Current Server Time: 20:25 (Central Europe)

#apertus IRC Channel Logs

2021/09/27

Timezone: UTC


01:14
ShinyCyril
left the channel
01:21
ShinyCyril
joined the channel
02:00
ShinyCyril
left the channel
02:05
ShinyCyril
joined the channel
03:33
ShinyCyril
left the channel
05:36
Bertl_oO
off to bed now ... have a good one everyone!
05:36
Bertl_oO
changed nick to: Bertl_zZ
06:54
se6astian
left the channel
06:54
felix_
left the channel
06:54
felix_
joined the channel
06:55
se6astian
joined the channel
08:34
se6astian
good day
13:04
Bertl_zZ
changed nick to: Bertl
13:04
Bertl
morning folks!
16:54
Dest123
joined the channel
17:00
se6astian
MEETING TIME
17:00
se6astian
who is here?
17:00
Dest123
is here
17:01
se6astian
great, Dest123 would you like to share progress and start the reporting right away?
17:01
Bertl
is here
17:02
Dest123
yeah sure
17:02
Dest123
So I'm back this week after a short break xD
17:03
Bertl
wb!
17:03
Dest123
The last problem I was facing was that the CMV12K simulation output channels weren't behaving as expected
17:04
Dest123
thx :) Bertl
17:04
Dest123
I got back to debugging the issue
17:05
Dest123
I started creating multiple configurations of the channels and I plan to test them and see if I can pick up anything
17:06
vup
is here
17:06
Dest123
I'm sending a constant pattern, supposed to be the test pattern, but somehow the pattern isn't easy for me to observe
17:06
anuejn
is here
17:06
Dest123
the data is messed up somehow
17:06
Dest123
so I hope these multiple configurations will help with that
17:07
Dest123
that's it for me :)
17:07
se6astian
do you have any new images for me for the gsoc article ready to be published otherwise?
17:09
Dest123
Just the ones I sent you.
17:09
se6astian
ok
17:09
se6astian
thanks!
17:09
se6astian
vup any news from your side?
17:09
vup
sure a bit
17:10
Bertl
0 or 1?
17:10
anuejn
badum-tsssss
17:10
vup
I continued working on narui, specifically working more on the rendering of filters like blur
17:10
Dest123
xD
17:11
vup
Previously could insert a "SubpassWidget" which would then cause all the children to render to a seperate render target and then use a custom renderpass to draw the children back onto the main render target
17:12
vup
This now gained the additional ability to to a additional pass right before the next higher "SubpassWidget" is encountered, which makes it possible to do a (very inefficient) backdrop blur
17:12
vup
See https://files.niemo.de/2021-09-22%2018-46-29.mkv for a demo of this
17:14
se6astian
looks cool
17:14
vup
I think the "SubpassWidget" stuff is now mostly finished, its mainly waiting on a pull request of the vulkan abstraction layer we are using to be merged and then can be merged into narui
17:14
vup
thats it from me this week
17:15
se6astian
anuejn: anything to add to that from your side?
17:15
se6astian
many thanks vup!
17:15
anuejn
just a tiny bit
17:15
vup
(oh I also have been thinking a lot about order independent transparency and how we could use it to make the rendering of these effects a lot more efficient, so if anybody has experience in this area I would be very interested to talk about that)
17:16
anuejn
I finished the work on the rewrite on the narui macros which give much nicer error messages
17:17
anuejn
also vup and me discussed some changes for axiom-recorder to be able to express the required frame-merging in our pipeline descriptions
17:17
anuejn
apart from that, nothing to report from my side
17:18
se6astian
many thanks, please don't forget to update your wiki project entry
17:18
se6astian
https://wiki.apertus.org/index.php/GUI_for_AXIOM_Recorder
17:18
se6astian
there are no links to code or blog post yet
17:19
se6astian
quick updates from me then, anyone else joined in the meantime who wants to report (bertl will have the closing words as usual)?
17:20
Bertl
yes please :)
17:20
se6astian
We reclaimed our mastodon account, rex did not reply to my emails so we had to go through the server maintainers
17:20
se6astian
so from now on project news again on https://mastodon.social/@apertus
17:21
se6astian
this years GSOC wrap article is also finished
17:21
se6astian
https://www.apertus.org/google-summer-of-code-wrap-2021-article-september-2021
17:21
se6astian
please give it a last minute review before I make it public later today
17:22
se6astian
today me and manfred visited the technical university institute of manufacturing technology
17:22
se6astian
in vienna
17:23
se6astian
a contact established through a past factoryhub manager
17:23
se6astian
they have lots of manufacturing and measurement related machines/projects/staff/students
17:23
se6astian
so definitely lots of potential opportunities for collaboration in the future
17:25
se6astian
they suggested applying for a "small" grant that supports and funds such industry meets university/research projects
17:25
se6astian
https://www.ffg.at/ausschreibungen/InnovationsscheckmitSelbstbehalt
17:25
se6astian
which would lay the foundation for CNC manufacturing of enclosures for example and optimization some aspects of it in the course of the project
17:25
se6astian
where the grant covers 80% of the cost
17:26
se6astian
they gave us the tour and manfred was "dribbling" when seeing al the machines they have
17:27
se6astian
