Current Server Time: 23:01 (Central Europe)

#apertus IRC Channel Logs

2020/05/14

Timezone: UTC


03:50
aombk
left the channel
03:51
aombk
joined the channel
04:32
Bertl
off to bed now ... have a good one everyone!
04:32
Bertl
changed nick to: Bertl_zZ
07:12
BAndiT1983|away
changed nick to: BAndiT1983
07:39
BAndiT1983
changed nick to: BAndiT1983|away
08:13
BAndiT1983|away
changed nick to: BAndiT1983
08:14
se6ast1an
good day
08:15
BAndiT1983
changed nick to: BAndiT1983|away
08:55
max_bxl
joined the channel
08:55
max_bxl
hello everyone!
09:05
anuejn
good morning!
09:06
BAndiT1983|away
changed nick to: BAndiT1983
09:07
se6ast1an
hi maxime!
09:07
se6ast1an
how are things going ?
09:08
max_bxl
fine !
09:09
max_bxl
waiting for Bertrand for online meeting
09:09
se6ast1an
great
09:09
se6ast1an
will reply to your email now
09:09
max_bxl
hopefully we'll resume IRL meetings soon !
09:09
se6ast1an
great
09:09
se6ast1an
regarding the raw2dng conversion of your timelapse clip is everything clear now with the darkframe?
09:12
berto_bxl
joined the channel
09:14
max_bxl
yep !
09:17
max_bxl
Bertrand checked it on it's big monitor, maybe he can tell us more about it
09:17
se6ast1an
<se6ast1an> regarding the raw2dng conversion of your timelapse clip is everything clear now with the darkframe?
09:17
se6ast1an
hi betrand!
09:18
berto_bxl
hi Sebastian
09:18
berto_bxl
Don't know if everything is clear with darkrame
09:18
berto_bxl
The re are change in the monitoring, but I guet weird stuff whenn "axiom-start"
09:19
berto_bxl
I dind't check yet the mail you send us about this, we'll do know with maxime
09:26
se6ast1an
sorry, occupied for a bit with a videoconference
10:04
se6ast1an
back
10:04
se6ast1an
let me know if you have any questions or need any support
10:06
max_bxl
yep, actually it seems there's progress on monitoring, but not on the raw2dng part
10:07
max_bxl
is it normal to have this kind of results : https://pastebin.com/xJMM0k0b
10:08
se6ast1an
looks good
10:08
max_bxl
?
10:08
se6ast1an
what does the result dng look like?
10:08
max_bxl
ok, but it goes on and on forever
10:08
max_bxl
so I don't have a DNG !
10:09
max_bxl
should I wait a bit more ?
10:09
se6ast1an
well you do this on the beta
10:09
max_bxl
yes
10:09
se6ast1an
so its supposed to be much slower than on your pc for example
10:09
max_bxl
on the PC it takes ages too, but I'll give it another try then
10:10
se6ast1an
with the Dark current: dcnuframe-x1.pgm x 1.0 I guess more sophisticated compensations are calculated
10:10
se6ast1an
which takes more time
10:10
se6ast1an
if it takes ages then something is wrong
10:10
se6ast1an
the log says you also do row noise compensation "Row noise from black columns..."
10:11
se6ast1an
3072 rows
10:11
max_bxl
ok
10:11
se6ast1an
but I would suggest to do this post processing on your PC indeed
10:12
se6ast1an
and make sure to snap the picture with -b enable black columns
10:12
se6ast1an
which you did not in the above command
10:13
se6ast1an
maybe it tries to use black coloumns which are not there
10:13
se6ast1an
could explain the long processing, but also would create weird artifacts in the image
10:14
max_bxl
ok
10:14
max_bxl
so the command would be : axiom-snap -2 -r -b -e 1ms > test_snap_r_x1.raw12
10:14
se6ast1an
yes
10:19
max_bxl
ok great, look much better!
10:19
se6ast1an
perfect
10:19
se6ast1an
is it faster as well?
10:20
max_bxl
yes
10:20
max_bxl
there's still some line of noise
10:21
max_bxl
what DCNU stands for ?
