-
Noticeable color banding-even in 32-bit images
I’m experiencing noticeable color banding on 16 and 32-bit image renders that aren’t evident in the Picture Viewer. The render looks fine when viewed in the Picture Viewer in C4D, but when opened in Photoshop or After Effects CS5, I’m seeing a lot of banding, even on the 32-bit image.
I generally work in 16-bit, rendering out 16-bit PNG sequences. The banding appears identical in 16-bit and 32-bit images in After Effects, regardless of whether I have the AE project color depth set to 16 or 32-bit. I generally work in the sRGB color space.
If I bring the 32-bit PSD back into the C4D Picture Viewer I don’t see the banding.
I took a screen grab of the C4D Picture Viewer and saved it as a TIFF. If I view that in the Mac Preview application I don’t see banding. If I bring that into AE or PS I see some banding, but not the many stages I see when I bring the rendered images into those Adobe packages.
I have a pretty good NEC LCD monitor that’s calibrated using NEC software and a Spyder 3 Elite Colorimeter. I don’t think calibration issues would result in a 32-bit image showing color bands.
I’ve posted a distilled version of the scene here:
https://www.marshall-arts.net/Support/
The scene includes a rounded dark grey CYC for the background to a product (which I’ve removed from the scene). The CYC is lit with various lights. The banding is visible in the falloff of the lights.
I see that in the update to 13.058 there’s a supposedly a fix of banding issues in the linear workflow. Is that relevant? I’m using a Linear workflow with all of the color management set to sRGB.
I’m using the Studio version of C4D, R13.061. Mac OS 10.6.8, Mac Pro, yadda yadda.
Thank you.
Shawn Marshall
Marshall Arts Motion Graphics -
5 Replies
-
I would guess that it’s got something to do with your AE/PS display settings.
-
I’m just guessing here but since 8 bit looks fine and 16/32 look good in the picture viewer I wonder if it’s because C4D is doing some fancy colour down-sampling to preserve the 8 bit look (since the display, Mac or PC, is going to be 8 bit/channel) and AE and PSD just assume you’ll put up with the visual banding when you are working above 8 bit… are there prefs in AE to address that I don’t see any in a fast search in PS CS5
-
Rest assured that your color values are all doing just fine. Photoshop and AE don’t dither color values before sending them to your display. Your display is incapable of showing all the color values in the gradient, so you see banding.
The only time Photoshop or After Effects will dither the values is when converting to a lower bit depth. If you convert your image to 8bpc by rendering to a QuickTime MOV through AE, for example, the image will be dithered and you shouldn’t see the same banding. The same is true for converting to 8bpc in Photoshop.
If knowing that your colors are okay isn’t enough then in 32bpc you can add a black solid, apply a small amount of noise (say 0.1%) to it and set it to Linear Dodge (Add). Your gradient should look nice (although you’ve now changed your rendered color values slightly).
An alternate method for checking your gradient would be to create a Levels adjustment layer and slide the gamma value around. What you should notice in 32bpc and 16bpc is that the color banding looks like it’s moving. This indicates that each band is in fact made up of several different color values (rather than just one, as displayed on your monitor). If you tried the same thing with an 8bpc image the boundaries of each band would not change, only the colors within the band.
Darby Edelen
-
Thanks for the replies. For now I ended up adding some noise in the 3D render which mitigated some of the banding. Now I’m noticing that adding RSMB to my 3D render adds banding. The gradients look pretty clean as-is in AE, but when I add RSMB banding appears.
Cheers.
Shawn Marshall
Marshall Arts Motion Graphics -
why worry about the banding at all? it’s not really there (it’s not there in the 8 bits images you sent), you are only seeing it since your display won’t do anything higher than 8 bit – your final output to 8 bit shouldn’t have it.
Reply to this Discussion! Login or Sign Up