Current Server Time: 20:37 (Central Europe)

#apertus IRC Channel Logs

2014/03/15

Timezone: UTC


23:02
FergusL
left the channel
23:05
aPinky
left the channel
23:14
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
23:14
se6astian
has set the topic
23:28
intracube
joined the channel
23:44
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
23:44
se6astian
has set the topic
23:44
troy_s
joined the channel
01:04
Bertl
off to bed now ... have a good one everyone!
01:23
intracube
left the channel
01:23
gwelkind
joined the channel
01:25
gwelkind
left the channel
01:39
Celia
changed nick to: FergusL
04:24
jerknextdoor
joined the channel
04:43
jerknextdoor
left the channel
06:18
niemand
joined the channel
07:25
niemand
left the channel
07:56
Bertl
morning folks!
08:25
troy_s
Bertl: Morn.
08:25
troy_s
Gosh the list stinks.
08:25
troy_s
Few points: FPS and Shutter is all that is required. Fractional is redundant redundant.
08:26
troy_s
Peaking is a wank. No clue why folks want something that belongs in the domain of displays on a camera. Yet to see a single one with it, nor a focus puller that would use it off camera.
08:35
Bertl
well, I don't think that peaking is a big issue, we will have several 'optional' plugins which work in the image pipeline anyway
08:35
Bertl
one more or one less doesn't matter that much, as long as it can be enabled/disabled ... and if something simple like peaking makes folks happy, why not
08:39
Bertl
there are some features which everybody wants to have, but nobody ever uses ... just let's hope they are all that 'simple' as this one :)
08:44
Bertl
I'm still missing descriptions of most of the essentials though, i.e. how something should be controlled/selected/etc ... like for example the actual image format (if you are not using the full sensor resolution), the white balance (I presume a color temperature is appropriate there?), the analog gain/offset etc
09:19
niemand
joined the channel
09:47
niemand
left the channel
10:06
Bertl
off for now ... bbl
10:09
rainer_vie
joined the channel
10:26
niemand
joined the channel
10:43
danieel
joined the channel
10:47
se6astian
joined the channel
10:50
se6astian
good morning
11:55
rainer_vie
moin moin
11:55
rainer_vie
sebastian, on 25th there is the meeting with departure. Do you have time to go with me to the meeting?
13:14
se6astian
could work
13:15
se6astian
but what exactly would be my part there?
13:16
Bertl
anybody around who would like to help a little with data entry? (not you se6astian! :)
13:21
rainer_vie
sebastian you are so talented in speeking with people and holding the focus in essential things ;)
13:23
Bertl
so he is a focus holder not focus puller :)
13:25
rainer_vie
he pulls into focus and hold it :)
13:38
se6astian
I can try :D
13:49
rainer_vie
have sent you a link to the document for their "brief information". would be great if you can have a look on it
13:49
niemand
Bertl: what data entry?
13:49
Bertl
conversion matrices for the hdmi chip
13:50
Bertl
they are given in the datasheet, the data needs to be reformatted for the shell
14:05
intracube
joined the channel
14:38
niemand
left the channel
14:39
niemand
joined the channel
14:49
rexbron_
joined the channel
14:50
rexbron
left the channel
15:06
rainer_vie
wow... debayer to tiff goes in less then a second
15:06
rainer_vie
same on exr
15:09
rainer_vie
time shows: real 0m0.687s
15:23
niemand
left the channel
15:25
niemand
joined the channel
15:31
se6astian
nice
15:36
rainer_vie
with GPU accelaration it could go up to realtime
16:10
Bertl
guess it depends on the debayering algorithm
16:18
se6astian
and the resolution
16:19
Bertl
well, simple debayering (i.e. combining 3/4 pixel) will always work realtime
16:19
se6astian
for preview playback half 1080p -> 960x540 might be more than fine
16:44
rainer_vie
such a proxy mode will be needed on the software side as well
17:02
se6astian
left the channel
17:25
rainer_vie
left the channel
17:48
troy_s
Bertl: I will do new shaper and matrices as soon as I finish painting my wall.