beside the usual huge and small CNC machines they also have an electron microscope or high power CT machine to "look through" metal parts
17:28
yasmin
joined the channel
17:28
Bertl
hello yasmin! glad that you could make it!
17:28
se6astian
and my last topic is of course the raw recording mode to capture raw video from the beta!
17:28
yasmin
Hello :D
17:28
se6astian
lots of progress on that front with work in progress documentation: https://docs.google.com/document/d/1ERt7XjSjaa7GC1EGPR6PFvjFwDJYQMQKQPc4xy9CwBk/edit#heading=h.b8y08xjnmd24
17:29
se6astian
and small tools and scripts to convert or analyze recordings
17:29
se6astian
https://github.com/apertus-open-source-cinema/misc-tools-utilities/tree/master/raw-via-hdmi/scripts
17:30
se6astian
today I started with darkframe creation and compensation
17:30
se6astian
in general it works pretty well already now
17:30
se6astian
whats happening now is mostly convenience with recording and file handling
17:30
se6astian
or improving image quality with the various things the raw2dng tool offers
17:31
se6astian
darkframe, gainframe, etc.
17:31
Bertl
that should give some idea of the current FPN and dynamic noise levels (dark frames)
17:31
se6astian
everything is currently designed for the cmv_snap full frames so some changes and evaluations here and there are reqired
17:32
Bertl
please do some simple statistics there as well, I'm quite curious about the (noise) levels ...
17:32
se6astian
well the raw2dng tool has a --check-darkframe parameter
17:32
se6astian
which shows:
17:32
se6astian
no darkframe:
17:32
se6astian
Not found : darkframe-x1.pgm
17:32
se6astian
Black refcol: ignored
17:32
se6astian
Average : 174.90
17:32
se6astian
Pixel noise : 6.15
17:32
se6astian
Row noise : 5.06 (82.4%)
17:32
se6astian
Col noise : 2.83 (46.0%)
17:32
se6astian
- odd rows : 3.80 (61.9%)
17:32
se6astian
- even rows : 2.88 (46.8%)
17:32
se6astian
against with darkframe:
17:33
se6astian
Dark frame : darkframe-x1.pgm
17:33
se6astian
Dark current: 0
17:33
se6astian
Average : 8.55
17:33
se6astian
Pixel noise : 4.83
17:33
se6astian
Row noise : 2.30 (47.7%)
17:33
se6astian
Col noise : 0.12 (2.4%)
17:33
se6astian
- odd rows : 0.15 (3.1%)
17:33
se6astian
- even rows : 0.15 (3.1%)
17:33
se6astian
the missing dark current is still an issue I think as corrected images are noticeably darker
17:34
se6astian
so still room for improvement
17:34
Bertl
what I'd like to know is the variance of the dark frames, i.e. how much each dark frame deviates from the average
17:34
Bertl
and what values/ranges we have to expect in those dark frames
17:35
se6astian
not sure the current codebase can output that...
17:35
se6astian
but I just started with darkframes today, so lets see
17:35
Bertl
okay, just keep the 'raw dark frames' around
17:36
Bertl
they should compress quite easily
17:36
se6astian
sure
17:36
se6astian
thats it from my side
17:36
se6astian
Bertl please finish us!
17:36
Bertl
maybe yasmin has something to report?
17:37
yasmin
i have downloaded Vivado, and prepared resources for VHDL as i usually write verilog
17:38
se6astian
sounds good!
17:38
yasmin
and by thursday i will finish installation of Vivado, and the simple project in VHDL
17:38
Bertl
excellent! just for information, yasmin is going to be our intern over the next two months if I remember that correctly?
17:38
yasmin
Yes :D
17:39
se6astian
great, welcome aboard yasmin!
17:39
yasmin
Thank you ^_^
17:39
Dest123
Welcome :)
17:39
Bertl
and she will be working on FPGA stuff, which we still need to define ;)
17:40
yasmin
looking forward to starting work :D
17:41
Bertl
great!
17:41
Bertl
anything else you would like to bring up?
17:42
yasmin
no
17:42
Bertl
okay, thanks then!
17:42
Bertl
so last week I spent most of the time on improving the raw mode and testing/verifying various aspects
17:43
Bertl
we ended up modifying the A/B frames so that they basically match the raw data in such way that the A frame contain the even and the B frames the odd lines of a monochrome image
17:44
Bertl
this simplifies the reassembly of A and B frames into raw data as it is basic interleaving
17:45
Bertl
the only drawback here is that the A frames cannot be used as preview image, but that doesn't seem to be something we want to do anyways
17:46
Bertl
we also added the raw markers, which are basically a special overlay mode which allows us to place framecounter and read/write buffer information inside the HDMI output
17:47
Bertl
this simplifies the reconstruction and also allows to identify missing or duplicate frames
17:48
Bertl
the next steps are to investigate how much data we can actually fit in a default 1080p60 frame and how much we can increase the clock rate
17:49
Bertl
ideally without losing compatibility with a preview monitor
17:49
Bertl
that's it from my side for this week
17:50
se6astian
many thanks!
17:50
se6astian
anyone else who wants to share/report?
17:51
se6astian
then: MEETING CONCLUDED, many thanks all participants!
17:52
Dest123
Thank you :)
17:52
yasmin
Thank you :D
17:54
Bertl
thanks for moderating!
17:58
se6astian
my pleasure
17:59
yasmin
left the channel
19:45
Bertl
off for now ... bbl
19:45
Bertl
changed nick to: Bertl_oO
19:56
se6astian
gsoc article released and promoted on twitter and mastodon
00:31
aleb
left the channel
00:43
aleb
joined the channel