Color Shift "Flashes" When Renering

bhurst wrote on 6/26/2012, 10:22 AM

I have been having a problem I call "color shift flashes" when rendering in VP11. I first noticed it when rendering content that contained hi-res photos, but now I see the problem with various source content and various output formats/codecs.

Then, I decided to try rendering the same content in VP9, and the problem is gone. Anyone aware of any existing problems of this nature in VP11?

Here are two sample rendered files ... one from VP9, one from VP11 ... using the same source file. The bad spots in the VP11 file are frames 2 - 3, frames, 76 - 77, frames 101 -102, etc.

VP11 (with problem)
http://www.mediafire.com/?tvw939262cgg69l

VP9 (no problem)
http://www.mediafire.com/?u6i7dz31f7ztfv7

Any suggestions would be appreciated.

Thanks!!

Comments

paul_w wrote on 6/26/2012, 11:05 AM

Looking at the problem v11 footage, there is a major swing from the red and blue hues to magenta. As viewed on the Vectorscope. Lasts for between 1 and 2 frames.
Worst point i found was at frame 00:00:03;11. Its almost completely magenta with tiny amounts of green left in.

Have you tried disabling GPU in v11 either during the render (i mean in the render options, use CPU only), or try switching it off in Vegas prefs.

Edit: the luminance level seems fine throughout, its only the chroma that gets effected.

Paul.

bhurst wrote on 6/26/2012, 11:59 AM

I had selected "Use CPU Only" during rendering and the problem persisted.

I just tried disabling GPU acceleration in prefs, and that seems to have fixed it.

Now, I wonder if I have a "faulty" GPU or a simple "incompatibility." According to the VP11 GPU Acceleration page, my card/driver is compatible, but obviously it's not working.

Thanks!

paul_w wrote on 6/26/2012, 12:07 PM

ok good.

Some thoughts:
make sure your GPU card has the latest drivers installed.
check the RAM preview setting, not set to zero (100 - 2000 is ok)

If thats all good, then start a ticket with SCS. Could well be a bug.

good luck
Paul.

Steve Mann wrote on 6/26/2012, 9:00 PM

Here's a side-by-side comparison. It happens once near the beginning and once near the end. I didn't see any others.

Definitely one for SCS.

bhurst wrote on 6/27/2012, 7:21 AM

Thanks! I will reference this thread for SCS.

bhurst wrote on 6/27/2012, 7:48 AM

I submitted the issue. I will post results, if any, here.