Current Server Time: 01:40 (Central Europe)

#apertus IRC Channel Logs

2014/03/08

Timezone: UTC


23:01
troy_s
eadmund: When did you speak with Mr. Knoll?
23:20
standesbeamter
joined the channel
23:20
standesbeamter
hoi
23:24
standesbeamter
left the channel
23:35
eadmund
Long long time ago - he had exactly the same points as you, so I guess you are in good company :)
23:36
eadmund
standesbeamter: willkommen, Geheimrat :)
23:48
eadmund
does anyone know why zedboard has gone out of stock, and microzed is available? Is a new zedboard coming?
23:53
norpole
joined the channel
23:58
norpole
left the channel
00:19
eadmund
left the channel
00:19
norpole
joined the channel
00:34
norpole
left the channel
01:16
gcolburn
left the channel
03:11
Bertl
off to bed now ... have a good one everyone!
06:15
marron
left the channel
07:06
sb0
left the channel
07:28
jucar
left the channel
07:42
jucar
joined the channel
07:56
Topic
apertus° - open source cinema | www.apertus.org | Currently Hot: Axiom Alpha Development http://axiom.apertus.org/ | IRC Logs available at: http://irc.apertus.org
07:56
se6astian
has set the topic
07:57
jucar
left the channel
09:09
Bertl_
morning folks!
09:09
Bertl_
changed nick to: Bertl
09:29
norpole
joined the channel
09:31
norpole
left the channel
09:34
se6astian
good morning
09:34
se6astian
some pictures from yesterday event: https://twitter.com/ApertusOSCinema
09:38
Bertl
nice
10:49
galvanic
joined the channel
11:52
danhanes
joined the channel
15:27
danhanes
left the channel
15:29
troy_s
Greetsies droogies
15:43
galvanic
left the channel
16:16
rainer_vie
joined the channel
16:18
rainer_vie
Hi Everybody!
16:18
jucar
joined the channel
16:28
troy_s
Greetings rainer_vie
17:17
sb0
joined the channel
17:26
Bertl
hey rainer_vie!
17:32
se6astian
have to leave
17:32
se6astian
bye
17:32
se6astian
left the channel
17:46
rainer_vie
oh too late cu
17:46
Bertl
yeah, he is quick when leaving :)
17:47
rainer_vie
:) probably he has something very important to do, as always
17:48
rainer_vie
Bertl = Herbert ?
17:48
Bertl
so what brings you here? let me guess, the metalab talk?
17:48
Bertl
yes, it's me
17:48
rainer_vie
fine... nice to meet you here
17:49
rainer_vie
yes the metalab talk, and I've met sebastian before...
17:50
rainer_vie
I want to take care for the part when the camera is generating data...
17:50
rainer_vie
backup/archiving/review
17:50
rainer_vie
transcode etc.
17:51
rainer_vie
the idea of the openCine software interests me
17:51
rainer_vie
I do a lot of that in my work...
17:51
Bertl
I think I remeber sebastian mentioning somebody working on a storage server or similar, is that you?
17:52
Bertl
*remember
17:52
rainer_vie
a mobile backup/hashing/storage/viewing solution is our idea...
17:53
rainer_vie
currently brainstorming on different versions
17:54
rainer_vie
that could be a small ARM based board to big file-vault
17:55
Bertl
I had a bunch of chats with various folks regarding requirements and existing solutions ... as I see it, it should be similar to a NAS, with higher data safety requirements
17:56
Bertl
doesn't necessarily need to be network attached, but I think you get what I mean
17:57
rainer_vie
yes, similar to a NAS. For offloading there should be all protocols implemented as well (SMB, NFS, iSCSI, etc)
17:57
Bertl
basically some kind of RAID setup (probably RAID 6), lots of space and software which ensures that nothing is deleted :)
17:57
rainer_vie
right... because of that we will hash all incoming files and putting into a database
17:58
rainer_vie
so in every step of copy you should be able to verify data integrity
17:59
rainer_vie
