#apertus IRC Channel Logs

"The spam filter installed on this site is currently unavailable. Per site policy, we are unable to accept new submissions until that problem is resolved. Please try resubmitting the form in a couple of minutes."

i tried a few times, guess I'll try again tomorrow, thanks

morning folks!

I tried last night & this morning the Axiom Beta with the Atomos Shogun Inferno, but I have some glitches. I've replied with details to the never-ending email thread we've been communicating on. cc. se6astian

you write that the blue LED stops, what happens to the red one?

how about ./pac1720_info.sh ?

nor did I run those scripts then

to give out those details

do you have any potential hypotheses as to why it might be behaving that way?

but it might be as well a problem with power supply or temperature, thus the scripts

first let's figure out what the problem is, then let's try to find a solution

well, that's always good :)

*reproducible

the photos in the email were taken this morning, everything was connected then, as I did the connecting last night (last night I did not get any image, except the pure black one)

but I can try connecting in a specific order, if that would help

looking at the pictures it might be a temperature problem ... to me it looks like there are no walls connecting the fan with the enclosure, which means that the heat is not transported outside but kind-of circles inside the enclosure

off for now ... bbl

I see - makes sense

I'm not next to it, but I did remember seeing 60 degrees Celsius at some point after running a while

Had also some trouble with black rectangle yesterday, but with stills, as I see bigger area there

Mine have shown normal outer area, but in the middle it was black, after reboot and manual kick it worked again

with stills I have some black bars (which I understand that are a feature of the sensor and hence could be removed) and I have a weird shift in the pixel order, such that the column of first 300 or so pixels on the left are moved to the right of the picture

in that you can see what I'm talking about

The black rectangle I'm talking about is exactly the hdmi area, se6astian had measured it

so you're saying that when I get a black screen on hdmi (does not always happen) I might be getting an equivalent black area in a snapshot I'd take(?)

I will test that hypothesis as well

Asked guys to get my script for you from the Dev camera

If no one has time to get the script, then I will do from home, after work

I have a small script from Bertl for capturing grayscale snapshots, but have no access to the cam from work, maybe someone from apertus core team is at location or available to get it through ssh

this should be easy to fix by a reboot of the system

and it should not happen if you always pair up the kick_manual.sh with halt_manual.sh

so halt_manual.sh followed by kick_manual.sh is the way to go?

I need to check again, but I don't think the reboot worked

aha, I see

if your camera is configured to auto start

then a system service will run the kick_manual.sh for you (once)

if this is disabled (i.e. no auto start on the camera)

you first stop the DMA engine and image output with halt_manual.sh

what is the correct way to shutdown? just `poweroff` command? and would it get in a weird state (the camera itself) if it abruptly stops (e.g. power loss)?

once the red LED stops flashing you can kill the power

unexpected power loss can lead to all kind of issues, including your camera not booting anymore, but all those problems are directly related to the data on the SD card

so having a backup SD card helps in such cases

you get information about the repair on the serial console

so is the pixel shift issue diagnosed to originate from running kick_manual twice?

I can check the hypothesis, but I'm not 100% sure how to bring the camera in a default/reset state, as to properly run that experiment

well, after a reboot (shutdown -r now) the camera is definitely in the proper state

how can I check for sure if I have autostart disabled?

I mean I can see that the blue led flashes like crazy a while after reboot

just capture an image with cmv_snap3

I'm trying to - ./cmv_snap3 -2 -b -r -e 100ms > image.raw12

waiting for sequencer ...

also if you do it again?

no

run ./cmv_snap3 -2 -b -r -e 100ms > image.raw12 again

ok

hdmi output is running and looking normal right?

and fil_reg 15 0 because I didn't want the image to suffer because of HDMI output

but I still get image on the shogun

after running fil_reg 15 0 you get "one" image or live view?

everything else was one image

the entire screen filled with noisy whites like the band I pointed out in the email

waiting for sequencer ... when trying to get a snapshot

what is the difference between disable overlay - gen_reg 11 0x0004F000 - and clear overlay - ./mimg -a -o -P 0 ?

clearing the overlay removes the image to be overlaid

I'm not sure if the disabling works, though

i.e. only ./mimg -a -o -P 0 seems to do the trick

and if it works you can edit your kick_manual.sh and do the "./mimg -a -o -P 0" instead of the line that loads the AXIOM Beta logo overlay

but I also get a thin, maybe 50px wide white noise column

then I tried ./mimg -a -o -P 0 again

lets try disabling the kick automatically starting at boot time

so*]

btw, kick_manual.sh has the mimg lines commented:

