Thread: [libdc1394-devel] dc1394_multiview
Capture and control API for IIDC compliant cameras
Brought to you by:
ddouxchamps,
gordp
From: Dan C. <Daniel.A.Christian@NASA.gov> - 2002-12-20 23:18:59
|
I can't get dc1394_multiview to work with multiple cameras anymore. This used to work under kernel 2.4.18. I've tried libdc1394-0.9.1 and CVS versions. It broke after I upgraded from 2.4.18 to 2.4.20-ac1. According to the FAQ, this means that the device major number changed, so I re-created /dev/video1394/0-3 dc1394_multiview works if I only have 1 camera, but seems to hang if there are two. Both cameras are Orange Micro iBots, connected through a Belkin hub. The controller is also from Orange Micro (NEC chip). -Dan If I do a strace on it, this is where it hangs: open("/dev/video1394/0", O_RDONLY) = 4 ... select(6, [5], NULL, NULL, NULL) = 1 (in [5]) read(5, "\1\371\10\0\31\0\0\0\1\0\377\277!\207\r\10\0\0\0\0\203"..., 32) = 32 read(5, ";\0\0\0\30\0\1\0\10\0\21\0ATI Radeon Video Ove"..., 100) = 100 write(5, "\220\20\2\0;\0\0\0", 8) = 8 read(5, 0xbffff700, 32) = -1 EAGAIN (Resource temporarily unavailable) select(6, [5], NULL, NULL, NULL) = 1 (in [5]) read(5, "\1\0\t\0\200\0\0\0\4\0\0\0\0300\v\10\0\0\0\0\204g8>\0\0"..., 32) = 32 read(5, "YUY2\1\0\0\0YUY2\0\0\0\20\200\0\0\252\0008\233q\20\1\0"..., 128) = 128 read(5, "UYVY\1\0\0\0UYVY\0\0\0\20\200\0\0\252\0008\233q\20\1\0"..., 128) = 128 read(5, "YV12\1\0\0\0YV12\0\0\0\20\200\0\0\252\0008\233q\f\3\0\0"..., 128) = 128read(5, "I420\1\0\0\0I420\0\0\0\20\200\0\0\252\0008\233q\f\3\0\0"..., 128) = 128write(2, "using Xv format 0x59565955 UYVY "..., 39using Xv format 0x59565955 UYVY packed ) = 39 ioctl(4, 0x3 |
From: Fernando B. <ISk...@gm...> - 2004-04-12 19:06:19
|
Hi Damien! > It should work with one card as far as you don't go over the max > bandwidth. But even in this case there would not be IOCTL errors. > How can i know if i got over the bandwidth? I mean the format size for one frame i dont know ( where i could read something about the difference between YUV411 and YUV22 ??). > > Coriander can set both cameras in send modus. So i can see in > Performances > > that both cameras send . > > ( mode MODE_640x480_YUV411 ) and i can Display one of the cameras at a > > moment, but not both. > > Dual display is not supported in coriander. Yes i know, i tried to modify coriander to do this, but i couldnt manage it. Would you be interested in doing this ? I mean i could help, if i knew more about coriander... The Performances in coriander get for one camera receiving in mode MODE_640x480_YUV411 39 % of the bandwidth, with 2 cam receiving 79 %. But i could only get a picture of one camera. When coriander says receiving what does it means exactly? that cameras[i].capture_buffer are filled or that dc1394_dma_single_capture(&cameras[0]); returned? > > Is this also possible with dc1394_multiview? > > I think so, but I did not used it recently. Just try it! ;-) MODE_640x480_YUV411 works great for one camera. I tried to get MODE_640x480_YUV422 but only when i change the bandwidth to 400 it shows pitcutres. i coulndt get the MODE_640x480_RGB at work at all. I wonder if dc1394_dma_multi_capture(cameras, numCameras); for two cameras in MODE_640x480_YUV422 will work at all. I will get the 2nd camera but only in the next 2 days. I will tell you then if i can receive. Thx a lot Fernando -- NEU : GMX Internet.FreeDSL Ab sofort DSL-Tarif ohne Grundgebühr: http://www.gmx.net/info |
From: Fernando B. <ISk...@gm...> - 2004-04-14 15:02:30
|
Hi! It works now somehow.... i got 2 different cams one from apple and the sony dfw-vl500... Tha apple one MUST be in the first slot, but it works then. But only in format YUV 4:1:1. How can i transform it to ppm? dc1394_multiview uses Xv functions. i would like to know how i could use the iyu12yuy2 to transform it to ppm images... Thx a lot -- NEU : GMX Internet.FreeDSL Ab sofort DSL-Tarif ohne Grundgebühr: http://www.gmx.net/info |
