Current Server Time: 01:06 (Central Europe)

#apertus IRC Channel Logs

2022/01/17

Timezone: UTC


00:16
aombk
left the channel
01:20
aombk
joined the channel
03:41
Bertl
off to bed now ... have a good one everyone!
03:41
Bertl
changed nick to: Bertl_zZ
07:38
se6astian
good day
10:47
alMalsamo
left the channel
11:24
alMalsamo
joined the channel
13:40
LordVan_
left the channel
13:44
Bertl_zZ
changed nick to: Bertl
13:44
Bertl
morning folks!
14:34
alMalsamo
left the channel
14:34
alMalsamo
joined the channel
16:12
intrac_
joined the channel
16:13
intrac
left the channel
17:00
se6astian
MEETING TIME, who is here?
17:01
Bertl
is here ...
17:02
vup
is here
17:02
se6astian
great, vup do you want to share some insights into the image analysis, or anything else new?
17:03
vup
sure
17:03
vup
not really a lot to report, but a quick tldr:
17:04
vup
Bertl suggested to put every two rows side by side and like this, a lot of the darkframe patterns seem to match up with the readout
17:05
vup
meaning, that columns 0, 128, 256 seem to be correlated, as well as columns 1, 129, 257, etc
17:05
vup
this can be seen here: sebastian_darkframe_2048_1x_12.990288.blob.zst
17:05
vup
whoops here: https://files.niemo.de/darkframe_pdfs/column_wise_averages.png
17:05
vup
unfortunately it does not seem like we have found the cause for the dynamic column noise yet
17:06
vup
se6astian tried some different settings for the sensor and these don't seem to have made any difference
17:07
vup
There are some interesting patterns in this column noise however
17:07
vup
it does seem to have a pretty interesting frequency composition: https://files.niemo.de/darkframe_pdfs/column_residual_noise_fft.pdf
17:09
vup
its also unclear if correcting this dynamic column noise will actually help that much
17:09
vup
even if we could perfectly correct it, it only brings the standard deviation of the residuals down from 5.39 (at 1x gain) to 4.9)
17:10
vup
so yeah thats pretty much it, se6astian took some full width darkframes today that should be interesting to investigate, as those should have the dark columns
17:11
vup
other than that I think capturing some flat fields and starting to prototype some first image correction flows is pretty much whats next to do
17:11
se6astian
sounds good
17:12
vup
if anybody has input on what metric one could use to measure the "success" of the image correction flow I would be pretty interested. For now the best I (and anuejn) could come up with is capturing a homogeneous signal and then evaluating the corrected image for the homogeneity.
17:12
vup
(For response curve one could then also use different exposure times to measure the linearity)
17:12
vup
oh
17:13
vup
I did not work on this, but somebody implemented 3d lut color correction for the recorder: https://github.com/apertus-open-source-cinema/axiom-recorder/pull/22
17:13
vup
that might be fun to play with
17:13
vup
Ok I think thats it from me
17:13
se6astian
oh cool, I created 3D LUTs to play around with for anuejn recently
17:14
se6astian
who is philipp? very cool addition
17:14
anuejn
philipp is my room-mate :)
17:15
se6astian
har! great :D
17:15
anuejn
somehow there is still a difference between davinci and the recorder but in principle it works :)
17:15
se6astian
could it be color space related?
17:15
anuejn
mmh?
17:17
se6astian
my color space mastery is very basic but a 3D LUT as I understand it is always a relative adjustment and depends on the color space the source image is coming from/resides in
17:19
vup
ah yeah, it could maybe be a srgb vs rgb thing?
17:20
anuejn
we checked a bit but couldnt find the problem (yet)
17:21
anuejn
we also added a parade view to analyze a bit but the difference is even there
17:21
anuejn
so it is not just output differences
17:21
se6astian
you mean the raw image already looks different?
17:22
se6astian
without any adjustments?
17:22
anuejn
no, i meant that the processed image looks different but also really _is_ different
17:22
anuejn
so it is not an output problem
17:23
se6astian
ok
17:23
se6astian
in resolve you can check with Project Settings -> Color Management
17:24
se6astian
Timeline color space in my case is Rec709 Gamma 2.4
17:25
se6astian
anuejn: anything else to report?
17:25
anuejn
nope, nothing from me this week
17:25
se6astian
right
17:26
se6astian
quick updates from me: as vup reported some darkframe captures were conducted
17:26
se6astian
I will send a polycarbonate sheet sample to manfred and Bertl tomorrow via envelope to test laser cutting it
17:27
se6astian
Bertl received the ultra felxible FPC hdmi cable sample last week I ordered from china
17:27
se6astian
GSoC for another season is approaching soon
17:28
se6astian
the org application starts second week of february
17:28
se6astian
so mentors please start to think if you want to mentor again this year
17:28
se6astian
or if someone new wants to mentor maybe
17:28
se6astian
and what could the tasks be, what was done/completed last year, etc.
17:28
se6astian
the idea page will play a central role again in our application
17:29
se6astian
not much else to report from my side
17:29
se6astian
the CCC stil requires more time to release our talk
17:29
se6astian
but it can be watched in a "relive" list
17:30
se6astian
where basically just the streamed original recordings reside
17:30
se6astian
anyone else with news/questions or anything to report?
17:31
Bertl
nothing here this week, was unfortunately otherwise busy ...
17:36
se6astian
alright then, MEETING CONCLUDED, many thanks all participants!
17:40
Bertl
thanks for the moderation!
17:41
se6astian
my pleasure
17:41
Bertl
off for now ... bbl
17:41
Bertl
changed nick to: Bertl_oO
17:47
intrac_
left the channel
17:47
intrac
joined the channel
19:55
alMalsamo
left the channel
19:57
intrac_
joined the channel
19:59
intrac
left the channel
22:05
intrac_
left the channel
22:05
intrac
joined the channel