23:00 | surami | it uses altera chip
| |
23:00 | surami | i don't know it could be reprogrammed to do raw capture
| |
23:00 | surami | maybe it's possible
| |
23:02 | Bertl | what's the price and is some schematic available?
| |
23:03 | surami | 459 euros + tax
| |
23:03 | Bertl | that's for two HDMI inputs, yes?
| |
23:04 | Bertl | or is that the version with 4 inputs?
| |
23:04 | surami | yes
| |
23:04 | surami | it's can be 4 too
| |
23:05 | surami | but i don't know any schematics about it
| |
23:08 | Bertl | well, it might work as is, if it can do raw recording
| |
23:08 | Bertl | but to reprogram it, one would need a schematic
| |
23:11 | surami | yes, and linux is not supported
| |
23:11 | Bertl | so not the best choice
| |
23:13 | surami | is there somebody, who is on this, i mean developing custom capture card?
| |
23:14 | Bertl | it isn't a project yet, but we are considering this option
| |
23:16 | surami | i see
| |
23:24 | surami | the before mentioned cards driver is directshow
| |
23:24 | surami | i mean with ffmpeg maybe the raw data could be captured without any loss
| |
23:25 | surami | or?
| |
23:25 | Bertl | directshow sounds like windows to me
| |
23:25 | surami | https://trac.ffmpeg.org/wiki/DirectShow
| |
23:25 | Bertl | I have no idea about windows solutions/options
| |
23:25 | surami | yes it is
| |
23:26 | surami | linux would be much more better.... eh
| |
23:27 | surami | this isn't a good solution
| |
23:28 | surami | that custom capture card should be developed
| |
23:30 | surami | good night Bertl!
| |
23:31 | surami | i go
| |
23:31 | Bertl | have a good night!
| |
23:33 | surami | left the channel | |
00:36 | fsteinel_ | joined the channel | |
00:39 | fsteinel | left the channel | |
00:50 | fsteinel_ | changed nick to: fsteinel
| |
03:23 | aombk2 | joined the channel | |
03:25 | aombk | left the channel | |
03:41 | Bertl | off to bed now ... have a good one everyone!
| |
03:41 | Bertl | changed nick to: Bertl_zZ
| |
06:30 | mooseboobs | left the channel | |
07:26 | intracube_afk | left the channel | |
07:28 | intracube | joined the channel | |
09:12 | surami | joined the channel | |
09:47 | surami | hi
| |
09:57 | g3gg0 | joined the channel | |
10:03 | surami | i found something "new" about that that capture card, it has linux driver too :)
| |
10:09 | surami | so the new dataflow theory: beta 3xhdmi out -> capturecard (pcie x4) 3xhdmi -> pcie x4 to pcie x1 converter on the jetson tk1 sbc (runing ubuntu) -> ffmpeg raw capture -> sata3 ssd on the jetson tk1 sbc
| |
10:09 | surami | something like this
| |
10:11 | surami | but the MLV format would be the best from ML
| |
10:12 | surami | there are several good options for MLV -> CDNG
| |
10:24 | rhavan | left the channel | |
10:48 | ItsMeLenny | joined the channel | |
10:48 | davidak | joined the channel | |
11:59 | Bertl_zZ | changed nick to: Bertl
| |
11:59 | Bertl | morning folks!
| |
12:00 | mars_ | morning
| |
12:00 | Bertl | surami: sounds doable, double check that the card allows raw recording though, otherwise it might become problematic
| |
12:02 | se6astian|away | changed nick to: se6astian
| |
12:03 | Bertl | morning se6astian!
| |
12:03 | danieel | what level of raw? 0: not raw, everything changed to yuv, 1: raw, active video only data, 2: rawraw: full frame with data in blanking (ehm, in hdmi rather data from data islands), also there are other constraints: A) support of only CEA resolutions, B) support of ANY resolution
| |
12:04 | Bertl | yep, thanks, that would be the interesting part
| |
12:04 | se6astian | good morning :)
| |
12:05 | danieel | one entry for your table: Deltacast SDI cards: 2A = rawraw but locks only to compliant resolutions/timings
| |
12:07 | danieel | blackmagic SDI = 1A, supplies only active video + parsed audio
| |
12:08 | Bertl | nice, do you know if there is a card which supports deep color?
