Rendering Quality Issue, (Striations in gradients) Options

[Sean delgatto]

Have been having this issue for a while now and not sure how to get rid of the problem. Basically you see striations in every gradient, particularly bad in areas of extreme contrast. Not sure if its a rendering option, a card issue, or what. No matter what card, amount of memory, or processor I have used over the past, and they are varied, I get the same results no matter what.
Info on PC is an Intel Duel core 2.6 G, 2 gigs of ram, 1.9 T hard drive space. Also have tried a variety of passes and between 2 passes and a 100 absolutely no difference.

Again if anyone can help me I really would appreciate it.

Also side note, when doing alpha layer rendering, I noticed that if there was a light source on an object similar in ways to the one in the attached photo, that it cute out an outline around the object rather than also include the light source and glow. Thats problem number 2. Again if anyone knows the answers to this I would be most grateful.

Thanks,

Sean

Attached thumbnail(s)

 

[robcat2075]

Some lens flare settings will create actual visible circles or bands because real lens do that sort of thing as artifacts of the lenses.

That banding looks quite slight, it may just be the limitation of 8 bit color. OpenEXR format has more precision but it can't change the fact that your display is almost certainly 8 bits per color also.

I have an LCD display and a CRT and i find the LCD is much worse when it comes to banding. This is true with your sample image when i compare it on both.




I think most LCDs are doing tricks to approximate 24bit color but dont' really do it like CRTs can.



Lens flare and other post effects aren't included in alpha buffers. If you absolutely have to composit a flare from A:M into something else, do a pass with everything else black and use a compositing mode to add it to your final image in whatever app you are using to composite with.

[NancyGormezano]

I have an LCD display and a CRT and i find the LCD is much worse when it comes to banding. This is true with your sample image when i compare it on both.

Aha! So that's why I could barely see the problem (uhh...couldn't see it at all). I use a CRT.

[Sean delgatto]

I have an LCD display and a CRT and i find the LCD is much worse when it comes to banding. This is true with your sample image when i compare it on both.

Aha! So that's why I could barely see the problem (uhh...couldn't see it at all). I use a CRT.

Definitely more noticible for sure but even if its a regular gradient I still see it and when What ever it is that I am doing is converted to DVD then plaid on a Tv whether it be CRT or LCD I still see it. I'll see what I can find out about OpenEXR

[robcat2075]

Definitely more noticible for sure but even if its a regular gradient I still see it and when What ever it is that I am doing is converted to DVD then plaid on a Tv whether it be CRT or LCD I still see it. I'll see what I can find out about OpenEXR

There's only 256 shades of gray in 24 bit color. Anything in between 134,134,134 and 135,135,135 for example, has to be simulated with something like 135,134,134 which will not be pure gray but red tinted. That's the first problem. In a long stretched out gradient like you have it may be visible.

Then when you compress to DVD or any digital TV format precision gets even worse. The DVD codec will compress areas of similar color to the same color making banding more obvious.

I see professional DVDs with obvious banding and some where it's not apparent. It may be that CG images are worse because they don't have film grain noise to break things up.

[bobbesch]

This is a common problem in photography also. Adding a small amount of noise usually helps.

[jimd]

i have run across the same thing
you'll have to mess with light setting when dealing with volumetric fx
also google, banding light in cg
its a common problem with no one fix
j