From: Dan D. <da...@de...> - 2003-01-07 03:46:53
|
On Mon, 2003-01-06 at 20:25, Dan Christian wrote: > I can't get libdc1394/examples/dc1394_multiview to work with multiple > cameras anymore. This used to work under kernel 2.4.18-10. I've tried > libdc1394-0.9.1 and CVS versions. This is my responsibility, and you have reported it before, but I am going to test it now. > It broke after I upgraded from 2.4.18 to 2.4.20-ac1. It's still broken > under 2.4.20-ac2. I am testing with a Marcello kernel.org 2.4.20 image only. I doubt there is a difference, but if you know where he lists information maybe about his changes please check for me. > According to the FAQ, the device major number > changed, so I've re-created /dev/video1394/0-3. FYI, the trailing numebers are only used to address more than one host adapter (technically, a port). I will test single port and multiple ports. Since it is supposed to work either way. BTW, I have been using rev 675 from the main linux1394 trunk for the past month and a half. I have successfully tested dc1394_multiview with 2 cameras on a single port and across mutliple ports. Checking the logs, there is not much changes in video1394 since 2.4 was branched both in trunk and the branch. Therefore, I will be surprised if it is not working. Otherwise, it ought to be easy to isolate the bug since so few changes. Well, its done compiling 2.4.20, time to reboot and test.... +-DRD-+ |
From: Dan D. <da...@de...> - 2003-01-07 03:56:12
|
Yes, it works fine for me. 2 cameras on one or multiple ports. On Mon, 2003-01-06 at 22:36, Dan Dennedy wrote: > On Mon, 2003-01-06 at 20:25, Dan Christian wrote: > > I can't get libdc1394/examples/dc1394_multiview to work with multiple > > cameras anymore. This used to work under kernel 2.4.18-10. I've tried > > libdc1394-0.9.1 and CVS versions. > > This is my responsibility, and you have reported it before, but I am > going to test it now. > > > It broke after I upgraded from 2.4.18 to 2.4.20-ac1. It's still broken > > under 2.4.20-ac2. > > I am testing with a Marcello kernel.org 2.4.20 image only. I doubt there > is a difference, but if you know where he lists information maybe about > his changes please check for me. > > > According to the FAQ, the device major number > > changed, so I've re-created /dev/video1394/0-3. > > FYI, the trailing numebers are only used to address more than one host > adapter (technically, a port). I will test single port and multiple > ports. Since it is supposed to work either way. > > BTW, I have been using rev 675 from the main linux1394 trunk for the > past month and a half. I have successfully tested dc1394_multiview with > 2 cameras on a single port and across mutliple ports. Checking the logs, > there is not much changes in video1394 since 2.4 was branched both in > trunk and the branch. Therefore, I will be surprised if it is not > working. Otherwise, it ought to be easy to isolate the bug since so few > changes. > > Well, its done compiling 2.4.20, time to reboot and test.... > > +-DRD-+ > > > > > ------------------------------------------------------- > This SF.NET email is sponsored by: > SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See! > http://www.vasoftware.com > _______________________________________________ > Mailing list for libdc1394-devel > lib...@li... > https://lists.sourceforge.net/lists/listinfo/libdc1394-devel -- Dan Dennedy <da...@de...> |
From: Dan C. <Daniel.A.Christian@NASA.gov> - 2003-01-07 23:22:13
|
