| ↑back Search ←Prev date Next date→ Show only urls | (Click on time to select a line by its url) |
| Who | What | When |
|---|---|---|
| svarbanov | hverkuil: hi, is there a legal way to set IDR period for encoders | [12:22] |
| hverkuil | V4L2_CID_MPEG_VIDEO_H264_I_PERIOD perhaps? Is that what you are looking for? | [12:26] |
| svarbanov | hverkuil: i checked this CID, but it sets the I-frames period for open GOP case, but what will happen if it is closed GOP? | [12:28] |
| hverkuil | Then it uses V4L2_CID_MPEG_VIDEO_GOP_SIZE. | [12:31] |
| svarbanov | hverkuil: the venus encoder has two properties related to I-frame period (which could be set by GOP_SIZE CID) and IDR period which will define which I-frame will be IDR frame. So I'm trying to used the standard controls to give to the user control over both settings
hverkuil: i.e I-frame period (gop size) could be 32, but IDR could be every 2nd I-frame | [12:37] |
| hverkuil | svarbanov: it's a good question. I'm not an expert in this area, so let me ask you a few questions to understand this better.
Is IDR specific to the open GOP case? Or is it used for both open and closed? | [12:49] |
| svarbanov | hverkuil: I'm not sure, but seems that it could be used for both open and closed GOP | [12:51] |
| hverkuil | Because if it can be used for both, then I wonder why the GOP_SIZE control was reused instead of making an IDR_PERIOD control.
what do the other two encoders that use this I_PERIOD control do? You may have to contact the author of this control for clarification. Or dig into the h264 spec. | [12:54] |
| *** | prabhakarlad has left | [13:06] |
| svarbanov | hverkuil: ok I see, just wanted to check with you in case you know | [13:07] |
| ............................................. (idle for 3h42mn) | ||
| *** | benjiG has left | [16:49] |
| ................... (idle for 1h34mn) | ||
| lucasrangit | Can someone help me understand how the UVC extension units get mapped to V4L2 controls ? I'm developing the firmware for a UVC camera. | [18:23] |
| ndufresne | lucasrangit, I'm far from knowledgable about all the details, but they are somhow exposed through the media controller, sudo media-ctl -d /dev/media0 -p | [18:35] |
| lucasrangit | ndufresne, thanks, i'll look into that but i think i'm having issues before the controls are mapped because my uvcdynctrl xml doesn't get imported without errors | [18:40] |
| ndufresne | ah, these controls | [18:42] |
| lucasrangit | ndufresne, sorry, i don't know the terminology yet | [18:42] |
| ndufresne | lucasrangit, so, rolling back, you have an XML file that sets controls, but it does not work, some controls are missing ?
lucasrangit, have you looked inside uvc_ctrl.c ? the mapping seems rather static if I was you, I'd follow the code using UVC_VC_EXTENSION_UNIT to figure-out ah, it's seems generic, but some names and description are being "standardized" | [18:44] |
| lucasrangit | ndufresne, the mapping for standard controls is fixed https://elixir.free-electrons.com/linux/latest/source/drivers/media/usb/uvc/uvc_ctrl.c#L423
ndufresne, i'm trying to a vendor specific control but don't understand how to define the ID for the custom V4L2 control in the XML. i'm getting import errors like "libwebcam Warning: the V4L2_CTRL_FLAG_NEXT_CTRL flag does not return the next higher control ID" | [18:56] |
| ndufresne | yes, but later in the code you'll see that it will accepts other controls, but it's not clear how you can access them from userspace
does uvcdynctrl support vendor CID ? | [18:59] |
| lucasrangit, ok, found some code in gst using vendor extension (the C920), it's calling UVCIOC_CTRL_QUERY
and UVCX_VIDEO_CONFIG_COMMIT see /usr/include/linux/uvcvideo.h | [19:10] | |
| lucasrangit | ndufresne, ok | [19:12] |
| ndufresne | interesting, I wonder what UVCIOC_CTRL_MAP does ... | [19:12] |
| lucasrangit | ndufresne, take a look at the control at https://github.com/cshorler/webcam-tools/blob/master/uvcdynctrl/data/046d/logitech.xml#L109
ndufresne, a V4L2 CID for focus control is added. but why is the value set to "0x0A046D04" ? | [19:15] |
| ndufresne | is it a fourcc ?
na, 04 is too small something interesting, V4L2_CID_FOCUS does not exist in the driver, instead we have V4L2_CID_FOCUS_ABSOLUTE/AUTO there is no LED CID either | [19:16] |
| lucasrangit | ndufresne, yeah those are vendor specific UVC extension units, not part of standard UVC camera spec | [19:19] |
| ndufresne | ok, if I understand right, these are controls of UVC_GUID_UVC_PROCESSING
still, it's not clear what this number is | [19:20] |
| lucasrangit | ndufresne, yeah :\ thanks for looking at it with me. | [19:22] |
| ndufresne | I'm actually curious, someday I might need to give gstreamer uvch264src some maintenance | [19:24] |
| lucasrangit | ndufresne, coincidence that Logitech's vendor id 0x046d, which matches bytes 0x0A_046D_72 ? | [19:24] |
| ndufresne | and it's using bunch of vendor controls to control the type of H264
I don't believe it's a conincidence no could it be VENDOR_CLASS_BASE + LOGITECH_BASE + N ? | [19:24] |
| lucasrangit | ndufresne, where is VENDOR_CLASS_BASE defined? | [19:27] |
| ndufresne | nowhere in the driver apparently, was the question, maybe it's inside uvcdynctl code ...
0x04 is not a conincidence either, fromt he spec, EXTERNAL_ VENDOR_SPECIFIC 0x0400 | [19:28] |
| lucasrangit | ndufresne, that last byte varies for each control though, see V4L2_CID_LED1_MODE 0x0A046D05 https://github.com/cshorler/webcam-tools/blob/master/uvcdynctrl/data/046d/logitech.xml#L113 | [19:32] |
| ndufresne | so I guess it's a control ID somehow, this is HW specific, with a vendor base value + index | [19:35] |
| .............................. (idle for 2h29mn) | ||
| lucasrangit | ndufresne, thanks for help and sharing your insights | [22:04] |
| ↑back Search ←Prev date Next date→ Show only urls | (Click on time to select a line by its url) |