#apertus IRC Channel Logs

good day

morning folks!

who is here?

hello I'm here

Hi sebastian that's right, I'm not working on an apertus project yet

kinda :) That's more or less the reason I'm here. I'm starting a new carrer in the IT on the softwear developpement side/coding

I see, great

dev kit to belgium has been received by oscar and he has tested it successfully

I started cleaning up and restructuring the wiki

also the raw hdmi documentation moved to wiki (work in progress) from a google doc

the 3rd version axiom recorder 3d print is done, assembled and works now

vup/anuejn do you want to share anything about the recorder/image analysis?

sure

great!

it has been quite a while since we used/tested those and the results were interesting

but we also have four monochrome engineering samples, and to my surprise, they seems to be working just fine (from the electrical point of view)

i.e. those do not give a good picture, but it is quite usable for testing sockets and similar, as they pass training and provide some output as well

I also started testing the ultra flexible HDMI cables, but the tests are still running there ...

that's it from my side for this week.

you also adapted the raw hdmi firmware a bit IIRC

right

sure

and that should now be in a state where it (atleast sometimes) produces the proper output

with the hdmi raw output it is now a lot easier to capture a lot of image data in a short amount of time, so hopefully some of the calibration processes will get a lot quicker

but that already uncoverd some pretty interesting patterns

and this occurs on atleast two different sensor so we can be pretty sure this is not just a artifact on one specific sensor

at higher gains they always look like the addition of two gaussians: https://files.niemo.de/4x_hist.png

are those monochrome or color sensors?

can you do the evaluations per channel there?

would be interesting whether they match up or not

in theory they should, as monochrome and color sensors should not differe electrically

yeah

no visible light, yes

that another thing, the exposure time seems to only have a really small influence on the mean dark current, and a rather weird one, the extreme exposure times (~0ms and 11ms) seem to have the biggest dark current and the further you get to the "middle" the dark current reduces

Bertl: what do you mean by column blocks?

ah right

so in column and row block?

and see whether they differ based on 'when' they were transferred compared to the exposure

yes, i.e. sum over 0,128,256,384,512, ...

yeah that could be interesting

you expect noise to "accumulate" when the data stays longer in the sensor memory?

interesting

same goes for the rows of course (i.e. I'd also expect some row pattern there)

Bertl: some row patterns are already visible in the last link i posted

thats quite tricky as the lens always vignettes

or what did you have in mind?

thats what a1ex recommended

I will see what I can do

(in an otherwise dark room ;)

or anyone else who wants to report?

i.e. where the data is one value per pixel

do you want to read the file and process it or just look at it?

just look at it for now

(these are always the average of 256 darkframes, the 2x for example is the gain, the number after that the exposure time in ms)

understood

I have to leave for another videoconference so if its alright for you I would conclude this weeks meeting

yeah, I think we are done for today

yeah fine with we

then MEETING CONCLUDED, many thanks

thqnk you, have a nice evening

off for now ... bbl

se6astian: also pretty interesting would be full width darkframes with the dark columns present