I just tried a plain 2.4.20 (no patches). It still doesn't work for me. Same as before: one camera is fine, two hangs in an ioctl. The system is a single Athlon XP2000+ with 1Gb SDRAM. High memory support is off MTRR enabled Local APIC enabled IO-APIC enabled All the IEEE-1394 stuff is loaded as modules: insmod ieee1394 && insmod ohci1394 attempt_root=1 \ && insmod raw1394 && insmod video1394 Any ideas? -Dan |
From: Dan D. <da...@de...> - 2003-01-08 04:57:49
|
On Tue, 2003-01-07 at 18:35, Dan Christian wrote: > OK, I have some more info. > > If it reload the drivers -- with the cameras plugged in -- then it > works. > > If I unplug both cameras and plug them back in, then it fails. > Unplugging only one camera still works. There is nothing else plugged > in (no hub). > > Any ideas why the hot-plug fails? I dunno. I did my tests by plugging in my second cam and just running dc1394_multiview. video1394 had already been loaded at boot time. Also, I was plugging and unplugging between host adapters (ports) testing those cases as well without reloading video1394. Seems rather quiet these days on the Digital Camera front. It would be nice to hear if others have a problem. Is the relative quiet a good (little problems) or bad (given up) thing? If it's very important, and since I can't reproduce it here, you could fly me out to your location to attempt to debug it. |
From: Ryutaroh M. <ryu...@it...> - 2003-01-11 04:05:07
|
> Seems rather quiet these days on the Digital Camera front. It would be > nice to hear if others have a problem. Is the relative quiet a good > (little problems) or bad (given up) thing? The following video1394 problem with kernel hacking options still occurs with kernel 2.4.21pre3, which has newer Linux1394 stack than 2.4.20. Ryutaroh | Subject: video1394: 160x120 resolution doesn't work. | From: ryu...@it... | To: lin...@li... | Date: Wed, 09 Oct 2002 22:22:37 +0900 (JST) | | Is there anyone succeeding in capturing 160x120 images from a 1394 | digital camera via the video1394 interface? | | Specifically, the following configuration doesn't work for me. | | kernel version: 2.4.19 through 2.4.20pre10 (2.4.18 worked fine) | resolution: 160x120 (other resolutions work fine) | interface: video1394 (raw1394, i.e. /dev/raw1394 works fine) | library: libdc1394 0.9.0-2, libraw1394 0.9.0 | camera: iREZ StealthFire, OrangeMicro iBot | | | Subject: Re: video1394: 160x120 resolution doesn't work. | From: ryu...@it... | To: da...@de... | Cc: lin...@li... | Date: Sat, 12 Oct 2002 14:29:13 +0900 (JST) | | Thank you for your information. | | Linux 2.4.19 + linux1394 rev 578 with almost standard kernel configuration | also works fine for me. But when I enable | | Debug memory allocations | Memory mapped I/O debugging | Compile the kernel with frame pointers | | kernel compilation options in "Kernel hacking" section, Video 1394 does | not work with 160x120 resolution. | | I also verified that Linux 2.4.20pre10 without "Kernel hacking" options | works fine. |
From: Dan C. <Daniel.A.Christian@NASA.gov> - 2003-01-07 23:35:47
|
OK, I have some more info. If it reload the drivers -- with the cameras plugged in -- then it works. If I unplug both cameras and plug them back in, then it fails. Unplugging only one camera still works. There is nothing else plugged in (no hub). Any ideas why the hot-plug fails? -Dan On Tuesday 07 January 2003 15:22, Dan Christian wrote: > I just tried a plain 2.4.20 (no patches). It still doesn't work for > me. Same as before: one camera is fine, two hangs in an ioctl. > > The system is a single Athlon XP2000+ with 1Gb SDRAM. > High memory support is off > MTRR enabled > Local APIC enabled > IO-APIC enabled > > All the IEEE-1394 stuff is loaded as modules: > insmod ieee1394 && insmod ohci1394 attempt_root=1 \ > && insmod raw1394 && insmod video1394 |