Current Server Time: 21:49 (Central Europe)

#apertus IRC Channel Logs

2014/11/20

Timezone: UTC


23:01
sb0
left the channel
23:02
comradekingu
joined the channel
23:02
sb0
joined the channel
23:28
sb0
left the channel
23:28
sb0
joined the channel
23:46
comradekingu
left the channel
23:47
sb0
left the channel
00:09
sb0
joined the channel
00:16
sb0
left the channel
00:16
sb0
joined the channel
00:19
sb0
left the channel
00:20
Bertl_zZ
changed nick to: Bertl
00:20
Bertl
back now ...
02:19
Bertl
off to bed now ... have a good one everyone!
02:19
Bertl
changed nick to: Bertl_zZ
02:47
mooseboobs
left the channel
05:12
troy_s
left the channel
05:12
troy_s_
joined the channel
05:24
lab-bot
kontrakatze closed T146: Forum restructuring proposal as "Wontfix". http://lab.apertus.org/T146
05:33
sb0
joined the channel
05:39
ItsMeLenny
joined the channel
05:56
g3gg0
joined the channel
05:57
sb0
left the channel
07:00
renearts
joined the channel
07:04
renearts
left the channel
07:52
lab-bot
philippej reopened T146: Forum restructuring proposal as "Open". http://lab.apertus.org/T146
07:54
PhilippeJJ
joined the channel
08:00
intracube
joined the channel
08:17
PhilippeJJ
is testing shout (https://github.com/erming/shout)
08:29
ItsMeLenny
left the channel
08:32
ItsMeLenny
joined the channel
08:46
PhilippeJJ
left the channel
08:47
jucar
left the channel
08:56
philippej|away
changed nick to: philippej
09:02
FransKanters
joined the channel
09:03
FransKanters_
joined the channel
09:04
FransKanters_
Hi All, as dicussed yesterday with Herbert, we might be able to help with a PCIe recording device.
09:05
FransKanters_
Our company Incatec is currently developing a PCIe framegrabber for our 4k Open Hardware Camera Platform which might also be suitable for Apertus users, maybe with slight changes
09:07
philippej
Hello FransKanters_ , and welcome! It seems Herbert is curently away but will probably soon arrive
09:07
philippej
His nick is Bertl_zZ (the _zZ meaning he is away :-) )
09:08
FransKanters_
We have already hardware available (PCIe gen2x4, a Xilinx Kintex K160T FPGA and 4x 3G SDI hardware interface (upgradable to 4x6G with pin compatible cable drivers) but we are working on the FPGA code
09:08
FransKanters_
Hi Philippej. Herbert told me he might be asleep or away...:-)
09:09
FransKanters_
I'll wait, but in the meantime feel free to discuss your opinion on the recording solution for the apertus camera!
09:10
ItsMeLenny
left the channel
09:15
philippej
Sure, my tech knowledge is limited, but we can already clear the licensing issue. Is the K160T usable with free tools?
09:19
FransKanters
yes, the k160t is the largest fpga that can be used with the free webpack software
09:21
philippej
Great!
09:21
renearts
joined the channel
09:25
FransKanters_
we are willing to share the full schematics and the FPGA code for normal operation as open source to allow other people to develop their own projects
09:26
FransKanters_
The framegrabber will be able to record 4k@30p RAW over 1x3GSDI, 4k60p RAW over 2x3GSDI and even 4k120p RAW over 4x3GSDI
09:27
philippej
this is really interesting
09:27
philippej
as for physicall connectors, I'd personnally favor sdi's bnc
09:27
philippej
I guess we can implement wathever protocol we want on top of this, and are not stuck to sdi
09:28
FransKanters_
We also plan to develop and sell advanced IP cores as closed source addition to the platform (both for the camera and the framegrabber)> We have to make some money too :-)
09:29
philippej
the licence allows it afaict
09:29
renearts
defenitely true Phillipej!
