Current Server Time: 23:09 (Central Europe)

#apertus IRC Channel Logs

2021/12/06

Timezone: UTC


04:13
Bertl_oO
off to bed now ... have a good one everyone!
04:13
Bertl_oO
changed nick to: Bertl_zZ
05:26
Spirit532
left the channel
05:26
intrac
left the channel
05:26
illwieckz
left the channel
05:26
balrog
left the channel
05:26
lexano
left the channel
05:26
kbeckmann
left the channel
05:26
tpw_rules
left the channel
05:27
intrac
joined the channel
05:27
Spirit532
joined the channel
05:27
illwieckz
joined the channel
05:27
balrog
joined the channel
05:27
lexano
joined the channel
05:27
kbeckmann
joined the channel
05:27
tpw_rules
joined the channel
05:37
Topic
apertus° - open source cinema | www.apertus.org | join the apertus° Lab: http://lab.apertus.org/ | IRC Logs available at: http://irc.apertus.org | Weekly IRC meeting: Monday 18:00 CET/CEST
05:37
vup
has set the topic
05:50
anuejn[m]
left the channel
05:50
Bertl_zZ
left the channel
05:50
Bertl_zZ
joined the channel
05:54
anuejn[m]
joined the channel
07:24
se6astian
good day
11:42
Bertl_zZ
changed nick to: Bertl
11:42
Bertl
morning folks!
11:50
anuejn
morning
13:24
vup
morning :)
13:29
se6astian
PLR HDR settings added to nctrl cooked parameters: https://github.com/apertus-open-source-cinema/nctrl/commit/adbc614fd6106979173e40a1cefc844e7fd00f88
13:40
vup
nice, did you try those on camera yet?
13:41
se6astian
yes works great
13:41
vup
nice, wasn't sure I implemented all the math operators needed :P
13:41
se6astian
now need some feedback from anuejn how to add it to UI elements in dashboard as well
13:41
se6astian
well hdr adds 2 kneepoint exposure times so I just copied the formulas from the normal exposure time
13:43
vup
right, makes sense
13:44
Bertl
you also have two voltage levels
13:51
se6astian
correct, but there is no big math necessary there
13:56
se6astian
for vtfl2 it says: Bit [6] = 0/1: Enable/Disable DAC
13:56
se6astian
that bit is enabled by default
13:56
se6astian
why would anyone want to disable the DAC?
13:57
vup
lower power consumption maybe?
13:57
Bertl
or could also be for external voltages?
13:58
se6astian
interesting
13:59
se6astian
the CMV12000 datasheet only has 2 occurances for "DAC" and thats exactly in the description of the VTFL registers for PLR HDR
14:02
se6astian
vup: how far are you with the hardware exports/viewer website?
14:02
vup
well its probably also the only DACs in the sensor
14:02
Bertl
VTF_LOW* pins are external bias pins
14:03
vup
se6astian: not a lot changed from last week, was quite busy
14:03
se6astian
regarding the mac address OUIs we requested: https://github.com/openmoko/openmoko-usb-oui/pull/37
14:03
se6astian
right
14:04
vup
but I can put some time into it this evening, lets see how far it gets
14:05
se6astian
great
17:00
se6astian
MEETING TIME, who is here?
17:01
Bertl
is here ...
17:01
vup
is here
17:02
se6astian
anything to share vup?
17:06
se6astian
ok, then quick updates from me:
17:07
se6astian
small recorder GUI improvements, wiki improvements, webui additions
17:07
se6astian
experimenting with 24fps mode currently, so 48fps recording
17:07
se6astian
ordered a powerbank with PD support, does not work with lattepanda still
17:07
se6astian
ordered another one that should arrive by the end of the week
17:08
se6astian
designed a preliminary recorder enclosure and 3d printed on friday
17:09
se6astian
grant applicaton with technical university is still work in progress after a VC last week, some internal discussions pending
17:10
se6astian
thats it from my side
17:10
se6astian
Bertl: any news from your side?
17:13
Bertl
nope nothing really new from my side
17:14
Bertl
the usual small work here and there but nothing to brag about :)
17:18
vup
sorry was afk for a sec
17:20
vup
not much from my side either, but anuejn and I started worked on the implementation of the new processing architecture of the recorder
17:20
se6astian
ah cool, can you give some insight?
17:20
vup
we got some basic stuff working again (like bitdepth conversion, debayering, input and output, etc)
17:21
vup
there are two basic (interacting) things that are missing still from the implementation: 1. support for multiple outputs
17:23
vup
and 2. caching that allows reuse of previous computations in case of diverging processing flows
17:23
vup
(for example imagine having a hdmi source and then two outputs, one for live preview and one for saving to disk)
17:25
vup
what we already have is priorization for the different paths, so for example the live preview should never cause frame drops when saving to disk if you have enough processing power for saving to disk but not saving to disk and live preview
17:26
vup
another topic we have been thinking about for the next iteration is tiling / cache blocking to overcome memory bandwidth limits we are now running into pretty easily. If anybody has any thoughts or inputs on that they would be much appreciated
17:27
vup
Other than that I spent some time on small webpage that collects information about all the pcb schematics and boards, maybe ill push some v1 later today
17:28
vup
thats it for now
17:28
se6astian
many thanks
17:29
se6astian
sounds great!
17:29
se6astian
anyone else who wants to report/discuss/share anything?
17:36
se6astian
