[Maximus3D]

Hi

I been searching, trying to find if anyone else posted anything about this and i can't find any other threads so here it comes.
I done a few testrenderings with alpha build v1.1.33.0001 and i get this nasty bug in the gradients in the sky, and i've seen it in other posted images too on the forum so it's just not me who gets this error.



See the clearly visible colorbanding here in this rendering that shows in the gradient in the sky. It's just not pretty and it ruins the whole image. I hope you guys can solve this with some dithering or something else.
Thanks in advance and keep up the good work

/ Max

[tom]

Which format are you using to save your output images?

[Maximus3D]

This is jpeg, but it's even visible even if i haven't saved my image. I see it appear in any format i render out to so i doubt it's due to compression.

/ Max

[tom]

I didn't want to mean its about compression but bitdepth maybe... Now I see it's not about it. You know, MXI is a high range format and it's being reduced to RGB gamut right after calculation for both displaying and saving the output. Thus the reduction is responsible for such artifacts. Since 24-bit is quite enough for many outputs, it's not even acceptable in such situations when you have many tones of similar colors as in your example. There are many ways of color reduction like "relative colorimetric", "absolute colorimetric", "perceptual", "saturation".... When reducing 32 or 64 bit per channel color down to 8 bit, M~R must be using one of them to prevent banding artifacts and maybe that's the way it's working already. But I'm not sure if it's perfectly possible to get rid of banding while reducing color depth this much. Even with dithering... Right after to be able to edit MXI we'll see what will happen there...

[tom]

I did some tests in photoshop and yeah dithering overcomes this very well as you say.
And I see your sky has no dithering, so this shows us MXI to 8-bit per channel RGB conversion doesn't do any trick about dithering.
This is not a bug because not announced as it works like that, instead this is a GREAT wish like texture filtering

[Maximus3D]

You're very much right about what you say Tom, i agree with you. However i have no bigger worries this problem will be resolved in future updates. It's just that until then a good idea would be to avoid any scenes containing large visible gradients in them due to this.

Btw Tom, thanks for your quick and interesting replies i appreciate that and it sure helps alot.

/ Max

[Tyrone Marshall]

I have been able to avoid this by using targa imag format for output. The banding is even more pronounced if you start changing the color curves.

[tom]

Maximus, is this problem still there? If yes, can you please send me a scene.

[Mihai]

We will have banding issues as long as Maxwells output is 8bit/channel, there's no way around that. But I really don't think it will remain at 8bit for long

[tom]

Mihai, you're right. I'm talking about the pronounced (non-dithered) banding as seen in Maximus' example.