Page 2 of 2

Re: reflective surface causing specs

Posted: Tue Aug 11, 2015 6:25 pm
by larsmidnatt
hard to say without pictures, that's for sure.

Re: reflective surface causing specs

Posted: Fri Aug 14, 2015 7:33 am
by Spectralis
I've had a similar problem recently with one of the animations I was rendering. The scene used two emissive planes with IES spotlights. I used RedSpec shaders on a G2M figure I was getting hot pixels in certain areas of the skin at certain points in the animation as the figure passed near the plane. The only way to cure this without setting hot pixels to zero was to double the resolution and move the emissive plane further away from the figure. This wasn't ideal because it changed the lighting of the scene. I tried reducing the emissive power but the lighting looked worse than when I moved the plane further away. It seems that the only way to get rid of this problem without using hot pixel removal is to render for a longer time at a higher resolution. This isn't ideal for animation. Strangely, exactly the same scene with the same settings but using a G2M figure with a generic skin didn't have this problem. I'd love to know what causes it and how to avoid it.

Re: reflective surface causing specs

Posted: Fri Aug 14, 2015 7:57 am
by sikotik13
At a guess, since I am not intimately familiar with RedSpecs shaders, I would guess specularity or gloss. The reflectivity tends to interact in all kinds of crazy ways with light, especially when also mixed with bump/normal/displacement. It is sort of significant, in that skin does possess said reflective properties, but that is most likely the difference between a more accurate shader and a generic one as far as the light interaction is concerned. Realism generally dictates a need for more computation to be correct.

Re: reflective surface causing specs

Posted: Fri Aug 14, 2015 8:20 am
by Spectralis
bepeg4d wrote:Hi,
it would be nice to see an image for better evaluate, but, in general, firstly check your bump and normal values, if they are too high, it's easy to get fireflies
Are you using IES files for the emitters?
Anyway, the faster way for killing all the fireflies is to render at double size with a quarter of sampling and then downscale the final render to the original resolution. In this way all the 1 bright pixels should be removed avoiding to use the Hot Pixel Removal that kills all the details ;)
ciao beppe
This solution has also been suggested for IRAY in 3ds Max and DAZ. Rendering at double size/half resolution and then downscaling 50% is quicker and produces better looking images apparently. Probably not practical for animation - not sure how long it would take Photoshop to batch convert 2200 frames.

http://buerobewegt.com/quicktip-renderi ... r-in-iray/

Re: reflective surface causing specs

Posted: Fri Aug 14, 2015 8:37 am
by Spectralis
sikotik13 wrote:At a guess, since I am not intimately familiar with RedSpecs shaders, I would guess specularity or gloss. The reflectivity tends to interact in all kinds of crazy ways with light, especially when also mixed with bump/normal/displacement. It is sort of significant, in that skin does possess said reflective properties, but that is most likely the difference between a more accurate shader and a generic one as far as the light interaction is concerned. Realism generally dictates a need for more computation to be correct.
Thanks! I'll try experimenting with the specularity and gloss settings to see if this reduces the problem. The RS shaders look pretty good though so don't want to mess them up.

Re: reflective surface causing specs

Posted: Fri Aug 14, 2015 2:35 pm
by larsmidnatt
Spectralis wrote: Probably not practical for animation - not sure how long it would take Photoshop to batch convert 2200 frames.

http://buerobewegt.com/quicktip-renderi ... r-in-iray/
Thanks for that link. Interesting results. I agree for a one hour animation that would be over 85,000 frames or so assuming 24fps.

Re: reflective surface causing specs

Posted: Fri Aug 14, 2015 5:45 pm
by Spectralis
larsmidnatt wrote:
Spectralis wrote: Probably not practical for animation - not sure how long it would take Photoshop to batch convert 2200 frames.

http://buerobewegt.com/quicktip-renderi ... r-in-iray/
Thanks for that link. Interesting results. I agree for a one hour animation that would be over 85,000 frames or so assuming 24fps.
Once I've finished rendering a project I'm currently working on I'll try resizing 2200 files using batch convert in Photoshop to see how long it takes. If rendering plus downscaling is quicker then I might use that process in the future.

Re: reflective surface causing specs

Posted: Sun Aug 16, 2015 8:38 am
by bepeg4d
Hi,
you can batch downsize all the images with PS or GIMP or you can avoid this and import all the images as sequence in QT Pro or any other video app and export it at 50%.
It's extremely more faster and you have the same result ;)
ciao beppe

Re: reflective surface causing specs

Posted: Sun Aug 16, 2015 8:08 pm
by larsmidnatt
bepeg4d wrote:Hi,
you can batch downsize all the images with PS or GIMP or you can avoid this and import all the images as sequence in QT Pro or any other video app and export it at 50%.
It's extremely more faster and you have the same result ;)
ciao beppe
that makes sense actually.

thanks