| *** tpb has joined #timvideos | 00:00 | |
| mithro | cr1901_modern: Yes, I was asleep | 00:17 |
|---|---|---|
| cr1901_modern | mithro: I can't find your PLL docs in either the upstream repo or your copy. Is it on a specific branch? | 00:35 |
| mithro | cr1901_modern: https://github.com/timvideos/HDMI2USB-misoc-firmware/blob/master/targets/opsis_base.py#L36 | 00:35 |
| tpb | Title: HDMI2USB-misoc-firmware/opsis_base.py at master · timvideos/HDMI2USB-misoc-firmware · GitHub (at github.com) | 00:35 |
| mithro | cr1901_modern: https://github.com/timvideos/HDMI2USB-misoc-firmware/blob/master/targets/common.py#L7 | 00:37 |
| tpb | Title: HDMI2USB-misoc-firmware/common.py at master · timvideos/HDMI2USB-misoc-firmware · GitHub (at github.com) | 00:37 |
| cr1901_modern | Okay, so looks like sys clock generation using PLL on Opsis makes sense, since you use so many of the PLL outputs. | 00:38 |
| cr1901_modern | (Although the DCMs could probably do it as well, it's prob more effort than it's worth) | 00:40 |
| cr1901_modern | On the minispartan however, I still think it's worth a shot to swap to DCMs to conserve PLLs. | 00:41 |
| cr1901_modern | mithro: The input is absolutely gonna need a PLL, no question. Is the HDMI output PLL used for anything besides frequency synthesis? | 00:41 |
| mithro | cr1901_modern: If we can free up the PLLs that would always be useful | 00:42 |
| cr1901_modern | If the HDMI output is only useful for frequency synthesis that could possibly also be moved to a DCM. But I'll just focus on the sysclk generation on minispartan for now | 00:44 |
| mithro | cr1901_modern: We still only have 4 on the Opsis | 00:44 |
| cr1901_modern | mithro: It probably can be done on the Opsis too, but it looks a bit tougher b/c now there's 5 different clock domains | 00:46 |
| cr1901_modern | I'll test on minispartan_base, just to get it working, and we can go from there. The PLLs should be used when we are expecting a specific frequency input (output?), but we don't actually have control over the signal we are receiving (generating?). | 00:48 |
| mithro | cr1901_modern: SGTM | 00:48 |
| cr1901_modern | :) | 00:48 |
| cr1901_modern | Because we have no control over what the actual frequency an external equipment will generate, all the HDMI inputs need PLLs- no way around it. | 00:48 |
| mithro | cr1901_modern: The main reason is you need the PLL is to generate the bit clock from the pixel clock | 00:50 |
| cr1901_modern | mithro: I suspect a DCM could generate the bit clock as well. | 00:55 |
| cr1901_modern | (On the output*) | 00:55 |
| cr1901_modern | On the input, if your input frequency is not quite, say 75MHz, a PLL will modify it's output to generate a "not quite 750MHz" bit clock, and track/predict changes in the input frequency. I am not sure if this tracking works with DCMs. | 00:59 |
| *** paradisaeidae has joined #timvideos | 04:25 | |
| *** rohitksingh_work has joined #timvideos | 04:37 | |
| *** ssk1328 has joined #timvideos | 04:51 | |
| mithro | xfxf: Just watching damion's keynote and we need live image stabilisation :-P | 05:31 |
| xfxf | we need better tripods and camera raisers | 05:32 |
| nueces | hola! | 05:32 |
| mithro | xfxf: image stabilisation only costs software | 05:32 |
| xfxf | I might invest in a few heavy ones for local usage | 05:32 |
| xfxf | They just tend to be megga exxy to ship around | 05:32 |
| xfxf | ya, but we can avoid it completely | 05:33 |
| nueces | one short questions. There is in the roadmap fix the issue that force us to use the redmere cables? | 05:33 |
