Monday, 2016-01-25

*** tpb has joined #timvideos00:00
mithro_florent_: so, with your changes the Opsis firmware seems to take ~3 hours to generate?00:14
_florent_which changes?00:14
_florent_(ie my fork or the official repo?)00:15
_florent_here it's 20 minutes, but that's still too long!!00:15
mithro_florent_: your jpeg ones00:32
mithro_florent_: when compiling my "florent" branch00:33
_florent_ah ok, strange here it's fast00:33
_florent_pfff the pixel issue is not easy to fix without simulation...00:34
mithro_florent_: how long till you pass out, if you haven't already :P01:03
_florent_a last implemementation and I sleep :)01:05
mithro_florent_: when you get up, we need to sync everything up so I can repo your work01:21
_florent_ok01:21
_florent_going to sleep now01:21
_florent_bye01:21
CarlFKmithro just drove off, will be back in hours01:23
mithroCarlFK: I'm right here01:23
CarlFKlol01:23
CarlFKnever mind01:23
mithroCarlFK: can you tell micolous and deeprave that the testing document is updated with some more logical instructions02:37
CarlFKmithro: k - deeprave went off to get his mac fixed.02:37
*** tija has joined #timvideos05:26
*** abhinav95 has joined #timvideos05:48
*** abhinav95 is now known as amod9505:49
*** amod95 has quit IRC08:01
*** se6astian has joined #timvideos08:44
*** Bertl_oO is now known as Bertl08:46
mithro_florent_: hey - I'm around now11:28
_florent_ok me too11:29
mithro_florent_: Dinner, Fix micolous's problems, Fix xfxf's problems, now to fix your problems :P11:29
_florent_Me?, I never have problems :)11:29
_florent_for the encoder buffer stuff11:30
_florent_can I create a branch in timvideos repo?11:30
_florent_and then do a pull request?11:30
_florent_I've been able to fix the bad pixel problem11:30
mithro_florent_: so normally you would just send a pull request directly from a branch in your own repo11:31
mithrobut your own repo appears to be in a pretty mixed up state11:31
_florent_yes but my repo has some others changes...11:31
_florent_yes11:31
_florent_so is it ok if I create a branch in the timvideos repo?11:32
mithro_florent_: yes11:32
_florent_ok11:32
mithro_florent_: so your plan is to create a branch in the timvideos repo, manually copy the changes accross and commit it?11:32
_florent_yes and wait travis result to merge it11:33
mithro_florent_: okay11:35
mithro_florent_: do you have travis enabled on your own repo?11:35
_florent_no11:35
mithro_florent_: do you want me to show you how?11:36
_florent_yes, but maybe later?11:37
mithro_florent_: okay sure11:39
mithro_florent_: so if I understand correctly, you have11:40
mithro * Fix for the encoder fps restriction (merged)11:40
mithro * Fix for the encoder in own clock domain (unmerged)11:40
mithro * Fix for reducing the buffering needed by the encoder (unmerged)11:40
mithro * Fix for the white bar pixel issue (unmerged)11:40
mithro_florent_: Is the last three correct?11:53
_florent_sorry?11:54
_florent_is it possible to stop a build on travis?11:55
mithro_florent_: yes12:01
mithro_florent_: not much reason too though12:02
mithro_florent_: click the little circle X12:02
_florent_where?:)12:03
_florent_(maybe I don't see it since not allowed to do that)12:04
mithroI have no idea how to give you permission to that....12:05
mithro_florent_: just let it fail12:05
_florent_ok12:06
*** 64MAAUVAO has quit IRC12:13
mithro_florent_: why 66MHz for the encoder?12:44
_florent_timing problems it seems if we set it faster12:46
_florent_I'll try to increase it12:46
_florent_ok going out for lunch12:47
_florent_travis should be working, and I hope to have good results when I'll be back :)12:47
mithro_florent_: welcome back :)13:55
_florent_hehe13:55
_florent_ok so I think it's good now13:55
_florent_I'll create the pull request13:55
mithro_florent_: so, what does this branch contain?13:55
_florent_it removes the large buf fifo on the jpeg encoder and fix the issues I had with this new buffer (artifacts that CarlFK saw while recording)13:56
_florent_and also enable 1080p3013:57
mithro_florent_: okay13:57
mithro_florent_: With these changes, does the encoder reach 30fps if the USB can reach that?13:59
_florent_yes14:00
mithro_florent_: great!14:00
mithro_florent_: does this also fix https://github.com/timvideos/HDMI2USB-misoc-firmware/issues/180 ?14:00
tpbTitle: Test pattern failure - black/white alternating lines · Issue #180 · timvideos/HDMI2USB-misoc-firmware · GitHub (at github.com)14:00