10:22
se6ast1an
Dark Current Non Uniformity (DCNU)
10:22
max_bxl
thx
10:22
se6ast1an
there are some additional compensations you can try
10:22
se6ast1an
https://wiki.apertus.org/index.php/Raw2dng
10:23
se6ast1an
Pattern noise correction:
10:23
se6ast1an
--rnfilter=1 : FIR filter for row noise correction from black columns
10:23
se6ast1an
--rnfilter=2 : FIR filter for row noise correction from black columns
10:23
se6ast1an
and per-row median differences in green channels
10:23
se6ast1an
and more experimental ones:
10:23
se6ast1an
--fixrn : Fix row noise by image filtering (slow, guesswork)
10:23
se6ast1an
--fixpn : Fix row and column noise (SLOW, guesswork)
10:23
se6ast1an
--fixrnt : Temporal row noise fix (use with static backgrounds; recommended)
10:23
se6ast1an
--fixpnt : Temporal row/column noise fix (use with static backgrounds)
10:23
se6ast1an
I would suggest trying some/all of those and saving different resulting DNGs
10:24
max_bxl
yep !
10:24
max_bxl
how do you specify output in raw2dng ?
10:24
se6ast1an
then you can apply the same raw developments to all of them and compare with increased exposure value to intensify the noise
10:24
se6ast1an
cat input.raw12 | raw2dng output.dng [options]
10:27
max_bxl
does it make sense to add the options, or we should only take one at a time ?
10:27
max_bxl
ie. cat input.raw12 | raw2dng output.dng --rnfilter=1 --fixrn --fixrnt
10:29
se6ast1an
I would not use multiple corrections at the same time
10:36
max_bxl
is it possible that "cat ..." doesn't pass metadata on ?
10:40
se6ast1an
no
10:43
max_bxl
I got this ... https://pastebin.com/sR1iEv9g
10:44
max_bxl
(the problem is after line 174)
10:45
se6ast1an
indeed seems to be processed differently
10:45
se6ast1an
alexML: you there?
10:45
se6ast1an
max_bxl: I would recommend to use the first method for now and just rename the output dng afterwards
10:46
max_bxl
yep, this process is on going ;)
10:57
se6ast1an
created task: https://lab.apertus.org/T1201
10:57
se6ast1an
alexML: if you are available please take a look at it
10:58
max_bxl
the --rnfilter=2 option seems better for our current setup
11:00
se6ast1an
great
11:00
se6ast1an
showing a side by side comparison of the different compensations would actually be a great addition to the wiki
11:00
se6ast1an
we have nothing like it yet
11:05
se6ast1an
off for lunch
11:11
max_bxl
ok good to know
11:11
max_bxl
will do with a color chart maybe
11:25
se6ast1an
back
11:25
se6ast1an
doesnt ned to be a color chat
11:25
se6ast1an
*chart
11:25
se6ast1an
I even think that a real life image is more interesting
11:25
se6ast1an
like your timelapse
11:28
max_bxl
ok will do then !
11:28
max_bxl
where ?
11:35
se6ast1an
https://wiki.apertus.org/index.php/Raw2dng
11:36
max_bxl
ok will do
11:36
se6ast1an
thanks
11:37
max_bxl
png on the wiki with links to the DNG files on the cloud ?
11:38
se6ast1an
you can zip dngs and upload to wiki as well
11:38
se6ast1an
let me check the max filesize
11:38
max_bxl
what does the dark frame calibration do ? (in order to add this to the wiki, as introduction of the section https://wiki.apertus.org/index.php/Factory_Calibration_(firmware_2.0)#Step_3:_Dark_frame_calibration )
11:38
se6ast1an
8 MB
11:38
se6ast1an
could be a problem
11:38
se6ast1an
please test
11:39
max_bxl
12,8Mo zipped, 13,8Mo tar.gz
11:39
max_bxl
...
11:40
se6ast1an
ah, right then files.apertus.org is probably a good place
11:40
se6ast1an
or cloud.apertus.org
11:40
max_bxl
yes, will do!
11:40
se6ast1an
<max_bxl> what does the dark frame calibration do? <- you mean how does it affect the DNG ?
11:41
max_bxl
yes
11:41
max_bxl
why we do it, how does it improve the DNG
11:47
se6ast1an
https://www.apertus.org/magic-lantern-getting-to-grips-with-axiom-beta-image-sensor-article-feb-2016
11:48
se6ast1an
https://www.magiclantern.fm/forum/index.php?topic=11787.msg129672#msg129672
11:48
se6ast1an
thats the long answer
11:49
se6ast1an
the short answer is: "it significantly improves image noise levels" :)
11:54
berto_bxl
left the channel
11:57
BAndiT1983
changed nick to: BAndiT1983|away
12:31
BAndiT1983|away
changed nick to: BAndiT1983
12:45
max_bxl
done : https://wiki.apertus.org/index.php/Raw2dng#Examples
12:48
max_bxl
off for lunch
13:19
max_bxl
off for today, bye everyone!