ok then, MEETING CONCLUDED, many thanks everyone for pushing this project forward, small steps and big ones to "brag about" :)
17:37
Bertl
thanks for moderating!
17:37
se6astian
as always: my pleasure
17:55
Bertl
off for now ... bbl
17:55
Bertl
changed nick to: Bertl_oO
18:13
ilia
joined the channel
18:14
ilia
Hello
18:15
ilia
left the channel
18:15
ilia
joined the channel
18:17
se6astian
hi there!
18:17
ilia
Ah hello it is working
18:18
ilia
Are my messages coming through?
18:18
se6astian
yeah, was just distracted :)
18:18
se6astian
so you said already you didnt have time for much recently but can you quickly tell us what you got done already?
18:24
ilia
I've started getting familiar with the code for raw2dng, got it compiled on macOS, but there seems to be issues related to 32 bit, so I've only maaged to run it on my linux laptop. I have not yet managed to get a correct DNG conversion yet. Am I right in thinking I'd need to use frame-merge and raw2dng to achieve DNG frames from the HDMI data?
18:24
se6astian
yes, if your input is bgr or frame data from hdmi
18:25
se6astian
but we also have snapshot raw12s
18:25
se6astian
http://files.apertus.org/AXIOM-Beta/snapshots/
18:25
se6astian
you might neeed to use the --swap-lines parameter with raw2dng for those
18:25
ilia
Yeah I've had difficulties converting .raw12 to DNG correctly
18:26
se6astian
as in an older camera firmware version we had a bug that swapped even and odd lines when saving stills
18:26
se6astian
what was the error or problem?
18:26
ilia
The DNG files seem to be all orange and crazy
18:26
ilia
Not simple line swap
18:26
ilia
Not sure what was going wrong
18:27
ilia
I was using raw12 generated using frame-merge
18:28
se6astian
if the colors are wrong but the general image does not look distorted then its likely a bayer order issue
18:28
ilia
Ah I see
18:29
se6astian
even/odd lines swapping also changes bayer order and leads to wrong colors
18:29
ilia
I will have a look right now
18:29
se6astian
https://wiki.apertus.org/index.php/Raw2dng#Further_Notes
18:29
se6astian
there is an example image
18:29
se6astian
its all pink then
18:29
se6astian
if its all orange it might be another issue
18:29
se6astian
which files did you try to convert in particular?
18:30
ilia
'magritte-bubbles'
18:31
se6astian
right
18:31
se6astian
there you need to be careful to merge the right A+B frames
18:31
ilia
Understood
18:31
ilia
I just merged the first two
18:31
se6astian
as its easy to accidentally merge B of one frame and A from the next frame
18:32
ilia
I would also like to replace the DNG writing code in raw2dng with that used in MLV App, as this code looks like embedded code, and seems to be broken on 64 bit. What do you think?
18:32
se6astian
the framecounter corner marker helps there
18:32
ilia
I see
18:32
se6astian
this script shows frame counter, etc
18:32
se6astian
https://github.com/apertus-open-source-cinema/misc-tools-utilities/tree/master/raw-via-hdmi/frame-info
18:33
se6astian
I deprecated the *.bgr extenson though and switched to *.frame now with recordings
18:33
ilia
Will try that as well
18:33
se6astian
because its no bgr anymore
18:33
se6astian
it rgb now
18:33
se6astian
but only for newly recorded clips :)
18:34
se6astian
raw2dng is supposed to work inside the camera as well so embedded code make sense
18:34
ilia
Ah I see.
18:34
se6astian
replacing that with code that does the same thing but does not work inside the camera anymore sounds like a bad "upgrade" :)
18:35
ilia
Then I should figure out why this doesn't work on 64 bit.
18:35
se6astian
yes please post the output/errors
18:35
se6astian
might be a simple fix
18:36
se6astian
we are not mac experts though but maybe we will be able to help
18:36
ilia
I suspect it might be due to the 32 bit pointer in the middle of the raw_info structure. I tried allowing a 64 bit pointer, but the code crashes.
18:36
se6astian
linux is likely also 64bit or is it a particularly old laptop?
18:36
ilia
The makefile always compiles the code to 32 bit
18:37
se6astian
ah I see
18:37
ilia
It sets the 32 bit flag unless already compiling on 32 bit arm I think
18:37
se6astian
and macos does not agree I guess
18:37
ilia
Yeah the issue with macOS is no more 32 bit support.
18:37
ilia
I'm sure there's a way of getting the code to work on 64 bit.
18:38
ilia
But for now I will just use my laptop with linux
18:40
ilia
So the format is now rgb instead of bgr with *.frame?
18:40
se6astian
right
18:40
se6astian
yes
18:41
se6astian
I updated https://docs.google.com/document/d/1ERt7XjSjaa7GC1EGPR6PFvjFwDJYQMQKQPc4xy9CwBk/edit# already accordingly
18:41
se6astian
but no new recorded smaples yet
18:41
se6astian
*samples
18:42
ilia
So the only change needed in processing should be choosing the right bayer order in raw2dng?
18:43
se6astian
the output from frame-merge should not be different
18:44
se6astian
and --swap-lines should also not be required for new raw12s
18:44
ilia
Ah ok
18:46
ilia
https://i.ibb.co/CWLp8Qr/image.png
18:46
se6astian
off for dinner now
18:46
ilia
Talk soon
19:02
ilia
left the channel
19:25
ilia
joined the channel
19:26
ilia
left the channel
19:53
anuejn
ilia: tho i think that replacing that dng writing code with something less smelly is a good idea
19:54
anuejn
and I dont really see why the mlvapp code wouldnt work on the camera