Current Server Time: 04:43 (Central Europe)

#apertus IRC Channel Logs

2014/09/29

Timezone: UTC


00:09
aombk
nice
00:21
intracube
aombk: thanks
00:22
intracube
one thing I've noticed is the FPN-subtract01.tif has noticable vignetting when you increase gain/contrast
00:23
intracube
I wonder if this is down to slight uneven illumination of the sensor when calibrating?
00:23
Bertl
probably, and the optics as well I guess
00:23
intracube
I'd sort of expect the fixed noise to run full width/height without any gradient
00:23
intracube
Bertl: there was a lens attached during calibration?!
00:24
Bertl
I'm pretty sure, otherwise even illumination is even more trickier
00:24
intracube
I'm experimenting with pre-processing the FPN tif to reduce the vignetting
00:24
intracube
and see if that gives slightly better results
00:25
Bertl
we built a contraption to illuminate the sensor without lens system, but that was rather complicated to use
00:25
intracube
with a contrast boost: http://www.pasteall.org/pic/show.php?id=77841
00:25
intracube
Bertl: ah I see
00:25
Bertl
also, depending on when the footage was acquired, the section doesn't need to be in the center of the sensor
00:26
Bertl
more likely it is at the top of the visual area
00:26
Bertl
3072/2 = 1536, but the output was 1080 (full HD)
00:27
Bertl
so it might be up to 768 rows off center
00:27
Bertl
(which is a lot :)
00:27
intracube
eep
00:31
aombk
global shutter footage is very pleasing
00:45
intracube
^ + 1
00:47
intracube
I read on here there were at least two different ways the data was read of the sensor
00:47
intracube
something to do with the capability of the recording device or HDMI interface?
00:48
Bertl
yes, we first had an odd 4:2:2 RGB encoding
00:48
Bertl
and later switched to a more sane YCbCr encoding
00:48
intracube
do you know if the FPN tiff was done using a different encoding to the test footage?
00:49
Bertl
I presume neither nor, i.e. I think that is most likely a snapshot
00:49
intracube
there is a (green) hot pixel that's in a slightly different position in the tiff compared to the test footage
00:49
Bertl
i.e. taken directly after the sensor, before the image output pipeline
00:50
intracube
the dead pixel is slightly further to the right (and smeared horizontally) compared to the tiff
00:50
intracube
the texture of the FPN also looks shifted in the same direction
00:50
intracube
that might make it impossible to do an accurate subtract
00:50
Bertl
the image is most likely 128 pixel off to the left as well
00:51
Bertl
because of the full HD vs sensor size
00:53
intracube
it doesn't appear to be that far out. only a pixel or so
00:59
Bertl
keep in mind, that it is probably a single sensel
01:00
Bertl
so that will show up as one bad color component in a snapshot
01:00
Bertl
but will get smeared in the YCbCr conversion for recording
01:08
intracube
Bertl: yep
01:09
intracube
so the FPN frame needs to be redone using the same capture method as the rest of the footage
01:09
intracube
then hopefully it would subtract that bit better
01:11
aombk
http://www.ibtimes.co.uk/fake-advert-apple-wave-convinces-people-charge-their-iphones-microwave-1467253
01:38
Bertl
intracube: yes, or you can apply the transformation used when the recording was done
02:29
Wescotte
left the channel
02:42
Bertl
off to bed now ... have a good one everyone!
02:43
Bertl
changed nick to: Bertl_zZ
05:17
Gegsite
joined the channel
06:04
danieel
se6astian|away: intracube - the sensor can not be made for any particular color space - as there is no single color which would generate response ONLY on its own color plane
06:04
danieel
the calibration matrix has therefore emphasis on the primary, with substraction of the other two, to reach e.g. sRGB
06:09
danieel
if you want to see the kodak chip in operation at live 4k... come to Brno this evening (pioneers unplugged)
06:19
Gegsite
left the channel
08:00
KurtAugust
joined the channel
08:04
KurtAugust
Hi there, nice to see that the Magic Lantern announcement seems to bring in a lot of donations! I have a question https://www.youtube.com/watch?v=E_WCZjPI7Bw#t=55 This was shot without an OLPF, right? Please consider adding one that soften the image enough. Nothing worse than over sharpened detail and color moiree. Unless you're going for a cctv look...
08:12
danieel
around 0.15 you can see the moire on the white buldings left edge...