_florent_probably yes (I did a test with a similar pattern, it was OK)14:01
mithro_florent_: that issue was fixed by https://github.com/enjoy-digital/HDMI2USB-misoc-firmware/commit/253e44dbbea7885490d9c76aaeed7ed70d4dc53a ?14:02
tpbTitle: gateware: fix bad pixel for each 8x8 block · enjoy-digital/[email protected] · GitHub (at github.com)14:02
_florent_yes with the fifo just before the encoder14:06
mithrookay, great14:06
_florent_mithro: can you cancel build 372 and 374?14:07
_florent_thanks14:11
_florent_if 375 is fine, I'll merge14:12
_florent_next task will be to look at the HV30?14:12
mithro_florent_: yes please14:13
mithro_florent_: as far as I understand, they should be output 1080i at the moment14:14
mithro_florent_: is there anything left in your repo that hasn't been merged into ours?14:14
_florent_I don't think so, but I'm going to do a cleanup of my repo14:15
mithro_florent_: yeah - it would be good if your repo was the same as upstream - much easier for me to follow what is happening then14:16
_florent_so for the HV30, from what I understand you can to capture 1080i, compress it, send it over usb, decompress it and then deinterlace it?14:16
mithro_florent_: first step with the HV30s is to detect a 1080i signal coming from them14:17
_florent_don't you think that the compression/decompression stages on the 1080i will mess up a little bit the image?14:17
_florent_we are detecting it correctly14:17
_florent_just that the image is black for now14:17
mithro_florent_: Should it be passing the image through?14:18
mithrowe also don't get anything in the status output?14:18
_florent_maybe not if the buffer is not updated in the software14:18
_florent_maybe the software fails to check the length of the DMA and then does not update the buffer number14:19
_florent_so on the output we are only reading the black buffer14:19
_florent_status output is fine14:19
_florent_same as the monitor I tested the HV30 with14:19
mithro_florent_: hrm... Last we tried, we aren't seeing any input detected here14:20
_florent_here it's fine14:21
mithro_florent_: do you know what "pull down" is?14:21
_florent_I read your issue about that14:21
_florent_but that's probably a bit of work to implement it in hardware14:22
mithroso, according to the web - if you put the camera into "cine" mode - it should be producing a HDMI signal which is pulldown14:22
_florent_ok14:22
_florent_let's try to get an image first14:22
mithro_florent_: so, we need to verify that14:22
mithro_florent_: if the camera is producing a pull-down signal - then we should some HDMI frames which have identical contents14:24
mithro_florent_: which is one reason why I had ysionneau working on the CRC generation of HDMI frames14:25
_florent_ok14:25
mithro_florent_: the detection algorithm for pull down is found at http://deinterlace.sourceforge.net/32Spec.htm14:26
tpbTitle: DScaler - Mark Rejhon's 3:2 Pulldown Algorithm (at deinterlace.sourceforge.net)14:26
mithrothe detection algorithm should only need to be written running on the lm32 - because it can just compare the CRC values14:27
_florent_ok, but I'm not going to implement that now :)14:27
_florent_first: get a picture14:28
_florent_then try to send that to the computer and do software processing14:28
mithro_florent_: yeah14:28
mithro_florent_: that is what I'm calling "stage 1" :)14:28
mithrothe (get a picture from the HV30 onto the computer)14:29
mithrostage 2 is trying to get the pull-down deinterlacing working14:29
_florent_ok14:30
*** tija has quit IRC14:31
mithro_florent_: the CRC generation would also be useful for a selftest mode (send a known HDMI signal, check the CRC value captured is the same)14:32
_florent_After Tim has too many projects --> Tim has too many ideas :)14:34
mithroI have too many projects because I have too many ideas I think :P14:34
mithroanyway, I should probably head to bed14:37
mithro_florent_: we could really use https://github.com/timvideos/HDMI2USB-misoc-firmware/issues/94 too14:41
tpbTitle: Have frequency (and hsync/vsync) information on the inputs in the status message · Issue #94 · timvideos/HDMI2USB-misoc-firmware · GitHub (at github.com)14:41
*** se6astian is now known as se6astian|away16:27
*** se6astian|away is now known as se6astian16:53
*** Joelw has quit IRC20:20
*** Joelw has joined #timvideos20:20
*** Bertl is now known as Bertl_zZ21:24
*** se6astian is now known as se6astian|away22:18

Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!