# ./mimg -a -o /opt/OVERLAY/overlay_04.rgb

can you put your kick.sh and kick_manual.sh on pastebin please

[root@beta ~]# service cmv12k status

https://thepasteb.in/p/wjh045vNznnuv

note that I just uncommented the "exit 0" line just before I issued the shutdown

so now I'm in a state of no kick being run

*having been run

systemctl status output ^

kick_manual.sh ^

https://thepasteb.in/p/wjh045vlrX9Fv

systemctl status cmv12k.service ^

check this out:

yes, but the status reports misleading information and the halt will be run

ah

:))

ofc

unfortunately the MicroZed doesn't have an RTC

yes good idea

(if you have network access for the camera)

yeah, not a big deal

so I commented ./mimg -a -o /opt/OVERLAY/overlay_05.rgb from kick_manual.sh

*uncommented

wow, I have live video on the shogun

well, I've been expecting it like crazy

and it has no glitches and anything

I've been expecting that as well :)) :P

but then now that I run ./mimg -a -o -P 0 the image goes away

still waiting for sequencer

I mean I will have to at least at some point

I probably should, anyways

and if I go for a reboot, should I then disable the cmv12k service via systemctl and comment the exit 0 in kick.sh?

yes please

unfortunately it does not have a version flag, as I see it

but that was on a different path (although it seems you moved it in the right one afterwards, /root/mimg) and who knows if it worked

so I've done that and rebooted

all good, same as before

next please capture a cmv_snap3 and see if the offset is gone

hdmi image is "still"

I have a feeling it might be misbehaving again

so I just tried again

then the blue led stops

I completely removed the case

so whenever I try to get a snap... "waiting for sequencer"

I did a reboot

then immediately took a snap

but otherwise it just stops taking shots

just let me clarify here: the cmv_snap3 takes a snap (which is okay) but then HDMI out freezes?

I've sent the captured image over email on that thread (same red brick wall background with my fingers in the foreground - except the fingers are not showing right, because I'm moving them (although it's a snap I assume it's somehow related to the movement of the fingers) and the hdmi stream is on

but I can reboot and run kick_manual and after a while, without taking a snap, the HDMI will freeze

and I can run halt_manual followed by kick_manual and the stream would work for a short while, then freeze again

after halt+kick it blinks for 3 seconds (blue), then it says adieu

okay, that all sounds like a hardware defect to me :/

and then it stopped blinking but stayed on

after halt+kick it blinked some 3-4 seconds, then led turned off and stream froze

(on previous halt+reboot+kick it blinked 10+ seconds, then led stayed fully on and stream froze)

so yeah, I assume that in most cases, upon halt+reboot+kick it will stream for 10+ seconds, then led will turn off and stream will freeze

but stream freezes anyways

some dmesg output ^

not sure why it says mmc1 card removed during transfer, but I assume it's probably irrelevant

sure

maybe they arent...

did that, they were quite shiny and dry

stream misbehaving the same way

(last thing is the ethernet becoming available (up))

right, guess we need to get it to the hospital

gotta go

BAndiT1983: Are you there?

He said that there is a long story with color artifacts at the edges.

"We tried some form of edge detection too, but we just used simple interpolation for the edges back then or tried to mix it with the Shoodak data."

He also said that a good part of the artifacts would be possibly elimated by a Low Pass Filter

"The way they are looking now, it will we far to „nervous“ on a moving image and it looks strange in the still image already."

hi TofuLynx, just for short time, will leave soon

I think I have an idea to improve the edge artifacts, I know that the artifacts occur mainly because the two green pixels inside the 2x2 block can be largely different, due to the edge crossing the block in the mid axis

Yup

Makes sense?

Okk!

still wondering if another 3 debayer algorithms are worth to implement later, maybe after gsoc: vng, ppg and amaze

ok, will try your OC fork in the next days, maybe also add dropdown box to switch between the algorithms

this are not important yet, just some which can be found in darktable or rawtherapee

some german paper, but for first impression on algorithms -> https://www.informatik.hu-berlin.de/de/forschung/gebiete/viscom/teaching/media/cphoto10/cphoto10_03.pdf

i can also translate parts of it, if you want to know specific stuff

Sure! It's bookmarked, I will look into it later

https://pixinsight.com/doc/tools/Debayer/Debayer.html

alright, enough links for next time ;)

By the way

The following days, probably until saturday, I will be very occupied with other stuff (University Exam and Family), so my presence can be less frequent here.

no problem, that would give me time to check your implementations and try to extend ProcessingTest UI with dropdown box

I would like to contribute to the UI of OpenCine, after GSoC

when you have time, let us discuss the memory pool stuff

Sure!

and this one should be inspected later -> https://pdfs.semanticscholar.org/9b61/9f4ab38191ecbe976644b7f50fa766ca2898.pdf

seems like GEDI (NEDI?) modification

so, off for today, sleep well, see you

See you! :)