08:25
PhilippeJ
joined the channel
08:31
KurtAugust
Well, there is a lot in the fence too.
08:46
KurtAugust
left the channel
08:47
KurtAugust
joined the channel
08:58
se6astian|away
changed nick to: se6astian
09:16
se6astian
KurtAugust: correct, the ALpha has no OLPF
09:32
aombk
http://prolost.com/blog/camcamcam
09:42
se6astian
anyone here to help me review a shortperk description
09:43
se6astian
native speaker maybe?
09:59
danieel
left the channel
10:23
Fred__
joined the channel
10:30
se6astian
changed nick to: se6astian|away
10:30
Fred__
If you insert a perk in front of EUR75 the Mega and Hyper supporter gets this to? (>> "All of the above") So at the moment there are 19 Buttons and T-shirts and not 5. is this right?
10:34
KurtAugust
left the channel
10:48
Fred__
left the channel
10:49
Fred__
joined the channel
10:53
Fred__
left the channel
11:16
KurtAugust
joined the channel
11:20
PhilippeJ
left the channel
11:43
PhilippeJ
joined the channel
12:01
Bertl_zZ
changed nick to: Bertl
12:01
Bertl
morning folks!
12:15
se6astian|away
changed nick to: se6astian
12:34
se6astian
back
13:19
masal
joined the channel
13:22
KurtAugust
left the channel
13:26
masal
left the channel
14:28
intracube
left the channel
14:40
intracube
joined the channel
15:27
PhilippeJ
left the channel
16:41
Gegsite
joined the channel
16:52
aombk
you know apart from surveys, polls on your web page are nice
17:27
se6astian
ah good that you remind me to remove the poll on the website :)
17:32
Gegsite
left the channel
17:33
se6astian
done
17:34
aombk
pm?
17:44
comradekingu
joined the channel
17:45
comradekingu
Is it possible to use C-type usb (which has DP) rather than hdmi
17:45
comradekingu
and/or DP in general instead of hdmi
17:46
Bertl
well, the problem with DP is that both integrated chips as well as connectors are rather proprietary and expensive
17:47
Bertl
but yes, from the technical point of view there is nothing which speaks against (e)DP
17:48
Bertl
both, HDMI and DP use TMDS as encoding and both can be supported with minimal change in the electrical connection
17:48
slikdigit
left the channel
17:49
Bertl
for the USB C-type, I've read about it, but I haven't seen any connectors/cables there yet, and it is unclear to me what bandwidth will be supported
17:50
Bertl
also I do not think that it supports DP at all, but I might be wrong there
17:56
comradekingu
Bertl: even the DP chips from TI?
17:56
comradekingu
Its royalty free to implement, so that probably weighs up a bit
17:57
comradekingu
Bertl: you can read about it here http://www.anandtech.com/show/8558/displayport-alternate-mode-for-usb-typec-announced
17:59
Bertl
ah, so in 'alternate' mode, basically using the USB signal pairs for something else :)
18:00
Bertl
yes, that would be doable I guess ... regarding TI chips, do you have any links/suggestions there?
18:01
Bertl
last time I checked I found a number of chips, from ST and similar vendors, but all of them required an NDA
18:04
Bertl
a hint for type-c receptacles would be also appreciated ... haven't found any right now on a quick search :)
18:09
Bertl
but I definitely like the idea
18:15
comradekingu
Bertl: TI has all sorts of DP solutions. They helped develop the dockport with AMD
18:19
Bertl
btw, royalfree to implement, yes, but the standard is not available (e.g. DP 1.3)
18:19
Bertl
you cannot even purchase it, as it seems, you have to become a vesa member to get it
18:22
Bertl
but the annual membership dues are a little more reasonable, starting with 3.500 USD
18:24
comradekingu
It used to be available with earlier DP
18:24
comradekingu
3500 USD isnt as bad as paying a per-device royalty. With 3 hdmis that adds up
18:39
legendin
joined the channel
18:54
aombk
this whole dslr blog/news sites thing is very disappointing. after having contacted many of them i am starting to believe most of them host ONLY paid articles/advertisements.
18:56
aombk
maybe i should make an independent and open cinema blog, like most of them used to be.
19:20
Bertl
welcome legendin!
19:20
Bertl
aombk: sounds good!