13:19
max_bxl
left the channel
13:25
Bertl_zZ
changed nick to: Bertl
13:25
Bertl
morning folks!
13:30
se6ast1an
good day
15:10
aombk2
joined the channel
15:13
aombk
left the channel
16:15
Bertl
off for now ... bbl
16:15
Bertl
changed nick to: Bertl_oO
17:30
se6ast1an
regarding the AXIOM Remote GUI, I would like to discuss a feature I have in mind: I want to add a switch/option that allows the user to choose between setting parameters while the value is being changed or only after the selection has been completed
17:32
BAndiT1983
toggle button for live/post?
17:32
se6ast1an
for example on a DSLR or broadcast/prosumer camera you expect to see a change in exposure time or aperture immediately, while turning a knob basically
17:32
se6ast1an
but with the current menu system we have in place it would be local that a change only propagates after pressing the set button or pushing the knob on the value you want
17:33
se6ast1an
so I would indeed like to add a button to each parametermenu where you can switch between LIVE/POST as you call it
17:33
BAndiT1983
this is where event handling is useful, you can tell to fire when a value changes
17:33
se6ast1an
the question is if the term "POST" is self expalantory
17:33
BAndiT1983
was just a quick shot
17:33
se6ast1an
and if this choice should be saved with each parameter individually
17:34
se6ast1an
or if there should be different behaviour while recording/not recording
17:34
se6ast1an
in the end this has a lot of potential to confuse people
17:34
se6ast1an
so finding the right term and way to display this is essential
17:44
se6ast1an
I guess I will also put these into a lab task
17:48
BAndiT1983
depends on the option, but with handlers you can decide when the change should be triggered, e.g. instantly, on button press, when something else happens etc.
17:49
se6ast1an
yes technically this should be easy to solve
17:49
se6ast1an
I think the main challenge is on UX side
17:55
se6ast1an
https://lab.apertus.org/T1202
18:05
BAndiT1983
changed nick to: BAndiT1983|away
18:34
Spirit532
left the channel
18:34
Spirit532
joined the channel
19:19
BAndiT1983|away
changed nick to: BAndiT1983
19:19
metal_dent[m]
Hi, sorry for not reporting today, I was still working on the makefile and finally finished it
19:20
BAndiT1983
hi metal_dent[m], very good! had any problems?
19:20
metal_dent[m]
it creates the image headers now the way we want
19:21
BAndiT1983
nice!
19:21
metal_dent[m]
actually not exactly the way we want as the pixel/hex data in the header are all in one line without indentations so there is one problem
19:22
BAndiT1983
it shouldn't really matter, as we want the data, also it's auto-generated, so don't worry
19:25
metal_dent[m]
ohh, and i was here trying to find ways to overcome that problem ':D
19:30
BAndiT1983
it can be solved rather easy with some tools, e.g. clang-format or similar, but focus on functionality first, can incorporate formatting when CircleCI is building
19:31
Bertl_oO
metal_dent[m]: how is the final output generated?
19:32
BAndiT1983
oh, Bertl is there, if you have 1 minute then i would like to know rather important info regarding the path for the PCB probe
19:33
Bertl_oO
sure
19:33
BAndiT1983
have posted 2 images of the paths -> https://lab.apertus.org/T1199
19:34
BAndiT1983
second one is just dumb sorting by x and y, but thought about adding like x and y threshold values to the command line, would be for the search grid
19:34
metal_dent[m]
makefile -> https://github.com/MetalDent/AXIOM-Remote/blob/dev/Firmware/Media/Icons/Makefile
19:34
metal_dent[m]
and header file -> https://github.com/MetalDent/AXIOM-Remote/blob/dev/Firmware/Media/Icons/ApertusRingLogo.h
19:35
Bertl_oO
BAndiT1983: can you use a script (or other way) to simply sum up the path lengths as metric?
19:35
BAndiT1983
problem is, that not all parts are defined by origin, like top and bottom connector pad rows, so i have to think of some method to wander between elements and pads have to weight it also
19:35
BAndiT1983
can do, as i know the coordinates exaclty
19:35
BAndiT1983
*exactly
19:36
BAndiT1983
metal_dent[m]: looks good!