| |
12:09 | danieel | there are 12 bit modes in SDI, the situation with deltacast is that it needs to run in RAW capture and you can parse it yourself
| |
12:09 | danieel | i am not into hdmi.. so probably easier to build one than find one
| |
12:11 | Bertl | yeah, HDMI seems to be problematic regarding deep color, although some devices claim to support it (at least in future versions :)
| |
12:12 | danieel | receivers generally support it
| |
12:13 | danieel | also, with my experience with hdmi chips - they do change the bitstream, so you probably never get RAW out of it
| |
12:14 | Bertl | you mean, raw raw, yes?
| |
12:14 | danieel | (look at ug180 revB for ADV7611, p112-113)
| |
12:15 | danieel | there are so many multiplications and offsets so that it will be a luck to configure it in a way that every single input maps to every single ouptut code
| |
12:16 | danieel | by defalut setup it does ugly things, i were developing display interfaces with that chip
| |
12:16 | Bertl | I see what you mean
| |
12:16 | danieel | in SDI it is better defined, and probably the cards respect bit to bit copies
| |
12:17 | danieel | probably the DVI receivers are more dumb too... so that might be the way
| |
12:17 | Bertl | good point
| |
12:21 | danieel | you might want to add crc over each line to hdmi, to make it sort of SDIish
| |
12:30 | Bertl | SDI does per line CRC?
| |
12:32 | danieel | yes, and has also a incremented line number :)
| |
12:32 | surami | here are the capture card what i found till now: http://www.magewell.com/hardware/sdi-cards/xi400de-sdi/xi400de-sdi_specifications.html?lang=en
| |
12:32 | danieel | read the smtpe standard...
| |
12:32 | surami | hdmi: http://www.magewell.com/hardware/hdmi-cards/xi400de-hdmi/xi400de-hdmi_specifications.html?lang=en
| |
12:33 | danieel | as the pcie bw is limited, they have made the scaler on the card to overcome that limit?
| |
12:41 | surami | i don't understand what you ask
| |
12:41 | surami | Bertl?
| |
12:42 | Bertl | he would like to know details you probably don't have about the card :)
| |
12:42 | danieel | seems to be a Gen1 card anyway
| |
12:44 | surami | Bertl: i just asked, what do you think?
| |
12:44 | surami | :)
| |
12:45 | Bertl | well, what is the price for the 4 port sdi/hdmi card?
| |
12:46 | Bertl | (we would need at least 3 ports)
| |
12:46 | lab-bot | sebastian closed T169: lsm.apertus.org down as "Resolved". http://lab.apertus.org/T169
| |
13:00 | surami | Bertl: it depends on supplier and card type, what i saw: from 459 euros to 648 euros + tax
| |
13:00 | Bertl | okay, that probably is in the same range as the BM cards
| |
13:29 | intracube | left the channel | |
13:34 | intracube | joined the channel | |
13:48 | intracube | left the channel | |
13:53 | intracube | joined the channel | |
14:14 | aombk2 | changed nick to: aombk
| |
14:17 | mooseboobs | joined the channel | |
14:43 | ItsMeLenny | left the channel | |
16:31 | danieel | Bertl: do not forget to support CEC or HEC for controlling the camera from the recorder
| |
16:46 | bcallebaut | joined the channel | |
16:51 | Bertl | good idea
| |
16:52 | danieel | power might be nice to have close by too
| |
17:04 | philippej|away | changed nick to: philippej
| |
17:39 | bcallebaut | left the channel | |
17:45 | bcallebaut | joined the channel | |
17:46 | troy_s | SDI is vastly more robust in field.
| |
17:48 | philippej | changed nick to: philippej|away
| |
17:56 | bcallebaut | left the channel | |
17:59 | Bertl | troy_s: how so?
