#apertus IRC Channel Logs

off to bed now ... have a good one everyone!

can we interface the hardware with an external computer or a processor (something like a raspberry pi or any thing cost effecient ) that would do the job ?

i mean axiom remote

Axiom remote uses a pic32

and what is your question?

and what has it to do with axiom remote? please read apertus pages about axiom remote and axiom beta, as you seem to confuse a lot here

Tmorning

Morning Rincewind

I'd like to take choice 5 please :)

Choice 5 is "Stay in bed and sink into a terrible depression"

Bertl : https://pastebin.com/N1DpMBy5

:)

so does the code look nice now? everything cleaned up and properly formatted?

https://github.com/gr8geek/apertus-challenge/blob/master/characterdevice/chardvr.c

hmm, I would ditch that online formatter ... it doesn't know much about kernel coding style

nope

nope

slowly getting there, still plenty of issues left

keep going!

Bert : I think it is okay

braces inside the kernel message should be avoided and has been fixed

Betl : I am really sorry for inconvenience

*been

Bertl : I can't pin point any of them

which could be solved by formatter, with right settings, automatically.....

;(

and what stops you from setting one up?

I was also not able to find a suitable kernel formatter

you are telling the formatte what you need and what have heuristics to do with it?

I really dont know to use a formatter

you should learn it, as it's mandatory for software development, to keep same style over all the code and not having a jungle of crappy code without proper indentation, whitespaces and whatnot

it goes overflow foe large character stream

Bertl : I was also thinking about a novel checksum algorithim that came to my mind

for formatting I have used linus torvalus 's .clang-format

any ideas why?

yes, that's the question

and why is that?

Bertl : like miss allignment of structure from line 143 onwars and also some misallignment at top which is fixed

The file screwed everything i fixed before

fixed

Oh my god

and in line 92?

that 'fix' actually made it worse

BUF_LEN is more scientific

yes

Hello supragyaraj

supragyaraj--: Hey!

brb on phone rn

supragyaraj-- I have an idea related to AI based debayering are you willing to mentor me

will brb

Bertl : I have fixed it

fixed

single line comments are discouraged

:(

fixed

I have fixed the before but I think I am astronomically unlucky

Bertl : Fixed

maybe train some AI to fix the code for you?

but it requires some love and proper settings

Thanks :)

metal_dent[m] hi, are you around?

yes!

metal_dent[m] you wanted to discuss something?

go on

yes BAndiT1983 and me wanted to have a discussion about the gsoc project

BAndiT1983 is only a supporter here, Supragya is the first mentor ;)

exactly my point XD

supragyaraj--: have you checked out my draft? Bertl checked yesterday and suggested some changes which I'm working on :)

actually, there are more than three :)

if i don't count the LCD logic chips

:)

I don't think sec 3.2 is required to be that extensive

yup

link them up

metal_dent[m], you should work on sec 2.2 also

Maybe create a google doc (going back and forth regarding comments and their resolution is much better on that)

challenge task is still not that important for the proposal, it's more about the main tasks

> Maybe create a google doc (going back and forth regarding comments and their resolution is much better on that)

you can mention the challenge task and place links to repo

is it me or are serif fonts harder to read?

Its related to AI based debayering

"is it me or are serif fonts harder to read?" -> +2000 on this

BAndiT1983: that's just you

yes Bertl is already there :)

pratyush so AI based debayering... how will it work? what problem will it solve? how will it be better? any example?

> metal_dent sec 4.3 is .... umm..... i don't know how to feel about it, maybe rewrite a little

you should stick to "apertus°" as name, i see different ones

supragyaraj--: i intend to use a convolutional neural network to perform debayering

wow, but why?

as it has surpassed state of the art sebayering algorithim

and what's the state of the art in debayering may i ask?

just coincidence

add that to your code submission

*link the repo

pratyush, had you emailed me earlier regarding a similar project?

okay

oh yup

?

at least not in the field, please read about making movies and what is required

I know it's there in breakdown weekwise, but this would really help

okay, i'll add the list before the schedule

markdown rocks

raspberry pi? but why? have you heard of debayering in post-processing?

for real time processing we can use any other lite algorithim

am I dreaming?

why are you adamant about AI, and which AI algo? and why ?

the future is AI!

because AI is the newest buzzword, it's not AI at all, it's pattern matching

AI is something like a very big if/else but hidden behind training DB iteration etc.

well backtracking can work

but since AI is there and so is compute power, who cares?

aham and i develop java at work and am cursing, as we are wasting compute power for unoptimized crap

even java could be better, but many people don't even know the basics of software development, not to mention memory management etc.

pratyush, yeah it works, but... there's something called ROI right?

aah return on investment

back to the topic, metal_dent[m], do you have everything you need for the moment?

supragyaraj--, i mean

very good

snowing? is it some code word for something? like new coke delivery

BAndiT1983: no, it is actual snow :)

*Vienna

BAndiT1983, looking at metal_dent[m]'s code RN, trying to understand ;)

how's Ahmedabad RN, metal_dent[m]?

