aye, it looks more of a cutoff issue than anything else: it’s a very faint reflection…
As a counterproof, try and brighten them, and see if they do the same.
the reflection is not so faint.
and the banding is still visible with cut-off set to 0.
here´s a slightly bigger region of the image ( unfortunately i cannot the show the uncropped image), cut-off set to 0
each band is varying by 1 point of gray in the first image, and 2 points of gray in the second.
Toggling clamping does change anything?
Also, for brighten them up, i meant considerably, where the reflection would approach a visible colour of around 192 or thereabout.
It’s not like i have the answer, hey, i’m simply poking to try and understand better.
I ran over it with a colour picker and each band you’re getting only has a variation in colour of 002 between each shade - one is 60,60,60 the next is 62,62,62 next is 64,64,64 and so on so you haven’t really got a lot of space to work with between each - definitely use 16 bit images for starters but the only thing I can really suggest is to add some grain to break it up - in this case I think it’ll be the only way to disguise the banding. Otherwise as Lele mentioned you may have to higher your reflection level or use a custom falloff curve so that there’s more range in the reflection going from weak to strong.
I think you may be hitting a limitation of 8 bit images more than anything else.
I think you are definetly running into a limitation with 8 bit color. you have a real shallow range of greys over a broad area. if you sampled those colors and tried creating a ramp in photoshop the same size I bet it would band up too. Like everyone else said, you can try to increase the contrast of your reflections to introduce a broader range of grey into that surface.
thanks guys,
i hadn´t looked at it that way. of course there is not much room for smooth interpolation.
i came to a acceptable solution combining 16-bit and some noise.