Current Server Time: 13:18 (Central Europe)

#apertus IRC Channel Logs

2022/11/07

Timezone: UTC


23:43
Bertl
left the channel
23:43
Bertl
joined the channel
23:47
aombk3
left the channel
23:50
aombk
joined the channel
23:51
aombk2
joined the channel
23:55
aombk
left the channel
01:46
Bertl
left the channel
01:46
Bertl
joined the channel
03:50
Bertl
left the channel
03:50
Bertl
joined the channel
05:54
Bertl
left the channel
05:54
Bertl
joined the channel
07:57
Bertl
left the channel
07:57
Bertl
joined the channel
09:37
mithro
left the channel
09:39
mithro
joined the channel
10:39
Bertl
left the channel
10:41
Bertl
joined the channel
12:45
Bertl
left the channel
12:45
Bertl
joined the channel
14:49
Bertl
left the channel
14:49
Bertl
joined the channel
15:28
aombk
joined the channel
15:29
aombk2
left the channel
15:55
aombk
left the channel
15:55
aombk
joined the channel
16:00
se6astian
MEETING TIME, who is here?
16:04
se6astian
anyone?
16:05
se6astian
its not halloween!
16:08
getzi
I'm here, I guess ;)
16:09
se6astian
hurray :D
16:09
se6astian
how is the color charting going?
16:10
getzi
working out the quirks of the raw recorder
16:12
se6astian
have you had the session with vup for darkframe capture ?
16:13
getzi
not yet, but we are in contact
16:14
getzi
for some reason I have some kind of a connectivity issue with the hdmi modules, only one is working reliably unfortunately
16:15
getzi
BUT the old lightmeter you gave me seems to be working :)
16:15
se6astian
ah, did the screw not solvethe problem..
16:15
se6astian
good
16:15
se6astian
we are assembling new hdmi plugin modules on the 21st
16:15
se6astian
but can you try to swap the 2 you have
16:15
se6astian
to verify if its the slot or the module
16:18
Bertl
is here as well (late but nevertheless ;)
16:19
se6astian
very good, any news Bertl ?
16:20
anuejn
is here as well
16:20
Bertl
yes, there has been quite some progress on the work we have been doing for LAM/INT
16:21
se6astian
great! please elaborate
16:22
Bertl
basically decoupling the input and output pipeline, getting high bandwidth HDMI raw data transfer working and testing both square center cutout full frame as well as binned data
16:22
Bertl
the address generators were redone, they now also contain a simulation mode where the settings can be tested on the actual firmware
16:23
Bertl
new scripts have been created to control those address generators (now used for both, capture and HDMI output) and a bunch of new raw output modes have been tested
16:24
Bertl
the sequencer was removed from the gateware (FPGA) and placed in userspace, utilizing the userio kernel module and the PL-PS interrupt framework
16:25
Bertl
we also increased the capture buffer to 3/4th of the entire memory of the microzed, allowing for 32 frames to be captured into memory
16:25
Bertl
(with binning, this allows for up to 128 "full" frames)
16:26
se6astian
cool, how is the realtime part on the CPU behaving, did you dedicate one CPU or is there no issue with the scripts latency ?
16:26
Bertl
the raw HDMI was tested with a Magewell four channel PCIe card, where we can capture A and B raw frames on two separate channels
16:27
Bertl
the script is currently not running at realtime and is a simple python script, the timing might not be perfect but it seems to be sufficient for testing
16:27
Bertl
there is not too much what the script actually does, it waits (with a select) for any interrupt, then updates the address generator and confirms the interrupt
16:28
Bertl
as the capture and the HDMI output is still handled from the FPGA, the script has quite some wiggleroom to update the generators
16:28
se6astian
I see, so the acquisition triggering is still an FPGA task?
16:29
Bertl
the trigger is part of the script, but it is okay to leave the trigger on, which will do one exposure after the other
16:29
Bertl
of course, if you want to do precise timing, it would be necessary to make that also based on some interrupt or have it run as a realtime process with known delays
16:30
Bertl
the entire setup was tested with full frame binning and we achieved 99.2 FPS with 10ms exposure there
16:30
Bertl
that is with overlapping read-out of course
16:30
se6astian
very nice
16:31
se6astian
ah that is also possible now, cool!
16:31
se6astian
would be interesting to measure what frametime and deviations we get from ideal FPS
16:32
Bertl
you mean when you try to fix the frame time to e.g. 25FPS or so?
16:33
se6astian
yes, if the 40ms between frames is maintained
16:33
se6astian
or if its 38, then 42ms, etc.
16:33
Bertl
I'd say there will be some jitter caused by interrupt latency as well as the script sending the trigger
16:33
Bertl
but in the future, we could use an FPGA based 'timer' to trigger the exposure as well
16:33
se6astian
but is it 39.99ms and then 40.01, or 35ms then 45ms, thats what I would be interested in
16:33
Bertl
that should not be too hard to do ;)
16:34
Bertl
also using something faster than python (C or Rust for example) should bring the jitter down there
16:35
Bertl
as I said, the test so far used continuous acquisition, so the frame rate is quite constant
16:35
se6astian
very nice
16:35
se6astian
will you write some documentation about this soon?
16:35
se6astian
and are the scripts released already?
16:35
Bertl
hopefully yes, we probably also want to integrate the scripts in our firmware
16:36
se6astian
exactly :)
16:36
se6astian
anuejn: any news on your front?
16:36
Bertl
the scripts are not secret, but they need some more testing and cleanup (so will be uploaded soon)
16:36
se6astian
great
16:36
Bertl
that's it from my side
16:37
anuejn
not too much on my side
16:37
se6astian
many thanks
16:37
anuejn
i have spoken to the photo lab at my uni
16:37
anuejn
sadly they only have one defined illuminant
16:37
anuejn
(thungsten)
16:37
anuejn
I would have to bring the other
16:38
anuejn
everything else they could provide
16:38
anuejn
I also started working on improving the dng writing in the recorder so that it is possible to user-specify metadata tags
16:38
anuejn
but that is not done yet
16:39
anuejn
thats it from my side
16:39
se6astian
very good, thanks
16:39
se6astian
quick updates from my side: hdmi plugin module pcbs have been received from oshpark and manual assembly is currently planned for the 21st
16:40
se6astian
Bertl: did you also have a chance to look at the new batch of boards already?
16:43
Bertl
no detailed inspection yet, just a few checks ...
16:43
Bertl
but it basically looks like last time, the automated placement is quite good, the manual soldering is rather bad
16:44
se6astian
but any showstoppers so far or just some manual rework required here and there?
16:45
Bertl
thanks to the BiSn workflow, everything should be reworkable
16:45
se6astian
excellent
16:45
Bertl
I haven't checked whether the connectors have been glued or not
16:45
Bertl
but IIRC we requested them not to be glued, so that should be fine as well
16:46
se6astian
we did yes, lets hope they remembered that
16:49
se6astian
ok, anyone else with news or things to report?
16:52
Bertl
left the channel
16:52
Bertl
joined the channel
16:53
se6astian
alright then, MEETING CONCLUDED, many thanks everyone for the news
16:54
Bertl
thanks for the moderation!
16:55
se6astian
my pleasure
17:57
aombk2
joined the channel
17:59
aombk
left the channel
18:21
elafon
left the channel
18:24
aombk2
left the channel
18:24
aombk
joined the channel
18:26
elafon
joined the channel
18:56
Bertl
left the channel
18:56
Bertl
joined the channel
21:00
Bertl
left the channel
21:00
Bertl
joined the channel
21:48
aombk2
joined the channel
21:50
aombk
left the channel