| ↑back Search ←Prev date Next date→ Show only urls | (Click on time to select a line by its url) |
| Who | What | When |
|---|---|---|
| hverkuil | syoung: ping | [12:05] |
| stevenm | hey I'm guessing if my device was supported by v4l2 i'd just plug it in and find another /dev/videoX device appear?
if I had to guess it's probably just that the vendor/product USB ID (which is 15e4:0140) just isn't associated with the right module to load in? | [12:12] |
| hverkuil | stevenm: that's likely to be the case. | [12:14] |
| *** | LazyGrizzly has left | [12:22] |
| .... (idle for 19mn) | ||
| stevenm | hverkuil, well it's basically a VHS VCR only it has a USB port on it - it's explicitly designed to let you play VHS tapes over USB (e.g. so you can convert home movies etc...)
an ION (that's the make) "VCR 2 PC" device it has no record button it - i think it's meant to solely just play them back over USB... i think it's just the most basic VCR ever with a normal USB video adapter inside it probably with quite a run of the mill chipset but announcing a strange VID/PID hverkuil, am I in need of a udev rule or something? | [12:41] |
| hverkuil | What are the VID/PID values? | [12:44] |
| stevenm | put them above ^ 15e4:0140
not sure if it uses the same chipset as "Video to PC" https://www.linuxtv.org/wiki/index.php/Ion_Video_2_PC it wouldn't suprise if this "VCR to PC" is just a VCR with a "Video to PC" adapter built in to it | [12:44] |
| ndufresne | anyone has been working / drafting some HDR10 and HDR10+ support ? | [12:47] |
| hverkuil | It has a Windows driver: so it is a proprietary format with a proprietary driver. | [12:47] |
| stevenm | stevenm begins to open it up for more clues
hverkuil, that's a leap isn't it? windows lacks many drivers - doesn't mean that linux will lack the one it needs | [12:47] |
| hverkuil | There is no support for usb video devices for that vendor in the kernel, I checked. | [12:49] |
| stevenm | got it open
it's one of these... https://www.linuxtv.org/wiki/index.php/Trident_TM6000 only a TM6010 | [12:49] |
| hverkuil | Ah, that helps. | [12:50] |
| stevenm | want a pic? :D | [12:50] |
| hverkuil | you can try adding support for it to drivers/media/usb/tm6000/tm6000-cards.c | [12:50] |
| *** | al has quit IRC (*.net *.split)
hnrk has quit IRC (*.net *.split) bbrezillon has quit IRC (*.net *.split) gnurou has quit IRC (*.net *.split) jherland_ has quit IRC (*.net *.split) broonie has quit IRC (*.net *.split) jmondi has quit IRC (*.net *.split) dagmcr has quit IRC (*.net *.split) mani_s has quit IRC (*.net *.split) arnd has quit IRC (*.net *.split) | [12:51] |
| hverkuil | ndufresne: HDR10 should be supported by V4L2, except for the fact that there still are no 10-bit pixelformats AFAIK. The colorimetry signaling is all there.
There is nothing for HDR10+ | [12:53] |
| stevenm | hverkuil, https://photos.app.goo.gl/QZeQFkfbabhwMiaXA | [12:53] |
| ndufresne | hverkuil, checking again, I don't recall seeing signalling for the mastering display and brightness level static metadata
do you know who was working on this, and for what HW ? | [12:54] |
| hverkuil | We have no drivers that implement the missing pieces. So there are definitely bits missing, but it shouldn't be difficult to expose that.
Nobody is. Inexplicably. If someone starts work on that, then I'd love to help with the missing API bits. | [12:55] |
| ndufresne | I know people working on this (learned yesterday) but inexplicably they have been solo riding this, which to be me is really weird, it's not that complicated, so coming here ensure you do this once | [12:57] |
| hverkuil | ndufresne: right, please redirect them. I'm happy to give advice and help with the v4l2 core code.
HDR10 I am fairly familiar with, HDR10+ not really. ah, HDR10+ == SMPTE ST 2094-40 Application #4 I do know about that. | [13:03] |
| ndufresne | nothing that crasy, in HDR10, the scaling data is just like colorimetry, it's pretty much static per stream, while in HD10+, it's up to per frame | [13:05] |
| hverkuil | what HW supports HDR10+? | [13:05] |
| ndufresne | for now, I got some FPGA based HW
I met some people at Netflix preping for HDR10 / HDR10+ QA (for appliance certification), but they didn't say (and probably could not) if there was actual HW ready "at Netflix" -> "from Netflix" hverkuil, short story, we are looking at getting the two form of metadata and pass them to encoders (or muxer depending on that format), and then take it back from decoder, parsers or demuxer and pass it back to the display (aka DRM) What we got now in gst is the static: Mastering Display Info, https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/blob/master/gst-libs/gst/video/video-hdr.h#L64 Content light level, https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/blob/master/gst-libs/gst/video/video-hdr.h#L113 and we has the colorimetry bits for a while already which reminds me, I how you a pointer to the time code hverkuil, so the timecode structure is there (the per frame one), https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/blob/master/gst-libs/gst/video/gstvideotimecode.h#L95 But we also got GstVideoTimeCodeConfig (line 68) and GstVideoTimeCodeInterval (line 117) | [13:06] |
| stevenm | hverkuil, it seems tm6010 is covered by the tm6000 driver these days, but it seems that's already in the mainline kernel? https://github.com/torvalds/linux/tree/master/drivers/media/usb/tm6000
so yeah just checked, the current kernel already has it :S s/the/my/ i.e. i'm on ubuntu mate 18.04 running 5.0.0-35 amd64 from the normal repo | [13:23] |
| still no idea how to actually make it react properly to being plugged in though | [13:33] | |
| hverkuil | ndufresne: you need a compound control containing the CTA-861 Static Metadata Type 1 data. That should do it. It is all based on the CTA 861 standard, so easy enough to define.