09:29
FransKanters_
We currently use the Xilinx SDI IP core, but in fact it is just a 3Gigabit link (the MGT transceivers connect to cable drivers) so it is possible to run whatever protocol you would like.
09:29
renearts
I'm an Incatec employee btw, I'll follow the discussion as well as answer questions if possible/needed :0
09:30
FransKanters_
We are considering this ourselves to push the framerate up.
09:32
philippej
Herbert will have more interesting questions / ideas than me
09:34
FransKanters_
No problem, all ideas/questions are welcome! The Apertus camera is a great idea and we share your vision on open innovation
09:35
philippej
what is the highest bandwidth you could push on a single cable (if it's not sdi) ? At some point I guess cable lenght becomse an issue. Depending on usecase, a short cable could be used (when the camera and recorder are in the same enclosure for example)
09:37
FransKanters_
with the current cable drivers / equalizers the theoretical limit is 3 Gigabit per second per cable. With new cable drivers (pin compatible, so no additional development needed) this can be 6 Gigabit per second.
09:37
philippej
cable driver = fpga code?
09:38
FransKanters_
So in total 12 Gbit/s and upgraded 24 Gbit/s although the PCIe gen2x4 is only theoretically 20 Gb/s
09:39
FransKanters_
No the cable driver is an IC on the board which enables higher cable lengths. According to the standard up to 90 meters with good cable. We have only tested it with 20 meters so far...
09:39
philippej
for your use case, I guess you need long cables (10m +)
09:40
philippej
this could be also a good use case for cinema : put the camera head on a crane, and have the recorder on the ground
09:41
FransKanters_
yes, that's true. HDMI cables over 10 meter for high bandwidth are probably more expensive than coax and less flexible.
09:41
philippej
(note to self, crazy that you can put 20Gbit on a 90 m cable :-))
09:42
FransKanters_
hehe, the 20Gb is with 4 cables. However, you do have multiple coax cables within one cable for broadcast solutions, we use a triple coax version.
09:43
FransKanters_
for longer distances you can always use standard 3gSDI to optical fiber converters.
09:44
jucar
joined the channel
09:47
philippej
is there a market for this grabber card outside your and our use you think?
09:48
philippej
I read quite often stuff like "manufacturer xxx screwed us again, their grabber card doesn't recognize my camera signal"
09:48
FransKanters_
we aim for the high speed/high resolution market, which in itself is very broad.
09:48
FransKanters_
Since we are specialized in image analysis algorithms and FPGA implementations thereof, our primary business is developing and selling IP blocks
09:49
philippej
we could for example provide better support and better drivers, maybe crowdsource the development of drivers for various uses
09:49
FransKanters_
Our Open Hardware Camera Platform is a means to demonstrate our IP and to find applications
09:50
FransKanters_
We know that problem you describe, which has actually led us to develop our own.
09:50
FransKanters_
By making the platform open, it is possible to extend it to all kinds of camera's which makes it a more solid investement I guess.
09:51
Bertl_zZ
changed nick to: Bertl
09:51
Bertl
morning folks!
09:51
philippej
Hello Bertl, some backlog for you already :-)
09:52
Bertl
yeah, _zZ actually means asleep btw, _oO is otherwise occupied
09:53
FransKanters_
We are already thinking of crowdfunding for this. We also think about using compression to get the max out of the sensor (4k@400p and 2k@850p) with automatic image analysis for detecting trigger events in the image. This would probably be closed source however, since it is a large investement and close to our core business.
09:53
FransKanters_
Hi Bertl, good morning!
09:55
Bertl
okay, I've finished the backlog now :)
09:56
FransKanters_
haha...fire away!
09:57
Bertl
the K160T has 8 GTX tranceivers, so you plan to use 4 for PCIe and the other 4 for the SDI, yes?
09:57
FransKanters_
that's right.
09:57
Bertl
what IP do you use for the PCIe side?
09:58
renearts
*not planning actually, it's already implemented in hardware
09:58
Bertl
okay, so the hardware is already available?