4k RAW uncompressed will generate a lot of data, this should be able to be playbacked also in realtime
17:59
Bertl
I was wondering if it wouldn't make a lot of sense to archive all the data on DLT for cold storage
17:59
rainer_vie
I thinking of LTO-6 and LTFS
17:59
rainer_vie
or LTO-5
18:00
Bertl
yes, LTO-5 has a sweet spot from the price/capacity PoV
18:01
rainer_vie
yeah I'm using it quite a lot... Data throughput is ok... but that should be also verified...
18:01
rainer_vie
I'm thinking of modules which will be then inside of openCine
18:01
rainer_vie
A software which can view the RAW files
18:02
rainer_vie
copy/verify it
18:02
rainer_vie
and archive it
18:02
rainer_vie
all that in a chassis with storage and LTO-Drive
18:02
rainer_vie
as well HD-SDI I7O
18:03
rainer_vie
i/O
18:03
Bertl
do you have any HD-SDI solutions you can recommend?
18:03
rainer_vie
Blackmagic is ok and has linux drivers as well as a SDK
18:04
rainer_vie
using their cards for a long time now
18:04
rainer_vie
they have a 4K quad 3G SDI card for under 1k EUR
18:05
rainer_vie
http://www.blackmagicdesign.com/products/decklink
18:05
rainer_vie
Linux Driver: http://www.blackmagicdesign.com/support/detail?sid=3945&pid=31786&leg=false&os=linux
18:06
rainer_vie
SDKs: http://www.blackmagicdesign.com/support/sdks
18:08
Bertl
okay, thanks, noted.
18:08
rainer_vie
a great project at all the camera....
18:09
rainer_vie
saw today that Xilinx has a quad HD-SDI board for their FPGA developing boards
18:10
rainer_vie
to get 12 Gbit/s picture data is really woodoo :)
18:10
Bertl
hehe
18:13
rainer_vie
The scaling from 4k to 1080p is done in the FPGA?
18:13
rainer_vie
for HDMI output?
18:14
Bertl
currently the scalining happens implicitely
18:14
rainer_vie
that means?
18:14
Bertl
i.e. we acquire the 4k image 4096x3072, reorder the data so that we have bayer blocks (RG/GB)
18:15
Bertl
then we transmit each bayer block to the HDMI encoder as RG,GB value to form a 2:4:2 RGB stream
18:15
Bertl
this reduces the image by 1/2 in each direction, i.e. to 2048x1536
18:15
Bertl
and we just cut out a 1080p region for now
18:16
Bertl
but yes, this happens in the FPGA
18:17
rainer_vie
ah ok... great... the option to get a scaled down picture from the hole sensor will be a nice feature as well... not everybody want to work with 4k data
18:18
rainer_vie
or a side by side recording function... 4k raw and a 1080p proxy for editing
18:19
Bertl
yes, I think scaling the image is not a big problem in the FPGA, of course, intregral changes will give better results than interpolation
18:19
rainer_vie
the debayer for the hdmi is done in the hdmi block?
18:20
Bertl
no debayer required in our case, as we send the data to the encoder directly, but typically it would happen after reading the buffer memory
18:21
Bertl
I don't think that it is a good idea to output both though (4k and FullHD) because of the huge bandwidth requirements
18:22
Bertl
i.e. it would be better to do the 4k->FullHD (probably encoded lossy) on the copy/archiving system
18:22
rainer_vie
yeah, probably a child module could do this later
18:22
rainer_vie
right... we will do this :)
18:23
rainer_vie
every flavor of codec and resolution will come out of this baby
18:24
rainer_vie
as said... you did a great work on the sensor frontend
18:34
Bertl
thank you very much!
18:36
reproduction
joined the channel
18:37
rainer_vie
:) the sensor captures the pictures so it's the most important part
18:37
Bertl
welcome reproduction!
18:37
rainer_vie
\o/ reproduction
18:42
rainer_vie
afk
19:16
reproduction
left the channel
19:31
rainer_vie
going off... cu @all
19:32
rainer_vie
left the channel
22:05
intracube
joined the channel
22:42
jucar
left the channel