| |
18:00 | mooseboobs | left the channel | |
18:01 | troy_s | A) Full size BNC connectors with lock vs Junk
| |
18:01 | troy_s | B) Full gauge cable vs Junk
| |
18:01 | troy_s | Connectors are the big one.
| |
18:01 | Bertl | nothing of that is specific to SDI :)
| |
18:01 | troy_s | Just saying..
| |
18:01 | troy_s | Good luck finding HDMI crap on an in field camera
| |
18:02 | troy_s | It is technological and usage based
| |
18:02 | troy_s | But you are better to listen to the screams of first assistants
| |
18:02 | Bertl | okay, so the argument is: because SDI commonly uses BNC and you normally don't find HDMI on in field cameras, it must be "better" yes?
| |
18:02 | troy_s | Ecosystem Bertl
| |
18:03 | troy_s | It isn't all this mythical "could" "if" "could"
| |
18:03 | troy_s | Which plagues these discussions
| |
18:03 | troy_s | It is "can" "does" "is" "proven over time"
| |
18:03 | Bertl | I wish you would simply stop generalizing
| |
18:03 | troy_s | I am doing my best
| |
18:03 | troy_s | Being deadly specific
| |
18:04 | Bertl | it would be much more productive if you said (for example):
| |
18:04 | troy_s | But the ecosystem doesn't care about HDMI by and large
| |
18:04 | Bertl | SDI has proven to be a good choice for in field because ...
| |
18:04 | troy_s | So the very real generalization is that it simply isn't supported
| |
18:04 | troy_s | This has been covered dozens of times in here
| |
18:04 | troy_s | Rexbron brought up at least five use cases for SDI over HDMI
| |
18:05 | troy_s | With practical real-world experience
| |
18:05 | Bertl | and if I remeber correctly, the protocol was not even remotely involved in those cases
| |
18:05 | troy_s | So yes. SDI is more robust in terms of cabling (according to just about every in field tech I have ever spoken with), transmission distance,
| |
18:06 | troy_s | And as silly as it seems, the actual connection protocols
| |
18:06 | troy_s | (And flange supports etc.)
| |
18:06 | troy_s | So when someone says "SDI" I doubt they are purely speaking of the technical protocol merits
| |
18:07 | troy_s | (Same for HDMI)
| |
18:07 | troy_s | But rather likely the larger ecosystem
| |
18:08 | Bertl | so, speaking from an eco system PoV
| |
18:08 | Bertl | SDI is currently defined up to 3Gbit
| |
18:08 | Bertl | how is that handled on 4k raw cameras/recorder?
| |
18:08 | Bertl | do they have 4 SDI cables connected?
| |
18:09 | troy_s | Bertl: Alexa has a built in codex system currently
| |
18:09 | troy_s | Codex was the glue for many years for ArriRaw
| |
18:10 | troy_s | And R3D uses their proprietary cards as does Sony
| |
18:10 | troy_s | All of those specifications are relatively easy to chase down online.
| |
18:11 | Bertl | so the question remains, how do those systems transfer raw 4k over SDI?
| |
18:11 | Bertl | is it simply a pack of 4 BNC coax cables?
| |
18:12 | troy_s | Not certain. Blackmagic rolls Decklinks
| |
18:12 | troy_s | Have a peek at the dox for the F65 and Alexa
| |
18:13 | troy_s | Dual link 12g SDI?
| |
18:13 | Bertl | not even remotely defined yet
| |
18:13 | Bertl | so I doubt there is an eco system around it
| |
18:13 | troy_s | 3g, 6g, and 12g SDI seems to point to your speculation.
| |
18:13 | troy_s | There is
| |
18:14 | Bertl | but I'm interested in the beforementioned "in field" usage and the "ecosystem" you were talking about
| |
18:14 | troy_s | Again, the ecosystem tends to evolve around the company that delivers it relatively robustly first
| |
18:15 | Bertl | basically you said so far that every camera manufacturer cooks their own soup and provides their own solutions
| |
18:15 | troy_s | A) you only find SDI on cameras
| |
18:15 | troy_s | B) raw formats generally record to the "mag" in camera these days
| |
18:16 | troy_s | Older days had a larger codex toaster with large umbilical cord
| |
18:16 | Bertl | so we can speculate that SDI is mostly used to connect view finders/monitors and recording is solved in a proprietary way
| |
18:17 | troy_s | Yes, with SDI being more and more viable via Blackmagic-like devices I believe
| |
18:17 | troy_s | Decklink has a whole page too.