19:28
Gegsite
joined the channel
19:35
comradekingu
aombk: the 43rumours site featured the indiegogo campaign, which is where i caught wind of it. I thought i was on the apertus newsletter, but guess not
19:46
hamster
joined the channel
19:46
hamster
Morning all...
19:48
legendin
left the channel
19:49
Bertl
hey hamster!
19:51
hamster
Saw your email - looking like you have a lot of work on your hands
19:51
Bertl
of course, but a lot of fun as well :)
19:52
hamster
Well, actually I was going to write "It looks like you have a lot of fun on your hands...."
19:52
Bertl
freudian slip? :)
19:52
Bertl
in any case, it will be both I guess, so right on the spot
19:52
hamster
:)
19:53
hamster
I'm more than happy to help if I can, but I am quite time poor - I only get a couple of hours a week to 'play' with stuff.
19:54
Bertl
we had a number of encounters already, so I thought I'll bring you in contact with the project, it might be your alley
19:54
hamster
However, I am working on one little project that might be of use. It is a low bandwidth (200Mb/s) serial bus for connecting buttons, switches, sensors and so on around an FPGA.
19:55
hamster
because it only has one wire between devices it leaves a lot of routing resource for fast, wide, stuff - you are going to need that with 4K.
19:56
hamster
What's the data rate from the sensors? about 1GB/s?
19:56
Bertl
you wish :)
19:56
Bertl
with the CMV12k we have 64 LVDS pairs with 600MHz
19:57
Bertl
but of course, we do not need to use the full bandwidth for 'normal' 4k recording
19:57
hamster
Oh, 5GB/s... just a little more than I'm use to!
19:57
Bertl
we have a preliminary design for the Beta, but we are open for improvements and additions where it makes sense
19:58
hamster
I saw the beta footage. Looks awesome.
19:58
Bertl
that is actually Alpha footage, i.e. from the AXIOM Alpha
19:58
Bertl
our shoe-box sized proof of concept camera
19:59
Bertl
basically uses a ZedBoard plus a custom sensor interface
19:59
hamster
How did you get started in this? is it a research project that has taken off?
19:59
Bertl
(the designs and code are on github if you like to check it out)
20:00
Bertl
apertus exists for quite some time, with the aim to develop FOSS/OH for movie making
20:00
Bertl
but the AXIOM basically got started a year ago (or a little more)
20:01
Bertl
I got involved last summer when sebastian was looking for some FPGA programmer to help :)
20:01
Bertl
half a year later we had moving pictures from the AXIOM Alpha
20:02
hamster
Nice.
20:03
hamster
Do you think that the desired design size is ambitious? That is a lot of stuff to pack in.
20:03
Bertl
yes, but we are basically matching the MicroZed size there
20:04
Bertl
and it seems like all the parts fit, except for connectors, where we have to go mini/micro for most of them
20:04
hamster
(joke) You could get the back of the case milled to be a heatsink for the Zynq(/joke)
20:04
Bertl
https://github.com/apertus-open-source-cinema (btw)
20:05
Bertl
we thought about that, and I'm sure we will do something similar on the Gamma (a little down the way)
20:05
hamster
Already found it :) Will have to have a good look when I get home from work.
20:06
philippej
joined the channel
20:07
hamster
The higher spec sensor had some pretty high frame rates - 500 fps IIRC that's HEAPS of data. Are you actually able to get that to storage. (of course you are, but I still wonder!)
20:08
Bertl
300FPS at 10bit full 4k, for the CMV12000, yes
20:08
Bertl
well, not everything needs to be stored, but yes, storing huge amount of data is a major problem
20:09
Bertl
we probably won't solve the storage problem completely in the Beta, which is designed for an external recording device
20:09
Bertl
but we plan to do so with the Gamma
20:11
hamster
I was also taking with somebody who was building something to capture high speed RF (250MB/s for an hour), and it looked like an array of high end SD cards was the only way to go. Or a very high spec PCIe SSD.
20:12
Bertl
yeah, we gave up on the SD idea, we have dual SD interface planned on the Beta for recording still images and similar
20:12
Bertl
but we figured we need at least two SATA 6G SSDs to record raw video
20:14
hamster
They were looking at Samsung SSD XP941 SSDs - e.g. http://www.amazon.com/Samsung-XP941-512GB-PCIe-MZHPU512HCGL/dp/B00JOSM3TK
20:15
Bertl
yeah, PCIe is one option, NVMe another and as I said, SATA 6G
20:16
Bertl
typically you want something which can be easily swapped for archiving
20:18
hamster
The whole professional video market is just a different world, where you can afford to keep content on expensive media because it costs so much to actually obtain it in the first place... quite foreign to me!
