PTB Update changes effect of Screen('BlendFunction'... on movie textures?

1
I just updated to PTB 3.0.12 from 3.0.11 today, and it seems to have caused a strange side effect in my experimental script, whereby movies now appear at ~50% contrast.

Previously I set Screen('BlendFunction', GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA) at the start of my multi-experiment function and alpha blending worked as desired for all experiment types (drawing images with Gaussian masks, dots, lines, movies, etc). Now it seems that I have to update alpha blending when I switch to the movie experiment using Screen('BlendFunction', GL_DST_ALPHA, GL_ONE_MINUS_DST_ALPHA) in order for movies to appear correctly (i.e. full contrast).

Perhaps this is the correct way of doing things and I was somehow getting away with my error up until now, but can anybody explain to me why this solution works? I'm using Screen('OpenMovie') with the 'pixelFormat' input set to 4 for RGBA8 (default) but I do have a mixture of .avi, .mp4, and .wmv video files. It's not clear to me why the alpha channel of these movies would be anything less than 1 though... (I'm running the latest GStreamer on Matlab 2013a, Windows 7 with an NVidia GTX 650 Ti).

Thanks!

Aidan Murphy, PhD
Section on Cognitive Neurophysiology and Imaging,
Laboratory of Neuropsychology, NIMH
2
Not here, as tested with v3.0.12 on Linux with Octave and Windows-7/NVidia GTX-970 with Octave and Matlab R2015b (running -nojvm) and latest GStreamer 1.6.0 for Windows. Adding blendfunction calls to PlayMoviesDemo.m and selecting pixelformat 4 does not have any effect on contrast for any of our demo movies.

So i'm not sure what's happening on your side? I guess you'd need to provide a minimal script that demonstrates the problem, but it sounds like this could be some other bug in your script, or a NVidia graphics driver bug?

-mario


XX---In PSYCHTOOLBOX@yahoogroups.com, <murphyap1@...> wrote :

I just updated to PTB 3.0.12 from 3.0.11 today, and it seems to have caused a strange side effect in my experimental script, whereby movies now appear at ~50% contrast.

Previously I set Screen('BlendFunction', GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA) at the start of my multi-experiment function and alpha blending worked as desired for all experiment types (drawing images with Gaussian masks, dots, lines, movies, etc). Now it seems that I have to update alpha blending when I switch to the movie experiment using Screen('BlendFunction', GL_DST_ALPHA, GL_ONE_MINUS_DST_ALPHA) in order for movies to appear correctly (i.e. full contrast).

Perhaps this is the correct way of doing things and I was somehow getting away with my error up until now, but can anybody explain to me why this solution works? I'm using Screen('OpenMovie') with the 'pixelFormat' input set to 4 for RGBA8 (default) but I do have a mixture of .avi, .mp4, and .wmv video files. It's not clear to me why the alpha channel of these movies would be anything less than 1 though... (I'm running the latest GStreamer on Matlab 2013a, Windows 7 with an NVidia GTX 650 Ti).

Thanks!

Aidan Murphy, PhD
Section on Cognitive Neurophysiology and Imaging,
Laboratory of Neuropsychology, NIMH