| nueces | I'm asking this because here that cable cost close to US 200 :/ | 05:33 |
| xfxf | you can't buy online? | 05:33 |
| mithro | nueces: Where are you located? I'm sure we can get you one for under $US20.... | 05:34 |
| xfxf | theyre about $15 from monoprice | 05:34 |
| xfxf | buying them locally from oz, they were like $70 plus | 05:34 |
| xfxf | make sure you are looking at the short cables too. most redmere cables retailers stock are super long, which is why they are expensive | 05:35 |
| nueces | I'm in Argentina. I thinking on buy online a few, but I have to pay the shipment plus %50 on taxes | 05:35 |
| mithro | nueces: We can probably get CarlFK to ship you a couple in an unmarked package | 05:35 |
| nueces | I test two cameras the last week, one Sony HDR-HC9E 1080i, but I could not change the format resolution to 720p@30fps. | 05:37 |
| nueces | and one Nikon D7000 and get WER problems | 05:38 |
| *** Bertl_oO is now known as Bertl_zZ | 05:38 | |
| nueces | mithro, I could pay the cables and shipment, but I'm waiting now for a friend that is going to do a work travel | 05:40 |
| mithro | nueces: BTW Have you looked at voctomix yet? | 05:41 |
| nueces | not really, just a quick look on it | 05:41 |
| CarlFK | nueces: https://github.com/CarlFK/voctomix-outcasts#getting-started--demo | 05:44 |
| nueces | any ideas if this https://www.amazon.com/dp/B014IZ276C/ref=wl_it_dp_o_pC_nS_ttl?_encoding=UTF8&colid=29U6I6T0H2FW4&coliid=I19RINIQNCPD4V could need two redmere cables? one for both ends? | 05:44 |
| tpb | Title: GitHub - CarlFK/voctomix-outcasts (at github.com) | 05:44 |
| mithro | nueces: I wouldn't recommend trying to use something like that with the Opsis | 05:44 |
| nueces | CarlFK, http://dpaste.com/2VJV1M9 http://img.ctrlv.in/img/16/08/24/57bd3492932a4.png | 05:50 |
| tpb | Title: dpaste: 2VJV1M9 (at dpaste.com) | 05:50 |
| nueces | mithro, ok, thanks | 05:50 |
| CarlFK | nueces: guessing it came up, but just black, then you hit ^c to halt it? | 05:53 |
| nueces | yes | 05:53 |
| nueces | but I wait a few seconds before hit ^c | 05:54 |
| nueces | :) | 05:54 |
| CarlFK | nueces: voctomix-outcasts/configs/light.ini | 05:54 |
| CarlFK | [videodisplay] - try the other 2 system= gl and x | 05:55 |
| *** sb0 has quit IRC | 05:59 | |
| mithro | xfxf: What happened at the start of https://www.youtube.com/watch?v=C19fFU-TVWU video? | 06:06 |
| xfxf | I can't look now, can you summarise? | 06:09 |
| xfxf | is that the one with clipping audio for the first 5 mins? | 06:09 |
| mithro | xfxf: I think there capture from the laptop was not working or something? | 06:09 |
| xfxf | i vaguely recall that one, I don't remember | 06:11 |
| xfxf | camera gets zoomed out to show screen? | 06:11 |
| mithro | xfxf: yeah - then I see you run down the front :P | 06:11 |
| xfxf | oh, right, somebody unplugged the laptop at the front | 06:11 |
| xfxf | ran out of battery | 06:11 |
| CarlFK | doh | 06:11 |
| xfxf | this is why we need the monitoring thing | 06:12 |
| mithro | xfxf: Yes | 06:12 |
| xfxf | there's one with horiffic clipping audio too until I readjusted it, MCEC adjusted house levels and the volunteer for some reasons didn't hear the clipping despite wearing headphones | 06:13 |
| xfxf | this is the known trade off training up volunteers though, just try to keep snafu's to a minimum | 06:13 |
| xfxf | that's only solved completely through more automation | 06:13 |
