[00:09] *** djrscally has quit IRC (Ping timeout: 480 seconds) [01:54] *** lplc_ has joined #linux-media [01:56] *** lplc has quit IRC (Ping timeout: 480 seconds) [02:32] *** danitool has quit IRC (Remote host closed the connection) [04:42] *** epoll has quit IRC (Ping timeout: 480 seconds) [04:51] *** epoll has joined #linux-media [04:57] *** prabhakalad has quit IRC (Quit: Konversation terminated!) [04:58] *** prabhakalad has joined #linux-media [05:22] *** Mo-jorisin has joined #linux-media [05:29] *** Mo_ has quit IRC (Ping timeout: 480 seconds) [05:43] *** Mo-jorisin has quit IRC (Remote host closed the connection) [06:19] *** dcz has joined #linux-media [06:51] *** dcz_ has joined #linux-media [06:54] *** dcz has quit IRC (Ping timeout: 480 seconds) [07:01] *** dcz has joined #linux-media [07:03] *** djrscally has joined #linux-media [07:08] *** dcz_ has quit IRC (Ping timeout: 480 seconds) [07:34] *** tmerciai has joined #linux-media [07:38] *** tmerciai has quit IRC (Remote host closed the connection) [07:39] *** tmerciai has joined #linux-media [07:44] *** ao2 has joined #linux-media [07:56] *** sgruszka has joined #linux-media [07:57] *** frieder has joined #linux-media [08:09] *** frieder has quit IRC (Ping timeout: 480 seconds) [08:18] *** frieder has joined #linux-media [08:48] *** frieder has quit IRC (Ping timeout: 480 seconds) [09:00] *** frieder has joined #linux-media [09:04] *** dcz has quit IRC (Quit: Konversation terminated!) [09:29] *** chewitt has joined #linux-media [09:38] <ribalda> I am not aware of `git fetch XXX --depth UNTIL_SHAX`. [09:38] <ribalda> we could probably have a remte service that calculates depths.... [10:15] *** lplc_ has quit IRC () [10:15] *** lplc has joined #linux-media [10:29] *** mvaittin has joined #linux-media [10:32] *** jmassot_ has quit IRC (Remote host closed the connection) [12:19] *** xecuter has joined #linux-media [13:23] *** mvaittin has quit IRC (Ping timeout: 480 seconds) [13:41] *** frieder has quit IRC (Ping timeout: 480 seconds) [13:45] *** Mo_ has joined #linux-media [13:50] *** frieder has joined #linux-media [14:21] *** sgruszka has quit IRC (Ping timeout: 480 seconds) [14:40] *** sgruszka has joined #linux-media [14:54] *** wallacer has quit IRC (Read error: Connection reset by peer) [14:54] *** wallacer has joined #linux-media [15:08] <pinchartl> ribalda: show about --shallow-exclude ? [15:24] *** frieder has quit IRC (Remote host closed the connection) [15:59] *** Mo_ has quit IRC (Quit: Leaving) [16:06] <ribalda> don't we want the oppossite of exclude. Exclude kill everything from that SHA. [16:26] <pinchartl> clone -next, exclude last -rc1 [16:26] <pinchartl> so everything earlier than last rc1 will be excluded [16:41] *** chewitt has quit IRC (Quit: Zzz..) [16:42] *** chewitt has joined #linux-media [16:44] *** chewitt has quit IRC (Read error: Connection reset by peer) [16:59] *** mvchtz has quit IRC (Remote host closed the connection) [16:59] *** mvchtz has joined #linux-media [17:15] *** dcz has joined #linux-media [17:16] *** BrianG61UK has quit IRC (Read error: Connection reset by peer) [17:49] <pinchartl> ribalda: I tried --shallow-exclude and get "fatal: expected 'packfile'" :-/ [17:49] <pinchartl> I wonder if it requires support on the server side [17:56] *** sgruszka has quit IRC (Ping timeout: 480 seconds) [17:57] *** tmerciai has quit IRC (Ping timeout: 480 seconds) [18:02] *** ayaka has quit IRC (Server closed connection) [18:03] *** ayaka has joined #linux-media [18:41] *** gnuiyl has quit IRC (Server closed connection) [19:04] *** epoll has quit IRC (Server closed connection) [19:05] *** epoll has joined #linux-media [19:42] *** ao2 has quit IRC (Quit: Leaving) [19:48] *** djrscally has quit IRC (Quit: Konversation terminated!) [19:54] *** djrscally has joined #linux-media [20:02] *** tmerciai has joined #linux-media [20:02] *** dcz has quit IRC (Read error: No route