sebkaine Posted July 26, 2016 Share Posted July 26, 2016 (edited) Hi guys , I have assemble a very basic scene with 1 sphere lighted by a sphere light. I can't remove my moire pattern problem ( see attach image ) i have try to - disable all shadows - augment light quality - play with output format - play with normals - play with subdivision mode - play with the shader - play with sample quality Nothing work, i still get an ugly step transition in light gradient. Any hint on what i am doing wrong would be great ! I just want a very smooth and elegant gradient in the light attenuation. Thanks for your time ! Cheers E ps : The attach image has been modify in PS to highlight the problem. light_moire_pattern.hiplc Edited July 26, 2016 by sebkaine Quote Link to comment Share on other sites More sharing options...
lukeiamyourfather Posted July 26, 2016 Share Posted July 26, 2016 If you use micropolygon rendering the banding goes away. It's like a floating point limitation or bug. Quote Link to comment Share on other sites More sharing options...
Atom Posted July 26, 2016 Share Posted July 26, 2016 (edited) Try increasing the Diffuse Limit on the Mantra node, you have it set to zero. Edited July 26, 2016 by Atom Quote Link to comment Share on other sites More sharing options...
lukeiamyourfather Posted July 26, 2016 Share Posted July 26, 2016 37 minutes ago, Atom said: Try increasing the Diffuse Limit on the Mantra node, you have it set to zero. The direct diffuse should be free of banding in this case. If you look at indirect diffuse and direct diffuse separately the direct diffuse still has banding. It's covering up the issue but doesn't solve the root problem. Quote Link to comment Share on other sites More sharing options...
sebkaine Posted July 26, 2016 Author Share Posted July 26, 2016 Thanks for your help guys ! @Atom > indirect diffuse add a bounce(s) of GI but i just want a basic lighting no GI. @Luke > that's interesting that micropoly doesn't have the issue. i share your diagnostic it looks like a bug in floating point interpretation but i would like to be sure that i haven't omit anything stupid before submitting the bug. Quote Link to comment Share on other sites More sharing options...
sebkaine Posted July 26, 2016 Author Share Posted July 26, 2016 (edited) @Luke > i double check and i still have moire pattern in micropoly, the problem is also there when i use a point light / tube light / grid light in any case i can't suceed to get a smooth gradiant in light attenuation. i'm on Houdini 15.5 and i haven't notice this problem before ? at first i thought of a geometry problem either - bad normal - bad smoothing - bad topology but i replace my sphere with a perfect nurbs sphere and get the exact same result so i would exclude any geo related problem. i have also try dany PBR shader to check if it was an issue cause by the new shader , but i get the exact same issue. i have also try all type light with the same issue switching to 2.2 / 8 bit reduce the problem but don't make it disappear. i will try to compare with octane to see what i get ... as a side note openGL playblast give me more accurate result and a nice transition in the gradient. Edited July 26, 2016 by sebkaine Quote Link to comment Share on other sites More sharing options...
Atom Posted July 26, 2016 Share Posted July 26, 2016 If you enable the Edge Falloff for the light it does smooth out a bit. Quote Link to comment Share on other sites More sharing options...
Guest tar Posted July 26, 2016 Share Posted July 26, 2016 (edited) Isn't it just banding of the display? Saving it out to Nuke where the viewer can be dithered, shows no banding. Edit: added a gif that shows how noise/grain reduces banding, even on terribly banded .gifs (sorry for the 4k monitor sized gif!) Edited July 26, 2016 by tar Quote Link to comment Share on other sites More sharing options...
lukeiamyourfather Posted July 26, 2016 Share Posted July 26, 2016 24 minutes ago, marty said: Isn't it just banding of the display? Saving it out to Nuke where the viewer can be dithered, shows no banding. I'm using a 10-bit per channel display. If it were the display the banding wouldn't be discernible by the human eye. The banding here is pretty bad and clearly visible. My guess is still a bug. Please send the scene file to SESI support. Quote Link to comment Share on other sites More sharing options...
sebkaine Posted July 27, 2016 Author Share Posted July 27, 2016 Thanks again for your help guys ! @atom > thx it does make things smoother ! @marty > the pb is that i am watching this in a gear VR that's why i need very smooth gradient and no noise, cause small noise look like a snow storm in the gear @luke > i will do that today Quote Link to comment Share on other sites More sharing options...
sebkaine Posted July 27, 2016 Author Share Posted July 27, 2016 bug submitted i'll keep you inform when i have an answer. Quote Link to comment Share on other sites More sharing options...
Guest tar Posted July 28, 2016 Share Posted July 28, 2016 (edited) On 7/27/2016 at 10:41 AM, lukeiamyourfather said: I'm using a 10-bit per channel display. If it were the display the banding wouldn't be discernible by the human eye. The banding here is pretty bad and clearly visible. My guess is still a bug. Please send the scene file to SESI support. Interesting that 10bit is showing banding. The attached shows the same exr in Nuke with dithering on/off on a 32-Bit Color (ARGB8888) OsX display. edit: click on the pic to display properly - the preview has destroyed the subtleness Edited July 28, 2016 by tar Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.