09:58
renearts
currently we are setting up the framegrabber functionality using the Xilinx PCIExpress IP block
09:58
FransKanters_
yes...we are opening a webshop in december...
09:59
Bertl
can I get a look at it (schematic and layout)?
09:59
renearts
we still need to develop the FPGA code for the PCIExpress functionality though
09:59
FransKanters_
although we will not yet sell the framegrabber since the fpga code is not yet ready
10:00
FransKanters_
We will start selling our open hardware camera platform in December.
10:01
Bertl
what precisely is the "open hardware camera platform"?
10:01
philippej
May I ask if you already have a price range for the grabber?
10:02
FransKanters_
I can send you the schematics if you like. This is sufficient for most users to fully exploit the possibilities. We want to keep the layout to ourselves for now, since it was a big investement and I don't want Chinese copies too soon :)
10:02
Bertl
so the hardware is not really open then, just documented
10:03
Bertl
(which is fine)
10:03
FransKanters_
Price range for the framegrabber will probably be about 2500-3000 euro.
10:03
Bertl
and yes, please send me whatever you can send me without getting into trouble :)
10:03
FransKanters_
If volumes increase, the price can be lower of course
10:04
Bertl
a picture of the card would be nice too though
10:05
comradekingu
joined the channel
10:06
FransKanters_
i have sent the schematics, will send an image soon.
10:06
Bertl
excellent! thanks!
10:11
Bertl
okay, so basically you're considering opening up the platform (how far?)
10:11
Bertl
and you are currently working on the software (FPGA code) or looking for somebody to work on that?
10:15
FransKanters_
sorry, have been on the phone for a moment...
10:16
Bertl
no problem
10:17
FransKanters_
We plan to open up the schematic and the FPGA code for normal operation (sensor control, SDI, PCIe, SPI communication to raspPi, etc.) up to 120p
10:17
FransKanters_
On top of that we plan to build some additional closed form IP (e.g. speeds up to 400p using compression) and improved processing
10:18
FransKanters_
this closed code will also include some application specific code such as we use for our sports projects which we cannot open up
10:19
FransKanters_
However, with the open source code you will have a fully functional 4k120p system.
10:19
Bertl
'p' in 120p means FPS?
10:19
FransKanters_
yes.
10:20
FransKanters_
we almost have our camera running at 4k/120 fps
10:20
FransKanters_
RAW bayer pattern over 4x3G SDI
10:20
Bertl
okay, and that one will be available soon-ish too I guess?
10:21
Bertl
do you have a price estimate for the camera?
10:22
FransKanters_
yes it will. Actually the board used in the camera is almost the same as the board of the framegrabber. They only differ in assembly, so you now have the schematics for the camera too :-)
10:22
FransKanters_
It will probably be around 4500 euro
10:23
Bertl
haven't got the schematics yet, but it might have been delayed by greylisting (if you used a new/different account)
10:23
Bertl
okay, sounds reasonable
10:23
FransKanters_
including simple aluminium housing, raspberry PI with touch screen for control and power cables.
10:24
FransKanters_
ok, I sent it by email ...
10:27
Bertl
so let's get back to the frame grabber, what is your vision what we can do for eachother there?
10:31
FransKanters_
it depends a bit on what is needed for you. We could for example together make sure that our framegrabber correctly communicates with your camera. That way the apertus users can have access to recording equipment as soon as the camera comes out.
10:33
FransKanters_
If electrically a different design is desired by you or the apertus community we could help in developing such a design.
10:34
FransKanters_
I'm not sure however what you need. Do you need more manpower for your development? Could we make shorter time to market by bringing in our framegrabber and camera knowledge?
10:35
FransKanters_
we are quite open for different types of collaboration, depending on what your community needs
10:35
Bertl
sounds good to me. to answer the questions
10:36
Bertl
we can definitely benefit from some manpower for development
10:36
Bertl
using the card as is would require to add a 4 channel SDI interface shield to the Beta
10:37
Bertl
which is something we haven't planned in the first step, mostly because it requires a lot of space
10:38
FransKanters_
I understand.So you are planning to use HDMI right?