stevenm: look in drivers/media/usb/tm6000/tm6000-cards.c, table tm6000_id_table : this has a list of VID/PID values for which the driver works. They map to specific board definitions, and those you need to figure out yourself (trial and error basically). If you are lucky, one of the existing boards might work out of the box. | [13:40] |
| ndufresne | hverkuil, so a RW compound for OUTPUT type of queues, and RO for capture ?, I think we need some spec for the capture side, as unlike other control, it's an information that we receive, not an information that we configure | [13:43] |
| hverkuil | Can't help with this, you'll need to do that yourself.
ndufresne: you're talking about a decoder or an encoder? | [13:43] |
| ndufresne | hverkuil, and capture / output
for decoder, we don't need anything if we can parse it from the bitstream, that's a choice we made, and encoder, we need it for stateful encoder only | [13:44] |
| hverkuil | ndufresne: can you reply to my "[ANN] Report of Media Summit: Codecs" post with that link to the gstreamer timecode header? That way it is recorded that that action item is done. | [13:45] |
| stevenm | hverkuil, it's not in that table | [13:45] |
| hverkuil | stevenm: correct, and that's why your device isn't recognized. | [13:45] |
| stevenm | i already know this!
but don't tell me I have to recompile the whole damn thing for one extra pointer surely there is a way of doing this via udev or something? | [13:46] |
| ndufresne | hverkuil, for capture (HDMI and SDI for what I'm interested in), I'm pretty sure it's a bit tricky, since the control might be updated
an example, you may unplug your HDMI cable and plug it into another source | [13:46] |
| hverkuil | Not just an extra pointer. It's a new device, you need to define how many inputs there are, if there are gpios involved, etc. It's NOT a matter of just adding an entry in a table.
Each vendor routes pins differently, and that has to be part of the board definition. I suspect it won't be that difficult to do since there is no tuner, so the TM5600_BOARD_TERRATEC_GRABSTER would probably be a good template to start with. | [13:47] |
| stevenm | hmm i wonder if I can get udev to lie about the device to make it look like a terratec grabster | [13:50] |
| ....... (idle for 30mn) | ||
| Fede_HC | Hello guys
I have a problem with a ELP USB camera connected to the raspberry pi... the camera image (under high light conditions) gets darker and then normal again every few seconds... Maybe do you know if there is some settings i can adjust? Or do you think its a camera problem? | [14:20] |
| ..... (idle for 21mn) | ||
| stevenm | hverkuil, got it working :)
echo 'ACTION=="add", ATTRS{idVendor}=="15e4", ATTRS{idProduct}=="0140", RUN+="/sbin/modprobe xpad" RUN+="/bin/sh -c '\''echo 15e4 0140 > /sys/bus/usb/drivers/tm6000/new_id'\''"' | sudo tee /etc/udev/rules.d/59-vcr2pc.rules sudo udevadm control --reload woops i need to swap 'xpad' for 'tm6000' :P but i already ran 'modprobe tm6000' before hand anyway :) | [14:43] |
| syoung | hverkuil: pong | [14:47] |
| hverkuil | syoung: ever worked with the synaptics rmi4 driver for touchpads? | [14:47] |
| syoung | No, I haven't. I once submitted an ill-thought-out patch which rightfully got rejected. | [14:54] |
| hverkuil | That's probably where I found your name.
I'm testing the v4l-touch support and it is totally broken :-( | [14:59] |
| ................. (idle for 1h24mn) | ||
| ndufresne | hverkuil, do you think you would have a bit of time tomorrow (probably too late for everyone now) to talk about VP9 references, and the fact that the references may not have the same resolution ?
the cookie/timestamp model just don't work for this CODEC and bbrezillon is pulling his hair now | [16:23] |
| hverkuil, ok, replied for the timecode, with slightly more info, hopefully that's enough
note that on decklink, the timecode does not have the fps, since it's always the same as the configured streaming one In GStreamer, the timecode fps may differ | [16:32] | |
| hverkuil | ndufresne: thanks!
I have time tomorrow, but I have to leave around 4pm (CET). | [16:35] |
| ndufresne | ok, noted, I'll ping you before that time
I'll try to write some context down and share, that might help | [16:37] |
| hverkuil | yes, please. That will help a lot. | [16:38] |
| *** | benjiG has left | [16:42] |
| .................... (idle for 1h35mn) | ||
| piggz | hi, i have a tablet that was originally android/windows, running an atom cherrytrail.....
it now has sailfish ported to it, using the android 3.18 kernel the cameras are ov2680 and ov5648 i can find very little docs on how to set these up how might i go about grabbing a bsic image from these using v4l what do i need to do with media-ctl ? | [18:17] |
| ↑back Search ←Prev date Next date→ Show only urls | (Click on time to select a line by its url) |