17:49
troy_s
Bertl: Very shortly. Excellent that se6 did them.
17:50
troy_s
Bertl: I also tried to break down the design flow on the wiki.
17:51
troy_s
Bertl: It is evolving, and I should likely show comparisons to existing menu flows. As much as it can be nice to see things with new eyes, the design context of expectation and learned response is a useful thing to keep in mind.
17:52
troy_s
Bertl: The real trick is to figure out the totality of information on the primary display. The menu flow can be sorted into a reasonable ontology.
17:52
Bertl
sounds good!
17:53
troy_s
(for example, the "display" area is likely reasonable for "Color Correct Axis", "Color Temperature", "Look Modification LUTs"
17:53
troy_s
etv.
17:54
troy_s
Bertl: You have a web based interactive control.
17:54
troy_s
?
17:54
Bertl
sebastian is working on one for some time now
17:54
troy_s
(CDL on that would be excellent. Sony does this already.)
17:55
troy_s
Bertl: How does our new data look?
17:55
troy_s
Bertl: You happy with the progress?
17:55
Bertl
CDL? Commercial Driver's License?
17:55
troy_s
Bertl: LOL. CDL is akin to EDL.
17:56
Bertl
English Defence League?
17:56
troy_s
Bertl: "Color Decision List" <- ASC developed method for primary grades.
17:56
troy_s
It defines the protocol.
17:56
Bertl
http://en.wikipedia.org/wiki/ASC_CDL
17:57
troy_s
Some open-endedness as it also does _not_ enforce a scene referred or display referred model, so you have to be careful when exchanging the values across boundaries.
17:57
troy_s
(EG: If the source is Cineon Log then you have to communicate that via ancillary methods)
17:58
troy_s
Bertl: Slope, Offset, Power.
17:58
Bertl
sounds like what we already have
17:59
Bertl
so we just need to read whatever format that has and 'apply' it
17:59
troy_s
Bertl: Advantage of CDL over LGG is A) canonized reference implementation document B) Happily works on scene referred data (Unlike LGG that requires a unity point, and is often hard coded to 1.0)
17:59
troy_s
Bertl: Yep.
17:59
Bertl
we are talking preview here or actual image acquisition?
17:59
troy_s
Bertl: See if you can access the ASC document?
18:00
Bertl
i.e. in the sensor pipeline or in the hdmi/sdi/whatever pipeline
18:00
troy_s
Bertl: Generally the raw data remains untouched, but it provides a very "at source" method to communicate a look to the dailies pipe.
18:00
Bertl
okay, so I understood that right
18:01
troy_s
Bertl: So if the shooter wants to communicate their shooting look, you would likely tack it in prior to 709.
18:01
troy_s
(or 2020)
18:01
Bertl
I don't see an ASC document on the wiki page
18:01
troy_s
Bertl: It is password protected. Let me dig for you. Hold.
18:01
Bertl
okay, thanks, take your time
18:04
troy_s
Bertl: PM
18:05
troy_s
Bertl: That is the official ASC point.
18:06
troy_s
(and apologies to everyone else in here regarding the bureacracy.
18:07
troy_s
Bertl: Raw -> Primary via CDL -> Trim pass to 709 / 2020
18:08
troy_s
(or even DCI-P3 etc.)
18:08
troy_s
Bertl: Also to answer your question re false color
18:09
troy_s
Bertl: There is no standard, but many seem to gravitate to hue rainbow using a Eurowestern idea of "heat" colors
18:09
troy_s
as opposed to kelvin factual color temps
18:11
troy_s
Bertl: F65 I believe is same as http://community.sony.com/t5/F5-F55/False-Color-Update/td-p/195831
18:25
intracube
left the channel
19:37
jerknextdoor
joined the channel
20:04
jerknextdoor
left the channel
21:27
niemand
left the channel
22:38
troy_s
Bertl: Great news
22:38
troy_s
Bertl: Preliminary checking on the Hutch charts (HCT) shows a pretty decent degree of average error. I suspect we are still getting large errors up high, but I haven't looked.
22:39
troy_s
Bertl: Also, the exposure now is a little on the lower side from what I can see...
22:46
troy_s
Bertl: http://www.pasteall.org/pic/show.php?id=68331
22:46
troy_s
(sRGB of course)
22:47
troy_s
Still some upper range strangeness as you can see via the streaking that I suspect is the noise issue Sebs was going to correct.