| xfxf | it horrifies me but go and look at a proper pro company doing this, their ratio of technical snafu's is higher | 06:14 |
| xfxf | surprisingly | 06:14 |
| xfxf | i think our ability to centrally manage + automate the solution to some degree beyond the way most commercial solutions do this negates the downside of not having trained guys | 06:16 |
| xfxf | plus the care factor of AV pro's is usually low for shows like this | 06:16 |
| xfxf | which i can understand tbh - if you were doing this every day... | 06:16 |
| xfxf | we like it because the content of the conferences is genuinely interested to us | 06:17 |
| xfxf | it'd be like us going and recording a sales conference | 06:17 |
| xfxf | i'd still hope i'd have a higher care factor though, but regardless | 06:17 |
| xfxf | s/interested/interesting/ | 06:17 |
| mithro | CarlFK: I have a crazy idea which is right up your alley.... | 06:21 |
| CarlFK | mithro: oh no... ;) | 06:22 |
| mithro | CarlFK: You should OCR the recorded video looking for things which look like URLs and then convert them to YouTube annotations so I can click on URLs in people's slides :P | 06:22 |
| nueces | lol | 06:23 |
| CarlFK | mithro: I have the OCR bit.. how about you give me some annotation code? | 06:23 |
| mithro | CarlFK: If you can generate the time period and the x/y location I can give you the annotation creation code bit | 06:24 |
| CarlFK | neat. oh wait.. x/y.. that's ... not something I currently do. but maybe I can... | 06:25 |
| mithro | CarlFK: Oh and the URL value :P | 06:25 |
| CarlFK | https://github.com/CarlFK/veyepar/blob/master/dj/scripts/ocrdv.py # ocrdv - reads frames from a .dv untill it finds a bunch of words | 06:26 |
| tpb | Title: veyepar/ocrdv.py at master · CarlFK/veyepar · GitHub (at github.com) | 06:26 |
| xfxf | DV? who uses DV? :P | 06:26 |
| CarlFK | https://github.com/CarlFK/veyepar/blob/master/dj/scripts/gsocr.py # uses gstreamer to send frames to gocr better? ;) | 06:27 |
| tpb | Title: veyepar/gsocr.py at master · CarlFK/veyepar · GitHub (at github.com) | 06:27 |
| CarlFK | mithro: can I get a job at youtube to do this? | 06:28 |
| CarlFK | should only take me.. um.. 6 months sounds good. | 06:28 |
| mithro | CarlFK: No :P | 06:28 |
| CarlFK | boo. | 06:29 |
| mithro | CarlFK: What is gocr and where does it come from? | 06:29 |
| CarlFK | mithro: http://jocr.sourceforge.net/ I hear tesseract is better. I think I have code to use that too | 06:30 |
| tpb | Title: GOCR (at jocr.sourceforge.net) | 06:30 |
| *** danielki has joined #timvideos | 06:31 | |
| mithro | xfxf: I think we could use ssk1328's code to send a "brightened" version of the slides to the projector and hence use a lower value during capture | 06:35 |
| mithro | ssk1328: Where did we get with the gamma correction bits? | 06:35 |
| mithro | CarlFK: btw are you still getting https://github.com/timvideos/HDMI2USB-mode-switch/issues/38 ? | 06:35 |
| tpb | Title: bytes swapped · Issue #38 · timvideos/HDMI2USB-mode-switch · GitHub (at github.com) | 06:35 |
| *** rohitksingh_wor1 has joined #timvideos | 06:36 | |
| CarlFK | mithro: nope. all fixed. want me to close it? | 06:36 |
| mithro | CarlFK: yes please | 06:36 |
| mithro | CarlFK: I saw a bunch of arm related fixes go into openocd in the last couple of weeks | 06:37 |
| CarlFK | mithro: openocd has been working for me ... I forget where I got it now | 06:37 |
| *** rohitksingh_work has quit IRC | 06:39 | |
| mithro | Damn, I put these PyCon talks on to be background stuff - but I keep getting actually distracted by them :P | 06:39 |
