Current Server Time: 13:46 (Central Europe)

#apertus IRC Channel Logs

2022/02/11

Timezone: UTC


00:14
vup
se6astian: looks good from what I can tell
00:15
vup
what was the motive for this?
00:29
vup
ohhhh
00:29
vup
I just figured out why the flatfields were so weird
00:30
vup
I fucked up the pixel order
00:30
vup
Now I also finally get the row noise
00:31
vup
it looks like maybe the flatfields are not completely "flat"
00:31
vup
https://files.niemo.de/flatfield_example.png
00:31
vup
now all the "weird" pixels are either on all color channels in the upper left corner, and the complete odd rows and even cols color channel
01:32
Bertl_oO
the upper left corner looks like an illumination issue, i.e. a shadow or reflection on the sensor
01:47
vup
yeah
02:04
vup
se6astian: so with the averaging built in, doing 1024 or even more should be easily doable, right?
02:23
Bertl_oO
off to bed now ... have a good one everyone!
02:23
Bertl_oO
changed nick to: Bertl_zZ
07:27
se6astian
good day
07:27
se6astian
the motive was darkframe IIRC
07:27
se6astian
great that it works
07:28
se6astian
and that you discovered the issue!
07:28
se6astian
for the flatfields some vignetting and light differences on the wall are indeed unavoidable yes
07:29
se6astian
yes 1024 frames with averaging is super convenient
07:30
se6astian
https://files.niemo.de/flatfield_example.png <- also looks like there is some residue on the sensor (the round spots)
08:55
anuejn
vup: very nice
11:43
Bertl_zZ
changed nick to: Bertl
11:43
Bertl
morning folks!
13:27
anuejn
morning :)
13:58
vup
se6astian: yeah looks like it re residue
14:04
vup
interesting then, that the dark columns are that much darker than the rest of the darkframe
14:36
vup
se6astian: do you still have any of the 2048 width darkframe stacks?
14:36
vup
now that I actually get row noise it would be interesting to try and correct it
14:41
se6astian
no I think I deleted everything in the cloud in that folder
14:41
se6astian
can check on recorder inn the evening though
14:41
se6astian
but I can just capture new ones of course
15:08
vup
right, no problem, would be interesting to investigate the row noise
15:18
Bertl
off for now ... bbl
15:18
Bertl
changed nick to: Bertl_oO
17:57
lexano
left the channel
18:14
lexano
joined the channel
18:36
se6astian
vup: starting darkframe stack collection now
18:39
se6astian
1024 frames averaged leads to this error though midway
18:39
se6astian
thread '<unnamed>' panicked at 'called `Result::unwrap()` on an `Err` value: OomError(OutOfDeviceMemory)', src/pipeline_processing/processing_context.rs:137:14
18:39
se6astian
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
18:39
se6astian
./capture_calibration_frame.sh: line 16: 3107 Aborted (core dumped) cargo run --release -- WebcamInput --device 0 ! DualFrameRawDecoder ! Average --n=$3 ! RawBlobWriter --number-of-frames=1 --path frame_averaged_$(ssh root@$1 "cat /axiom-api/devices/cmv12000/computed/exposure_time_ms/value").blob
18:40
se6astian
using 256 frames for average for now
18:40
se6astian
works fine this way
18:54
se6astian
vup please check: https://cloud.apertus.org/index.php/s/aJwdQ9BHQnm4k4p
18:54
se6astian
all darkframes in 2048 width with black coloumns on
18:54
se6astian
256 averages each + snaps
19:35
vup
will check later
19:35
vup
Re the oom error, did you update to the latest version?
19:38
vup
(commit 575b540)
19:39
se6astian
I did git pull yes
19:40
se6astian
will double check the commit name tomorrow
20:28
vup
Hmm curious
20:29
vup
With the latest version there should not be any difference in memory usage between 256 and 1024 (or any number greater than the number of cores for that matter)
21:50
se6astian
the error keeps returning at around 480 frames averaged
21:53
vup
Interesting
21:53
vup
anuejn: was reporting memory leaks aswell, but I have none...