10:38
Bertl
HDMI or DP or for recording maybe a custom interface
10:39
FransKanters_
Do you plan to use MGT pairs for these?
10:39
Bertl
i.e. whatever connector we have but a custom protocol
10:40
renearts
Bertl: what about 4x3Gbit serial channels? :+)
10:40
Bertl
the GTX on your framegrabber card should be capable of 12Gbit per channel no?
10:41
renearts
so anyone can decide which protocol he/she wants to use, whether it be SDI or not
10:41
renearts
our speed grade limits us to 6.6GT
10:41
renearts
but with -3 12GT should be possible IIRC
10:41
Bertl
okay, so let's say 6Gbit then :)
10:42
FransKanters_
Our board has basically MGT's connected to a 3Gigabit cable driver into a HDBNC connector. We can do whatever protocol we want on that interface, as long as the physical connection is 3Gb or 6Gb and adheres to the physical SDI standard. The driver chips are small (5x5mm) and the HDBNC are also not very large.
10:42
Bertl
yes, agreed
10:43
FransKanters_
It is also possible to remove the cable drivers and connect directly, but cable lengths will be limited...
10:43
Bertl
so, if you want to help AXIOM, maybe a good idea would be to create an "SDI" shield which can connect to the framegrabber but is not limited to SDI as protocol?
10:43
renearts
Just checked the datasheet: with speed grade -3 and FFG package the GTXs go up to 12.5Gb/s
10:44
FransKanters_
sure, sounds good to me.
10:44
Bertl
of course, it would be great to open up the design for the shield
10:45
Bertl
(which I think shouldn't be a problem for you at all)
10:45
FransKanters_
that would be no problem of course
10:47
Bertl
I also think it shouldn't be a problem for you to get us one or two of the frame grabber cards for testing in the near future
10:48
Bertl
which would allow the AXIOM Beta community to have a generic as well as AXIOM Beta specific recording solution
10:48
FransKanters_
I think we could arrange that.
10:48
Bertl
in return, we can provide an AXIOM Beta (once we have something working) to test with as well
10:50
FransKanters_
sounds good!
10:51
Bertl
okay, so if you still want to help with current development, then definitely having a review of our sensor and interface designs would be much appreciated
10:51
FransKanters_
One drawback in our current design is that you need two slots since we have components on the back side of the board.You can see that on the pics I sent by e-mail.
10:51
FransKanters_
We still want to develop a custom frame with coolinf gor the FPGA that is GPU like dual slot.
10:52
Bertl
yeah I don't think that's a big problem nowadays
10:52
FransKanters_
we'll be on a lunch break now. Talk to you later!
10:53
Bertl
okay, great! some time to grab breakfast :)
11:07
surami
joined the channel
11:08
surami
left the channel
11:13
comradekingu
left the channel
11:55
FransKanters_
we're back...if you have further questions...
12:05
Bertl
well, no questions from my side at the moment ... everything sounds great so far!
12:06
Bertl
we should probably do some technical discussions on the SDI shield soon, but that doesn't need to happen right now
12:06
Bertl
what I would appreciate is feedback to the current sensor and interface design ASAP as we plan to order the PCBs soon
12:07
Bertl
http://vserver.13thfloor.at/Stuff/AXIOM/BETA/
12:08
Bertl
specifically the axiom_beta_sensor* and the axiom_beta_interface_dummy*
12:21
FransKanters_
didn't check every connection :) but it certainly looks good!
12:22
FransKanters_
Did you actually calculate the differential impedance on the LVDS channels?
12:23
Bertl
yes, we did, but a second opinion (check with expensive tools :) would be welcome
12:24
FransKanters_
:-) We don't have these tools. Our board is designed with a partner who has these tools...
12:25
Bertl
yeah, well, it was worth a try :)
12:25
FransKanters_
How is the stack for the dummy board? Top and bottom signal and ground in between? Or also power?