| *** evilRails has joined #timvideos | 06:44 | |
| *** ssk1328 has quit IRC | 07:06 | |
| nueces | I like this player, you could see the slides in any moment, is a nice thing to have :) https://gstconf.ubicast.tv/videos/gstreamer-webkit-html5-and-real-time-communication/ | 07:12 |
| tpb | Title: Gstreamer, Webkit, HTML5 and Real-Time Communication - GStreamer conferences (at gstconf.ubicast.tv) | 07:12 |
| mithro | nueces: I like the side-by-side modes but xfxf hates them | 07:12 |
| mithro | nueces: This is what the CCC does -> https://media.ccc.de/v/emf2016-198-my-ubertooth-year | 07:13 |
| tpb | Title: C3TV - My Ubertooth Year (at media.ccc.de) | 07:13 |
| xfxf | mithro: if we did 1080p i'd certainly consider the side by side | 07:18 |
| xfxf | it makes some presentations unreadable at 720p | 07:18 |
| mithro | xfxf: We could do 1080p output with 720p input | 07:19 |
| mithro | I like the side-by-side mode as it removes people needing to "do" something :P | 07:24 |
| nueces | mithro, there is a difference, in my link the player could show the slide even if they are not showing the slide in a side by side capture | 07:24 |
| xfxf | well, sort of, there's actually a benefit in giving the team something to do during the presentation | 07:24 |
| xfxf | i'd actually put money that we'd end up with worse audio or camera placement if we made it mostly hands off | 07:25 |
| xfxf | there's little incentive to actually look at the recording | 07:25 |
| mithro | well - we replace the camera placement with speaker tracking :P | 07:25 |
| xfxf | ack, if we were going to do that, that's exactly the solution | 07:25 |
| xfxf | and have some audio analysis going on w/ central reporting | 07:26 |
| evilRails | xfxf: i honestly barely touched the camera, i just set-and-forget 2 mins in | 07:28 |
| mithro | Did you guys see this -> http://www.jaycar.com.au/pan-and-tilt-action-camera-bracket-mount-for-9g-servos/p/XC4618 | 07:28 |
| tpb | Title: Pan and Tilt Action Camera Bracket Mount for 9G Servos | Jaycar Electronics (at www.jaycar.com.au) | 07:28 |
| *** paradisaeidae has quit IRC | 07:37 | |
| *** ssk1328 has joined #timvideos | 07:55 | |
| mithro | I like how the gstreamer one has little green bars where the slide changes | 08:26 |
| mithro | ssk1328: Where did we get with the gamma correction bits? | 08:32 |
| mithro | CarlFK: you might like https://www.youtube.com/watch?v=qjTc5q7MsMg :-P | 08:35 |
| ssk1328 | mithro: I am not doing anything on it right now, I did a bit of research earlier | 09:06 |
| ssk1328 | mithro: gamma = 2.2 will be difficult to implement, But we can easily do gamma= 2, a lot work has already been done in floating point point square root | 09:07 |
| *** hozer has joined #timvideos | 11:31 | |
| *** Bertl_zZ is now known as Bertl | 11:54 | |
| *** ssk1328 has quit IRC | 12:46 | |
| *** danielki has quit IRC | 13:00 | |
| *** rohitksingh_wor1 has quit IRC | 13:31 | |
| *** Bertl is now known as Bertl_oO | 15:02 | |
| *** evilRails has quit IRC | 15:14 | |
| *** rohitksingh has joined #timvideos | 15:29 | |
| *** nueces has quit IRC | 15:44 | |
| *** rohitksingh has quit IRC | 16:50 | |
| *** nueces has joined #timvideos | 17:57 | |
| *** nueces has quit IRC | 19:15 | |
| *** springermac has quit IRC | 19:28 | |
| *** springermac has joined #timvideos | 19:40 | |
| *** Bertl_oO is now known as Bertl_zZ | 20:27 | |
| *** CarlFK has quit IRC | 23:55 | |
Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!
