally, hard to come back from merging points you dont want merged.

  3. #3
    For starters flickering can be an issue with Interpolated GI. For something like you are doing here I'm not so sure GI is even needed, but if it is you might want to try an environment light. Also make sure all your settings are divisible equally. Having reflection and refraction samples set at 5 and then having adaptive sampling set at 6 is generally not a good thing. Also there is hardly ever a need for Diffuse Bounces to be set higher than 2. If you are insistent on using GI try Brute Force only with a Rays setting of 32, when using Interpolated, Brute Force Rays don't compute - completely meaningless - and jack up your Interpolated settings to say 256 primary and 128 secondary to start and go from there. But again, see if you can use the environment light or well placed area lights. Good luck.
    Tim Parsons

  4. #4
    Electron wrangler jwiede's Avatar
    Join Date
    Aug 2007
    Location
    San Jose, CA
    Posts
    7,205
    Quote Originally Posted by Tim Parsons View Post
    Also make sure all your settings are divisible equally.
    Please explain why that's beneficial?

    Having reflection and refraction samples set at 5 and then having adaptive sampling set at 6 is generally not a good thing.
    Why? What negative outcome does it cause?

    I'm not trying to put you on the spot, I just haven't run into the "sample divisibility" issue you're describing, and would like to understand it better.
    Last edited by jwiede; 01-27-2020 at 01:39 PM.
    John W.
    LW2015.3UB/2019.1.5 on MacPro(12C/24T/10.13.6),64GB RAM, NV 980ti

  5. #5

    Having reflection and refraction samples set at 5 and then having adaptive sampling set at 6 is generally not a good thing.
    qurious to know this too.

    as for flicker, is it interpolated Gi ?
    LW vidz   DPont donate   LW Facebook   IKBooster   My vidz

  6. #6
    I thought this was pretty common knowledge in the realm of rendering in computer graphics - keep all rendering inputs divisible by two. Same goes for images. I'm no math wiz so I can't speak to its validity, but I've been doing this over twenty years and that's what I have always heard. Seems to work for me. If I come across a reliable source for this information I'll put up a link.
    Tim Parsons

  7. #7
    Curmudgeon in Training Ma3rk's Avatar
    Join Date
    Mar 2003
    Location
    Near Beaverton, OR
    Posts
    2,034
    FWIW, it looked pretty clean w/o any flicker on my HP laptop.
    Earth can't be flat otherwise cats would have pushed everything off the edge!

    The Big Apexx System:
    AMD RYZEN THREADRIPPER 2990WX Processor (3.00GHz)
    64GB
    2x NVIDIA Quadro RTX 4000 8GB 32 Core

  8. #8
    If you want to save a bit on render time then you likely only need 1 diffuse bounce or maybe 2 at the most. Anything after 2 you likely won't see any visible difference. Your flickering problem is most likely being caused by overlapping geometry or 2 polygon faces occupying the exact same plane as Squarewulf has already pointed out.
    Threadripper 2990WX, X399 MSI MEG Creation, 64GB 2400Mhz RAM, GTX 1070 Ti 8GB

    https://www.dynamicrenderings.com/

  9. #9
    Quote Originally Posted by Tim Parsons View Post
    I thought this was pretty common knowledge in the realm of rendering in computer graphics - keep all rendering inputs divisible by two. Same goes for images. I'm no math wiz so I can't speak to its validity, but I've been doing this over twenty years and that's what I have always heard. Seems to work for me. If I come across a reliable source for this information I'll put up a link.
    I generally do that 8, 16, 32, 64, 128, 256 thing as well. I think I picked it up years ago and it's become a habit for me more than anything.
    Threadripper 2990WX, X399 MSI MEG Creation, 64GB 2400Mhz RAM, GTX 1070 Ti 8GB

    https://www.dynamicrenderings.com/

  10. #10
    Electron wrangler jwiede's Avatar
    Join Date
    Aug 2007
    Location
    San Jose, CA
    Posts
    7,205
    Quote Originally Posted by Nicolas Jordan View Post
    I generally do that 8, 16, 32, 64, 128, 256 thing as well. I think I picked it up years ago and it's become a habit for me more than anything.
    IIRC, these sample field values are all multipliers internally, though, against a base sample value (some low even number), so whether the multipliers themselves are divisible by two (or powers of two) shouldn't make a difference because two's already a factor of the base value. Further, he specifically mentioned the fields should all share a common divisor, and in context that made it even less clear what the actual issue was that was being solved/prevented.

    I asked for an explanation to try and pin down the actual issue behind these "rules".
    Last edited by jwiede; 01-28-2020 at 12:36 AM.
    John W.
    LW2015.3UB/2019.1.5 on MacPro(12C/24T/10.13.6),64GB RAM, NV 980ti

  11. #11
    Just a quick update.
    1st I checked geometry, merged points, and checked vertices, etc... and that isn't my issue in this case.
    2nd I adjusted sample values to be equally divisible (although, like others above, I'm curious as to why this is important) by setting Camera AA at 6, reflection/refraction at 3/3, light samples at 36. FWIW I also reduced diffuse bounces to 2. But none of that fixed the flicker.

    Turning GI off completely did take care of the flickering, but resulted in several new issues since I set up the scene with GI in mind (it's critical to parts later in the scene). So GI is the culprit but I still don't know exactly how/why.

    Going the way of Brute Force GI (32 rays) seemed to help, but resulted in unacceptable render times (about 4x longer)

    Back with interpolated GI, I used the suggested 256 primary and 128 secondary rays, but this still didn't fix the issue.

    Amazingly, what ended up working the best so far was adjusting these interpolation settings: min pixel spacing to 1, max to 100, and interpolation softness up to 80%
    That has resulted in this render (quick clip).
    https://vimeo.com/387695231

    If you could watch it in full HD, you would still notice some flickering but it's drastically reduced.
    What I've gotten to is probably acceptable for my company's purposes, but I'd love more insight on what exactly is going on and why those particular settings worked (more or less). And obviously if there is a more "correct" solution, I would love to hear it; and any other suggestions that might work better.

    Thanks for the help and suggestions so far.

  12. #12
    Medical Animator mummyman's Avatar
    Join Date
    Aug 2009
    Location
    Connecticut
    Posts
    1,078
    Did you try using an Environment light and no GI? Shouldn't flicker... and could give similar results. I keep hearing that... and watching some tutorials and it seems to workout pretty well.

  13. #13

  14. #14

    try this >

    LW vidz   DPont donate   LW Facebook   IKBooster   My vidz

  15. #15
    Registered User
    Join Date
    Jan 2004
    Location
    Stavanger, Norway
    Posts
    181
    Simple solution: Do not use interpolated radiosity

    I never do. For it to look good you have to tweak it so that the time gain isn't that significant anyway, so I find it simpler to just leave it alone.
    - Ignorance is bliss...

Page 1 of 2 12 LastLast

Bookmarks

Posting Permissions

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