Color Banding of XFrog Model

Started by choronr, September 29, 2011, 01:18:41 pm

Previous topic - Next topic

choronr

I have used this XFrog model many times before but never have run into this problem of color banding. I tried several different setting changes in atmosphere, lighting and in the render settings and still get the banding issue. Would appreciate your thoughts on this.

Bob

Kadri


Bob are you sure it is from the model?.
It looks like a bucket render discrepancy too.
Did you tried a different POV to be sure?

choronr

Quote from: Kadri on September 29, 2011, 07:57:57 pm

Bob are you sure it is from the model?.
It looks like a bucket render discrepancy too.
Did you tried a different POV to be sure?


Thanks Kadri, I had moved the model farther back at bit and got the same results. Then, I chose ambient occlusion instead of global illumination it the atmosphere and it seems the banding went away. Not sure I know what the differences are when you use AO vs GI - but, after a quick render of the scene, the image is noticeably lighter.

Not sure I know what you mean about a 'bucket discrepancy' - i.e., what is default; and, what should it be?

Kadri


'bucket discrepancy'
Sometimes there are render errors in TG2.
Some parts are rendered different.
They look like different rendered squares in the images.

I am curios too what is happening with the object Bob.

choronr

Quote from: Kadri on September 29, 2011, 08:47:53 pm

'bucket discrepancy'
Sometimes there are render errors in TG2.
Some parts are rendered different.
They look like different rendered squares in the images.

I am curios too what is happening with the object Bob.

The current bucket reading is 256/256. I had no render errors. All other elements of the scene looked very good - except the banding within the cactus.

I've made some changes to the gamma correction that should darken the scene to where it was before I changed to AO - then, I'll do a final render (about 13 hours).

dandelO

Are your cloud layers set to acceleration=none, Bob? I suspect that might be the problem if they aren't.

choronr

Quote from: dandelO on September 30, 2011, 10:40:06 am
Are your cloud layers set to acceleration=none, Bob? I suspect that might be the problem if they aren't.

Hi Martin,

All cloud layers are set at 'acceleration/none'. I'm doing two full renders; one with AO and one with GI to show the comparison. The one with GI will produce the banding in the cactus. Should be done by late today.

choronr

Ordinarily, I'd place these images in the image sharing section; but, will present them here to illustrate the problem.  Here are two renders; the first using 'ambient occlusion'. The second using 'global illumination'.
The AO version seems to eliminate the banding issue while the GI version displays the banding issue at the lower left. The overall best image quality still lies with the GI version  except for the banding.

Some borrowed elements are:

•   FrankB's 'Simple Camoflage Surface' added to my surfaces.
•   dandelO's  'PGS Evolution' grass surface on distant mountains.
•   dandelO's 'Altocumulus' clouds.
•   Walli's 'Dry Shrubs' and XFrog cactii.

Since I cannot find the reason for the banding  issue, I'll use the 'sweep it under the rug' fix by just placing a single model of the gray/green bush directly in front of the cactus to hide the banding. Additionally, I'll tone down the color of the small straw colored bushes.

Here is the AO version.

choronr

And, here is the GI version.

Oshyan

I still see the issue a bit in the AO version. These are classic examples of small differences between lighting in different render tiles. Changing the size (smaller) *may* help, but maybe not. Instead you could/should try increasing GI Sample Quality and/or Relative Detail. Try values of 2 or 3 for Rel. Det. and 8 for Sample Quality.

- Oshyan

choronr

Quote from: Oshyan on October 12, 2011, 07:34:16 pm
I still see the issue a bit in the AO version. These are classic examples of small differences between lighting in different render tiles. Changing the size (smaller) *may* help, but maybe not. Instead you could/should try increasing GI Sample Quality and/or Relative Detail. Try values of 2 or 3 for Rel. Det. and 8 for Sample Quality.

- Oshyan

Thank you Oshyan for these suggestions. This was the first time I ran into this issue. I've gone onto another project now; but, have noted your settings in the event I run into it again ...appreciate it.