Hey all,
I'm back to working on the installation and have noticed an irritating bug in pix_buffer_write (I think).
Once and a while (about 15 times out of 600) the buffer gets messed up when it is saved, the top 113 pixels of the image get copied over with what looks like a bitmask from another part of the image. Sometimes its totally black and opaque, other times you can see it resembles other parts of the image.
I can't see any pattern to it, and it seems to quite infequently (as long as your not writing 100s of images as I am)
I uploaded some of the messed up frames at:
www.ekran.org/ben/research/pix_buffer_bug.tgz
I would really like this fixed so that I can depend on the images getting written into buffers properly after 1000s of frames.
I wish I could see a pattern as to when this happens, it does always happen If I'm writing a lot of frames, but I can't force the creation of one of these messed up frames.
Thanks b..
What OS?
I use pix_buffer_write daily on the Mac with 600 or more frames and can't say i've seen that problem. All that object does is do a memcpy() into the buffer of the exact image dimensions. There should not be partial frames.
cgc
On 1/3/06, B. Bogart ben@ekran.org wrote:
Hey all,
I'm back to working on the installation and have noticed an irritating bug in pix_buffer_write (I think).
Once and a while (about 15 times out of 600) the buffer gets messed up when it is saved, the top 113 pixels of the image get copied over with what looks like a bitmask from another part of the image. Sometimes its totally black and opaque, other times you can see it resembles other parts of the image.
I can't see any pattern to it, and it seems to quite infequently (as long as your not writing 100s of images as I am)
I uploaded some of the messed up frames at:
www.ekran.org/ben/research/pix_buffer_bug.tgz
I would really like this fixed so that I can depend on the images getting written into buffers properly after 1000s of frames.
I wish I could see a pattern as to when this happens, it does always happen If I'm writing a lot of frames, but I can't force the creation of one of these messed up frames.
Thanks b..
GEM-dev mailing list GEM-dev@iem.at http://lists.puredata.info/listinfo/gem-dev
Hu Chris,
Ah this is on linux, not OSX.
(installation = linux) (performance = osx)
my cpu usage is low and each frame captured is done by an instance of the same abstraction, I have no idea why some would mess up and not others.
Any insight Johannes?
..b.
chris clepper wrote:
What OS?
I use pix_buffer_write daily on the Mac with 600 or more frames and can't say i've seen that problem. All that object does is do a memcpy() into the buffer of the exact image dimensions. There should not be partial frames.
cgc
On 1/3/06, B. Bogart ben@ekran.org wrote:
Hey all,
I'm back to working on the installation and have noticed an irritating bug in pix_buffer_write (I think).
Once and a while (about 15 times out of 600) the buffer gets messed up when it is saved, the top 113 pixels of the image get copied over with what looks like a bitmask from another part of the image. Sometimes its totally black and opaque, other times you can see it resembles other parts of the image.
I can't see any pattern to it, and it seems to quite infequently (as long as your not writing 100s of images as I am)
I uploaded some of the messed up frames at:
www.ekran.org/ben/research/pix_buffer_bug.tgz
I would really like this fixed so that I can depend on the images getting written into buffers properly after 1000s of frames.
I wish I could see a pattern as to when this happens, it does always happen If I'm writing a lot of frames, but I can't force the creation of one of these messed up frames.
Thanks b..
GEM-dev mailing list GEM-dev@iem.at http://lists.puredata.info/listinfo/gem-dev