to host) [20:05] *** dcz has joined #linux-media [20:05] *** Spr0cket has quit IRC (Server closed connection) [20:05] *** Spr0cket has joined #linux-media [20:07] *** DarkOK has joined #linux-media [20:10] *** tmerciai has quit IRC (Ping timeout: 480 seconds) [20:22] *** dcz_ has joined #linux-media [20:26] *** dcz has quit IRC (Ping timeout: 480 seconds) [20:53] *** lvrp16_ has quit IRC (Server closed connection) [20:54] *** lvrp16_ has joined #linux-media [20:59] *** dcz_ has quit IRC (Quit: Konversation terminated!) [21:01] *** sgruszka has joined #linux-media [21:03] *** BrianG61UK has joined #linux-media [21:13] *** djrscally has quit IRC (Quit: Konversation terminated!) [21:14] *** arnd has quit IRC (Server closed connection) [21:15] *** arnd has joined #linux-media [21:19] *** conchuod has quit IRC (Server closed connection) [21:19] *** conchuod has joined #linux-media [21:25] *** djrscally has joined #linux-media [21:27] *** BrianG61UK has quit IRC (Read error: Connection reset by peer) [21:36] *** ribalda has quit IRC (Server closed connection) [21:36] *** ribalda has joined #linux-media [22:01] *** sgruszka has quit IRC (Ping timeout: 480 seconds) [22:27] <DarkOK> Hi! Not sure if this is a good place to inquire about this, but during my time trying to capture from my USB capture device, I've noticed that the general brightness of the image is different between software. Most notably, qv4l2 is the brightest, but interestingly only with it set to YUYV (unlike other software), with MJPEG being fine. [22:28] <DarkOK> ffplay and OBS seem to provide the most accurate image (with very slight difference), but OBS can lag eventually and ffplay noticeably slows down my X session. qv4l2 seems to provide the best latency with streaming method set to read() with 1 buffer (which doesn't work with mmap), which is why its accuracy is an issue for me. Could anybody look into this? I'm not nearly knowledgeable about V4L2 to do so. [22:28] <DarkOK> As a visual representation, if interested: https://darkok.xyz/v4l2/badbrightness [22:31] <pinchartl> DarkOK: that could be a colorspace issue [22:31] <pinchartl> is it an HDMI capture device ? what driver does it use ? [22:32] <DarkOK> Yeah, I thought as much at first, but I'm not able to change it to anything else other than sRGB. It is an HDMI capture device, and does use uvcvideo. [22:33] <pinchartl> you said qv4l2 provides the best latency with read() [22:33] <pinchartl> but the uvcvideo driver doesn't support read() [22:34] <DarkOK> That's strange, it is shown as an option in qv4l2, and if I set it to memory mapped instead I'm unable to set the buffer count to 1, as it provides no video until I set it to 2. [22:35] <pinchartl> qv4l2 has good colorspace support, but expects the device to report accurate colorspace information [22:35] <pinchartl> I'd investigate that [22:36] <pinchartl> I don't think your device produces different brightness for different applications, it's more likely that different applications convert YUV to RGB with different colorspace handling [22:36] <pinchartl> it could also be that some applications would set V4L2 controls such as contrast and brightness, if your device supports them [22:37] <pinchartl> you can use v4l2-ctl to list controls and their values while applications are running, and check if they differ [22:38] <DarkOK> Hmm, it doesn't seem to be the controls, as they're the same values when switching between qv4l2 & ffplay [22:39] <pinchartl> then I'd investigate colorspace handling [22:42] *** djrscally has quit IRC (Quit: Konversation terminated!) [22:42] <DarkOK> Not entirely sure how I would. One thing that could be of value though is that I can't change the options in qv4l2 though beyond colorspace = sRGB, transfer function = Rec. 709 and capture image format only being yuyv & mjpeg [22:42] <DarkOK> I'm no expert in this :P [22:43] *** djrscally has joined #linux-media [22:46] <DarkOK> I've went between colorspaces on what I'm capturing too, but qv4l2 does always end up brighter. Setting it to YCbCr709 does make it have less colour banding, though (which I suppose would make sense because that matches what I said before?) [22:47] *** djrscally has quit IRC () [22:49] <pinchartl> I've just checked, and read() support is available because qv4l2 opens the device through libv4l2 by default [22:49] <pinchartl> that's shown by "Device /dev/video0 (wrapped)" [22:49] <pinchartl> libv4l2 emulates read() [22:50] <pinchartl> if you use Open Raw Device [22:50] <pinchartl> you won't get read() [22:51] <pinchartl> as for setting the colorspace, you can't do that [22:51] <pinchartl> UVC supports reporting the colorspace produced by the device [22:51] <pinchartl> but not setting it [22:52] <pinchartl> can you share the output of `lsusb -v` for your device ? [22:52] <pinchartl> (you can use paste.debian.net) [22:52] *** Mo has joined #linux-media [22:55] <DarkOK> Right, opening as raw shows user pointer instead. With that however I also can't select 1 buffer. As for lsusb I'll provide that, one moment [22:57] <pinchartl> mmap() is the optimal option performance-wise [22:57] <pinchartl> and the streaming method shouldn't affect the contents of the image [22:57] <DarkOK> https://paste.debian.net/1361563/ (as an aside, I'm really ought to update my kernel, but my laptop on the latest is also affected) [22:59] <DarkOK> Yeah, I know mmap should be optimal. It isn't affecting the contents of the image either it's just that my reasons for preferring qv4l2 in this case rather than the other software that already works is that I can get better latency with it [23:00] <pinchartl> so your device reports [23:00] <pinchartl> bColorPrimaries 1 (BT.709,sRGB) [23:00] <pinchartl> bTransferCharacteristics 1 (BT.709) [23:00] <pinchartl> bMatrixCoefficients 4 (SMPTE 170M (BT.601)) [23:01] <pinchartl> that's translated in V4L2 terms by the driver to V4L2_COLORSPACE_SRGB, V4L2_XFER_FUNC_709, V4L2_YCBCR_ENC_601 [23:02] <pinchartl> I suspect that qv4l2 honours those [23:02] <pinchartl> and decides that YUYV is in limited range while MPJEG is in full range [23:02] <pinchartl> while your device likely uses full range or limited range for both [23:03] <pinchartl> that would explain the difference between YUYV and MJPEG in qv4l2 [23:03] <pinchartl> other applications may not honour the colorspace information reported by the uvcvideo driver, I haven't checked [23:04] <DarkOK> I see. I've noticed options to change between full/limited range in qv4l2 but these don't seem to be effective. They flash for about a frame but then go back to the same old [23:07] <pinchartl> I believe those try to configure the device to change the quantization produced by the hardware [23:07] <pinchartl> and as the device doesn't support changing it, it has no effect [23:08] <pinchartl> it would be nice if qv4l2 had an option to override the colorspace information reported by the device [23:09] <pinchartl> ah it does [23:09] <pinchartl> in the Capture menu [23:10] <DarkOK> Right, those are also the very options I've tried. It's also available from the context menu with the video window open [23:14] <pinchartl> the overrides don't seem to have much effect on my camera [23:19] <DarkOK> Hmm. If you have an idea of what can be patched in the source code though I'm willing to apply patches for the time being [23:19] *** Mo has quit IRC (Remote host closed the connection) [23:19] <pinchartl> it's getting a bit too late for me to dive into qv4l2 [23:20] <pinchartl> maybe hverkuil could help you tomorrow, he knows more than me about qv4l2 [23:20] <DarkOK> Ah, alrighty. Thank you for all your help though [23:30] *** tfiga has quit IRC (Server closed connection) [23:30] *** tfiga has joined #linux-media [23:40] *** ldts has quit IRC (Server closed connection) [23:40] *** ldts has joined #linux-media