12:26
Bertl
ground and power in between
12:26
Bertl
(same as for the sensor board)
12:30
FransKanters_
Not sure, but if i remember correctly we have signal planes always next to a GND plane and not a power plane. Is the power plane split up into multiple voltages?
12:31
FransKanters_
i.e. does the differential signal on top or bottom cross a voltage change?
12:32
FransKanters_
probably not really a problem at 600 MHz, but for the high speed serial transceivers this should be avoided.
12:33
davidak2
MOX crowdfundinf was succesfull and reached all of it's streched goals
12:33
davidak2
S20postfix -> ../init.d/postfix*
12:33
davidak2
lrwxrwxrwx 1 root root 15 Apr 22 2014 S20rsync -> ../init.d/rsync*
12:33
davidak2
lrwxrwxrwx 1 root root 24 Apr 22 2014 S20screen-cleanup -> ../init.d/screen-cleanup*
12:33
davidak2
lrwxrwxrwx 1 root root 17 May 27 13:09 S20sysstat -> ../ini
12:33
davidak2
https://www.indiegogo.com/projects/mox-file-format
12:33
davidak2
lol, some crap pasted
12:34
davidak2
then we can plan to integrate it in the beta software
12:40
FransKanters_
looked it up for our design: for the sensor connections we also cross power planes, but not for the high speed SDI signals...
12:41
FransKanters_
So doesn't seem a problem here :-)
12:49
Bertl
davidak2: sounds good!
12:50
Bertl
FransKanters_: great!
13:00
mooseboobs
joined the channel
13:34
jucar
left the channel
13:44
jucar
joined the channel
13:49
philippej
changed nick to: philippej|away
14:01
comradekingu
joined the channel
14:40
philippej|away
changed nick to: philippej
15:08
Bertl
off for now ... bbs
15:08
Bertl
changed nick to: Bertl_oO
15:13
FransKanters_
left the channel
15:13
FransKanters
left the channel
15:15
renearts
left the channel
15:43
Daves
joined the channel
15:44
Daves
i'm a noob who missed out on the indiegogo campaign by a day! can anyone help me
15:44
Daves
I want an f-mount beta
15:44
Daves
will they ever be for sale openingly on this website?
15:47
Daves
left the channel
16:01
Bertl_oO
changed nick to: Bertl
16:01
Bertl
back now ...
16:03
aombk
live pseudorandom audio and video data stream : http://aombk.attinom.net/otr/betahive
16:07
lab-bot
Bertl created T174: Investigate Micron HMC Technology (Memory). http://lab.apertus.org/T174
16:09
lab-bot
Bertl created T175: Investigate Open Source USB 3.0 Stack. http://lab.apertus.org/T175
16:12
lab-bot
Bertl created T176: Investigate CoaXPress. http://lab.apertus.org/T176
16:18
lab-bot
Bertl created T177: Investigate (e)DisplayPort. http://lab.apertus.org/T177
16:31
niemand
joined the channel
16:33
lab-bot
Bertl created T178: Investigate FPGA Options (Low Power as well as Processing) and Licenses.. http://lab.apertus.org/T178
16:39
slikdigit
joined the channel
16:39
slikdigit
left the channel
16:39
slikdigit
joined the channel
16:43
niemand
left the channel
16:47
comradekingu
left the channel
16:56
sb0
joined the channel
16:57
davidak2
aombk: is it just a random playlist or somehow generated?
16:57
aombk
the audio is generated
16:58
aombk
by a software synth i am building the last couple of days
16:59
philippej
it sounds really good (and creepy)
16:59
aombk
haha thanks
16:59
davidak2
what software do you use for streaming?
16:59
aombk
wirecast
17:00
aombk
but Open Broadcaster Software is also good
17:00
comradekingu
joined the channel
17:03
niemand
joined the channel
17:03
sb0
left the channel
17:08
philippej
changed nick to: philippej|away
17:19
davidak2
OBS looks really nice. will try it if i had the need for video streaming
17:32
philippej|away
changed nick to: philippej
17:41
philippej
changed nick to: philippej|away
17:48
jucar
left the channel
17:48
se6astian|away
changed nick to: se6astian
17:53
se6astian
good evening
17:54
Bertl
evening se6astian!
