GL860: more devices, colour images
Lorin Olivier also has a GL860 with a different USB ID (05e3:f191) in an Asus F5RL laptop. He had some success with my code but the images look to be in a different format when running my software. He’s contributed traffic logs from windows which I’ve put alongside mine in the git repository.
Simon (Sur3) also tried it with his 05e3:0503 device and got a seemingly different image format as well. He also took an image that came back from mine and decoded the Bayer colour space, so I can now get images back in colour!
It’s great to see other people getting involved in these efforts, as I will probably not be able to put much time towards this for a while.
April 8th, 2008 at 8:44 pm
hm, in my technical specification i read:
Integrated 1.3 MP Webcam (USB: ID 05e3:0503 Genesys Logic, Inc.)
but 640×480 would be just 0.3MP perhaps that explains why i just get the upper right corner, but the pictures i took in windows are also 640×480, that’s strange, i’ll do more research in that…
i hope i will get a bayer matrix too someday, then i could experiment with 2dim-tensor-interpolation and such strange ressource-eating algorithms :-D
greets
Simon
April 8th, 2008 at 9:48 pm
well, i did some usbsniffing: http://homepages.uni-paderborn.de/neuron/gentoo/gl860/05e3_0503.tbz2
i hope it is correct, because in amcap i didn’t see any picture, but i think perhaps it was just too dark for the camera ;-)
April 8th, 2008 at 11:06 pm
under windows I can change resolution and make it much higher (up to 1280×1024 if I remember correctly). I just started with 640×480 as that is the amcap default.
April 8th, 2008 at 11:31 pm
Your sniffing is quite strange, it seems to me that there’s not any pictures but maybe a continuous stream with nothing looking like a frame separator. The packet which marks the end of an image is present one times in the log after the reception of more than 3,500,000 octets instead of 307,200. If there were an image, it will be too dark : data range often in 0-4 (/255). Moreover you have receive a lot of times a byte from the webcam, which was not the case in the log of Daniel and mine.
I have a 5e3:f191 and the first image received under Windows is the lower right corner, the next ones are normal, although quite often there ‘s a vertically flipped frame after the first one.
April 9th, 2008 at 12:39 am
@dsd
i just can reduce the image size in amcap, but perhabs i’m blind, perhaps i’ll search for it tomorrow…
@O Lorin
yes, it was really dark and as i interpret the data from capture_to_file right my camera needs about 10 pictures before you can see anything usefull.
@*
well as i mentioned i experimented a bit with the capture_to_file and recognised, that the picture slowly fades-in from the earlier mentioned ( http://homepages.uni-paderborn.de/neuron/pix/cap09.pgm ), so after about 40 pictures i get an “normal” (ok not really normal) image in a strage format/color-coding, but it still moves and is just a corner.
thanks&greets
Simon
April 9th, 2008 at 2:36 am
Some good news for my cam: the initialization sequence isn’t usefull. I made sniffings while changing the image size and most of the traffic during the modification equals the initialization sequence. So I suppress the do_init function
and it works perfectly. This is pertinent as the light of my cam switch off and switch on later when I use capture_to_files. I’ll study what are the minor changes in the sequences.
April 9th, 2008 at 9:23 am
hehe, what a strange hack, when i suppress the do_init, then the i see quiet the same pictures as with it, but the light of my camera doesn’t turn on at all.
April 9th, 2008 at 9:25 am
…and i think this is a feature and not a bug: “no, i’ m not recording… lalalaa” :-D
April 9th, 2008 at 2:22 pm
:-)
Finally got something with the capture_x11 program. Also disabled do_init().
Image is:
- black-white
- and upside-down
- Right-top corner only. Have not tried to see what percentage of the screen.
Patrick
April 10th, 2008 at 2:35 am
Lastest news : I can see the video in 640×480, 800×600 and 1280×960. In both biggest sizes, the image is verticaly flipped…
April 10th, 2008 at 9:01 am
ok i did some more usbsniffing…
this is the log from activating the webcam in the device manager:
http://131.234.130.200/neuron/gentoo/usb_05e3_0503/aktivieren.log
and this is amcap:
http://131.234.130.200/neuron/gentoo/usb_05e3_0503/amcap.log
and finally amcap with changing the size and format of the images:
http://131.234.130.200/neuron/gentoo/usb_05e3_0503/changing.log
but i dont see any pictures in amcap, when usbsnoop is installed…
also i recognised, that i can choose between I420 (standard), RGB and YUY2, with YUY2 i can select a maximum image-size of 1600×1200 but with I420 and RGB only 640×480
April 10th, 2008 at 5:17 pm
In changing.log, there is some images but some important part of the traffic is lost. Maybe the logging by sniffusb takes too many times and data are lost. A lot of End-Of-Image markers are lost, this may be why amcap doesn’t succeed in showing any image. Here is the program I use to study your log : http://sharefiles.us/download.php?id=BB78BD2D1 (just make “relog file.log”). It try also to write the images in the log.
April 10th, 2008 at 10:06 pm
thanks, i will check it out.
April 10th, 2008 at 11:22 pm
Lastest news : 800×600 and 1280×960 aren’t only verticaly flipped, they are also horinzontaly flipped. My webcam isn’t able to do 160×120, 320×240, 1024×768, under 640×480 the frames are in 640×480, this is the viewer which to do the size reduction. The same for 1024×768 which is a to-be-reduced 1280×960. Maybe it’s the same for your model, three real image size.
I control also the luminosity, light source and retrolighting of the cam. These three parameters are obviously part of the initialization sequence. Often these commands are sent twice. I’ll try other feature and try to see if these infos are received during an initialization sequence.
October 21st, 2018 at 11:16 am
see
dsd