20:19
philippej
aloha everyone !
20:19
Bertl
hamster: yeah, it was a completely new world for me as well
20:19
hamster
I would really like to chat a lot more, but I've got to go and do 'real work'.... 8-(
20:19
hamster
I'll leave the window open, and keep an eye on things....
20:19
hamster
Hi Phillippej
20:20
Bertl
no problem, we'll chat later ...
20:21
Bertl
and of course, feel free to hang around ... the channel is also logged in case you loose connection, or want to read up on discussions
20:27
hamster
I've just been scanning the logs from 2014-09-14 - where Anton has had pretty much the same conversation we just had :-)
20:29
Bertl
yeah, sometimes it's good to have archives :)
20:35
se6astian
any native speakers around now
20:35
se6astian
still need some text reviewed
20:35
se6astian
I again managed to smuggle some mistakes into the last update posts :)
20:36
hamster
left the channel
20:36
Bertl
yeah, always impressive :)
20:39
legendin
joined the channel
20:40
legendin
Greetings!
20:40
se6astian
hey there
20:47
legendin
Really excited about this initiative. Finally decided as I was filling out the beta survey that I would like to get involved in Axiom development on a personal basis, and maybe offer modules and other accessories on a professional basis in the future
20:48
Bertl
great!
20:49
legendin
Been itching to get back to programming and electronics for a while now, but it's hard to say now to film work ;)
20:50
Bertl
well, so AXIOM is the perfect combination then :)
20:51
legendin
certainly seems so yes :)
20:53
se6astian
what modules or accessories did you have in mind, I am curious
21:00
legendin
Im am currently in the process of rummaging through old ideas to see if I can draw some inspiration. There are so many possibilities with the AXIOM that open up possibilities that up until now have been hindered by proprietary or incomplete interfaces. The key is to find the old ideas that were put on the shelf for reasons such as "that won't be possible to integrate with a camera" etc.
21:04
philippej
left the channel
21:04
se6astian
I see, well thats exactly the kind of things we were hoping to enable with our approach
21:04
legendin
So far I have one concrete project in mind that would benefit from automation via rec information. Additionally I'd like to tap into live IMU data to control lights on set.
21:06
legendin
And that's what I love about this project. Coming from a linux/oss background, it is quite frustrating at times not to be able to tinker with / hack professional film cameras
21:07
legendin
I just recall how liberating it felt when I installed ML on my 550D a couple of years ago
21:11
se6astian
:)
21:13
Gegsite
left the channel
21:16
legendin
Ooh! With an active mount, a software addon could log lens data, time, duration, clip number etc. on the fly and let the on set script add descriptive metadata on eg. a tablet through wifi
21:16
legendin
that'd be nice as hell
21:18
legendin
pour some processing in there and the addon could search for a frame with the slate and embed a thumbnail in the log as well
21:20
aombk
nice idea
21:21
legendin
Curious as to how the runtime will facilitate such software addons
21:22
aombk
do you know of pcie recording solutions for desktop/studio work and usb recording solutions for laptop recording?
21:22
aombk
like this one http://www.bhphotovideo.com/c/product/1073122-REG/inogeni_4k2usb3_4k_hdmi_to_usb.html
21:24
aombk
i dont like the idea of getting an hdmi monitor/recorder when there are cheaper and powerful laptops available
21:28
aombk
se6astian? bertl?
21:29
Bertl
not on my end, at least not for a reasonable price
21:39
aombk
lets say, the one i linked, would it work on a laptop bandwidthwise?
21:41
Bertl
if your laptop has USB 3.0, yes
21:42
Bertl
but it does do some data conversion as mentioned on the overview
21:58
se6astian
at least it supports Linux and is rather affordable
21:58
se6astian
interesting find
21:58
Bertl
yup, probably worth testing at some point
22:54
se6astian
Canon reacts to the AXIOM - ML team up
22:54
se6astian
http://www.slrlounge.com/canon-teams-stella-mccartney-limited-edition-bagcamera-combo/
22:55
Bertl
LOL :)
22:56
se6astian
with this I sign of for today
22:56
se6astian
good night :)
22:56
Bertl
have a good one!
22:57
se6astian
changed nick to: se6astian|away