Finn-Arne Johansen wrote:
I've finally found time to test this, and found that the bug with v4l-devices is still there. I created a small project, with 2 still-pictures only. It's meant to be the excact same picture, One is taken with xawtv (.jpg) and one is taken from stopmotion (.png). As you can see, the one from stopmotion is nearly useless, while the one taken from xawtv is almost clear. The .sto file was originally named "test_2.sto", but I renamed it as I placed it availible for download at developer. Please take a look athttp://d.skolelinux.no/~finnarne/v4l_bug_stopmotion_0.1.1.sto
Thanks for the testing.We are currently moving away from using gstreamer for capturing images and importing video as this caused us no end of problems.
We are moving to a design where we will use external programs such as mplayer, vgrabbj, etc to grab images, and then import the images realtime, therby creating a video.
We feel that this way of doing it is much better as it frees us from gstreamer and all the bug associated with it, and allows people to specify any program for capturing frames from the input device.
This means we will be able support all devices, as long as you can find any program that supports it, and if you have a command line option to grab images from xawtv you can even use this as an direct input source.
Most of this will be in the next release, and I pretty sure that it will completely eliminate this bug =)
-- Fredrik Berg Kjølstad