17:57
davidak
joined the channel
18:03
sb0
joined the channel
18:29
jucar
joined the channel
18:42
comradekingu
left the channel
18:44
comradekingu
joined the channel
18:59
FransKanters
joined the channel
19:00
FransKanters
Hi Bertl, I noticed your task to investigate the Micron HMC memory.
19:00
FransKanters
That is cool technology to include :-)
19:01
Bertl
yes indeed, not sure how practicable it is atm though
19:02
Bertl
btw, regarding SDI, you are using the Xilinx IP for that, yes?
19:02
FransKanters
I know some of the companies involved in the coaxpress standard by the way. They do up to 4x6.25Gb/s giving 25 Gbit/s. Not sure if they use specific drivers.
19:02
Bertl
if you have some input to those tasks, please go ahead and comment
19:02
FransKanters
Yes, we currently use the IP core from xilinx, but it is in fact very well possible to do it yourself.
19:03
FransKanters
We probably will do it at some point to go to 6Gb/s
19:03
Bertl
yeah, it shouldn't be too complicated as far as I checked
19:03
FransKanters
It is just a wrapper that includes the right headers and CRC.
19:04
FransKanters
Of course it will take some time to develop but it shouldn't be too difficult.
19:04
Bertl
I see more a problem with compliance testing
19:04
Bertl
do you have resources to do that?
19:04
FransKanters
yes, of course we don't comply to anything...:-)
19:05
FransKanters
but hey, it works!
19:05
Bertl
yeah, not a big deal in a known environment, but for generic interfaces, it might become a problem
19:05
Bertl
we've seen that with HDMI out to the Atomos
19:05
FransKanters
yes, that's true. By using the Xilinx IP block it is easier to assure compliance.
19:06
FransKanters
We noticed that some companies use black magic in their products (and name) and do not adhere to the SDI standard...
19:07
FransKanters
We had a hard time interfacing with these products...
19:07
Bertl
*G*
19:08
FransKanters
So I guess it is the same everywhere...More reasons for open platforms...
19:09
FransKanters
i'm away for a moment...
19:26
danieel
FransKanters / Bertl : CxP is special chips, different than SDI, with inclusion of coupled power its quite hard to get it to work
19:31
comradekingu
left the channel
19:52
philippej|away
changed nick to: philippej
19:55
lab-bot
sebastian closed T165: Create TAG Festival article for apertus.org as "Resolved". http://lab.apertus.org/T165
20:12
niemand
left the channel
20:17
intracube
left the channel
20:30
lab-bot
sebastian closed T170: add sample providers to website as "Resolved". http://lab.apertus.org/T170
20:45
se6astian
time for bed
20:46
se6astian
long day at the university ahead
20:46
se6astian
good night
20:46
se6astian
changed nick to: se6astian|away
20:46
davidak
good night se6astian
20:46
Bertl
night se6astian!
20:46
FransKanters
left the channel
20:52
wescotte
joined the channel
20:52
wescotte
left the channel
20:59
lab-bot
davidak closed T130: Images not loaded as "Resolved". http://lab.apertus.org/T130
21:02
philippej
thanks davidak !
21:03
philippej
new unified faq here, now with a table of content
21:03
philippej
https://apertus.org/en/faq
21:04
lab-bot
philippej closed T154: move crowdfunding FAQ questions to AXIOM FAQ as "Resolved". http://lab.apertus.org/T154
21:44
philippej
good night everyone
21:47
Bertl
nighty night
21:50
philippej
changed nick to: philippej|away
21:53
designbybeck
left the channel
22:27
lab-bot
davidak created T180: Update Donation site. http://lab.apertus.org/T180
22:29
surami
joined the channel
22:30
surami
left the channel
22:39
g3gg0
left the channel