| |
18:19 | troy_s | (And I am not certain, but the Alexa M has an umbilical cord, possibly proprietary?)
| |
18:24 | danieel | Bertl: 4k is done over 4x SDI links (might be 4x1.5G for 4k/30 and 4x 3G for 4k/60), the big overhead is caused by compatibility and need to transfer YCbCr, not raw
| |
18:25 | danieel | for raw, most 4k cameras are fine with 2x 3G SDI, (canon C500, arri alexa)
| |
18:26 | danieel | mine does it with 2x1.5G even
| |
18:27 | danieel | BM makes 6G, for 4k yuv 30p (=4x 1.5G payload), now jumping to 12G for 60fps yuv
| |
18:39 | Bertl | i.c.
| |
18:45 | surami | are the firmmware of the BM cards upgradeable?
| |
18:46 | Bertl | most likely, nobody nowadays gives away that option
| |
18:47 | surami | :)
| |
18:50 | danieel | there is an usb port with a mcu for that
| |
18:50 | danieel | funny enough, for dual cards they have that circuit twice there too :)
| |
18:51 | surami | :)
| |
19:07 | Bertl | off for a nap ... bbl
| |
19:07 | Bertl | changed nick to: Bertl_zZ
| |
19:10 | troy_s | danieel: Do you use two actual connectors to deliver the signal?
| |
19:10 | troy_s | So a wound pair of cables with two standard BNC connectors at the feed end?
| |
19:11 | danieel | yes, i use a pair of cables
| |
19:12 | danieel | (in protocol/sw i have cable id, so it can be connected in any order)
| |
19:16 | troy_s | danieel: Does yours roll to a DeckLink?
| |
19:17 | danieel | nope, it cant capture raw data
| |
19:17 | danieel | same situation as with canon raw / arriraw
| |
19:30 | troy_s | Urf. Is the YCbCr transform non-float?
| |
19:30 | troy_s | Because you could get 1:1 in /out if it were.
| |
19:31 | danieel | ??
| |
19:35 | troy_s | Apparently there is an RGB mode
| |
19:35 | danieel | raw in all the mentioned cameras is bayer
| |
19:43 | troy_s | Should be plausible though if it transmitted non-mangled RGB
| |
19:44 | troy_s | (16 or 32 bit greyscale linear would be ideal of course)
| |
19:49 | danieel | it is usually transmitted log (canon, arri), i did linear as it was simpler and keeps all codes out of sensor
| |
19:50 | danieel | it is hard to get 16 bits out of the sensor, so it makes not much sense
| |
19:50 | troy_s | Sorry
| |
19:50 | troy_s | I meant that it must be protocoled at linear
| |
19:50 | troy_s | Could be log or whatever you want
| |
19:50 | troy_s | But the protocol can't be futzing with the bits.
| |
19:51 | danieel | it is not.. so what are your worries?
| |
19:51 | troy_s | Huh?
| |
19:53 | danieel | canon/arri push the 14/16 bits into 10/12bit by log mapping.. (probably they use the limited 10 bits), i transfer bit perfect copy of 12 bit data
| |
19:54 | danieel | the protocol does not change the bits.. but the stream is unwatchable on usual equipment
| |
20:58 | se6astian | changed nick to: se6astian|away
| |
21:20 | davidak | left the channel | |
21:21 | davidak | joined the channel | |
21:33 | g3gg0 | left the channel | |
21:39 | surami | good night!
| |
21:39 | surami | left the channel | |
22:00 | Bertl_zZ | changed nick to: Bertl
| |
22:00 | Bertl | back now ...
| |
22:58 | davidak | left the channel |