19:37
BAndiT1983
we can reduce the makefile even further, but not important right now
19:37
metal_dent[m]
alright then i'll sleep now! nn!
19:37
BAndiT1983
sleep well, good night
19:38
BAndiT1983
Bertl_oO: regarding the lengths, currently it's only between the element positions, but i suppose that the element BETA is in the center or so, it contains the connector pads
19:39
BAndiT1983
what i would like to achieve is to create a proper path from top-left or bottom-left, which goes from left to right, then turns and in the next row it goes from right to left
19:40
BAndiT1983
maybe i will just calculate all pad paths, then sort like now plus a threshold for position, as you will precalculate the paths, then i could add some command line threshold values for path "smoothing"
19:40
Bertl_oO
yeah, just a sum of the part to part moves should give a rough idea how much better this path is
19:41
BAndiT1983
ok, will implement that, just simple trigonometry
19:41
Bertl_oO
the problem with optimizing pad paths is that becomes a traveling salesperson problem which is non trivial
19:42
BAndiT1983
i know, that's exactly the first thing i've stumbled when started to search for a solution
19:42
Bertl_oO
simple sorting top to bottom will give very bad results for vertical connectors
19:42
BAndiT1983
maybe there is a module in python for this
19:42
Bertl_oO
similarily sorting left to right will cause problems in the other direction
19:42
BAndiT1983
let me reflect a bit on that problem, it's solvable, will use pad position and element placement as factor and weighting or so
19:43
Bertl_oO
it is solvable but solving it for this number of pads will take way longer than the probing :)
19:44
BAndiT1983
there is always a module for python ;) https://pypi.org/project/tsp/
19:44
BAndiT1983
https://ericphanson.com/blog/2016/the-traveling-salesman-and-10-lines-of-python/
19:45
Bertl_oO
well, you'll see :)
19:46
BAndiT1983
am rather confident that we can get it up and running soon, panelizer also has some tests done, maybe we should move the PCB probe repo to apertus and alter also the panelizer
19:49
BAndiT1983
*later
19:52
se6ast1an
BAndiT1983: I need some event handling support with function pointers, should I commit and you take a look tomorrow or so?
19:52
se6ast1an
it currently breaks the build though...
19:53
BAndiT1983
yes, please commit, also you can take a look at the handlers which were already there, i think it was IButton::Activate() or similar, name is still not fitting, but should be ok for now
19:53
se6ast1an
I did, thats how I created them
19:53
se6ast1an
but now get
19:53
se6ast1an
UI/Screens/MainMenu.h:54:64: error: no matching function for call to 'NumericMenuItem::SetHandler(void (MainMenu::*)(void*))'
19:53
se6ast1an
_lcdBrightness.SetHandler(&LCDBrightnessMenuItemHandler);
19:54
BAndiT1983
it has to be static in the class
19:54
BAndiT1983
static void AnalogGainButtonHandler(void* sender);
19:54
se6ast1an
ah!
19:54
BAndiT1983
like in MainPage.h
19:55
BAndiT1983
my implementation is not that polished yet though
19:55
se6ast1an
I looked at mainpage.cpp but not .h, thanks
19:55
BAndiT1983
no problem
19:56
se6ast1an
yep that did it, compiles
20:00
se6ast1an
pushed fix that should fix the build again
20:05
BAndiT1983
yep, CircleCI is green again!
20:14
se6ast1an
handler works, off to tv evening
20:14
se6ast1an
only thing I still need to figure out is how to get the ili object into that function now :)
20:34
comradekingu
left the channel
20:51
comradekingu
joined the channel
21:30
se6ast1an
off to bed
21:30
se6ast1an
good night
21:34
Bertl_oO
nn
22:01
uberardy
left the channel
22:03
uberardy
joined the channel
22:55
BAndiT1983
changed nick to: BAndiT1983|away
22:57
anuejn
register dumps from the camlink: https://paste.niemo.de/higaxonabi.makefile
22:59
anuejn
(comparing output from my laptop and the micro)
23:10
anuejn
good night everyone!
23:35
megora
joined the channel
00:32
Bertl_oO
anuejn: ah, nice, tx
00:32
Bertl_oO
vup: any interesting differences?
00:32
vup
did not look at them yet
00:50
Bertl_oO
ah, sorry, assumed you had ...