> how's Ahmedabad RN, metal_dent[m]?

35* here :/

degrees

where?

metal_dent[m], just one google search will disclose my location XD

my society is very strict in that manner, noone was aloowed inside or outside w/o any valid reason

not everyone is a stalker

but have workstations on desk

was used to real one and have even developed software while on the train, as i was asked to work on saturday in the former company, but our trip was already booked, so sent the code to the server and testers have checked it

something works and then stays for 10 years

why should someone care who constantly gets money

metal_dent[m]: i see bootloader as title of your proposal on the gsoc2020 page, it's actually firmware

yes, but till now we have only looked at firmware, as BL is more difficult and requires a lot more hardware knowledge

ah, okay, missed the timeline before

just look at the wording as sentences like "Placing the Axiom Beta software to PIC16 and PIC32 firmware via USB for camera flashing" is confusing

what do you mean there?

and also the timeline tasks are drifting apart, if self-programming and so on are first, then why USB comes later, second thing is, i would suggest to do LVP first, before even touching the pic32 self-programming, this one would follow a bit later, as it requires quite some knowledge of linker script and pic32 structure, this is where i have also to discuss with Bertl often, as it's not always obvious what's going on

LVP and self-programming do not really belong together, but in this order you would see what the next step would require

metal_dent[m]: as reference -> http://ww1.microchip.com/downloads/en/DeviceDoc/40001714C.pdf

it's called ICSProgrammer

hi max_bxl

Hello max

hi max_bxl!

just designed a new part to hold connectors (like HDMI, etc.): https://cloud.apertus.org/index.php/apps/gallery/s/LqkCfTLRJYd9wFB#Nato%20Rail%20Concept/hdmi-clamp.jpg

it seems vup already pushed a fix yesterday

se6ast1an told me

I don't know if we are going to flash the sd card with the new firmware

the fix is to remain at 0.0.18, but pypng guys won't return the code, so it shouldn't get forgotten to update the script at some point

max_bxl: did you try: http://irc.apertus.org/index.php?day=21&month=03&year=2020#340

high priority is getting the problem fixed on your beta now

excellent

can someone check is T1167 is well described ? https://lab.apertus.org/T1167

or the pastebin link

I did method 2 verification and got this https://pastebin.com/khTTqKx9

added IRC logs to the task description

(from what I see of df -h)

64 frames of around 18 mb each

ok

By testing

i.e. doing this ssh root@$BETA ". ./cmv.func; fil_reg 15 0" # disable HDMI stream

Did you see the email response about that?

just _our_ shield, or has this been an issue for someone else?

Bertl_oO: any ideas?

would cleaning pins help?

the external recorder has been confirmed to be working with the beta before?

the shield is tightly connected (there is a white plastic bar above it tightened with two screws)

No I mean the recorder and the beta have worked together before?

se6ast1an : I don't find cmv.func or hdmi.func on the beta in both locations you mentionned => https://pastebin.com/4HPMneDn

Did you try a monitor already?

Pc Monitors not so much

se6ast1an pc display worked (interesting!)

so this could be related to the hdmi-output settings of the axiom?

Please check your setup.sh content

Could be that it's not selected currently

it uses `./gen_init.sh SHOGUNp24` and a few lines later `./gen_init.sh SHOGUN`

I did not (yet) change that script, so it the vanilla: https://github.com/apertus-open-source-cinema/axiom-beta-firmware/blob/master/software/scripts/setup.sh (at least the one from the latest nightly release)

Please disable the 24p line

did just that, currently rebooting

looks good, video even feels snappier. so it could be that I just looked at the 24 (fps?) output until now

thank you

So it tried syncing and got something

oh ok, but the `SHOGUN` preset is 60fps or lower framerate?

FPS

See: https://wiki.apertus.org/index.php/AXIOM_Beta/Manual#1080p60.2F1080p50_Mode

Sorry not halfing

25 or 30 FPS

ok, thx. I realize the "stable" hdmi output did not last that long. I rebooted the axiom and have the same issues again...not sure why

Damn

maybe I should test the pc-display again for a longer period of time and see if those issues appear there as well...

To remove any grease residues

ok

It's also possible that the shogun flame is slightly different than the shogun

well see, I'll report back then. thank you and have nice evening!

You are welcome

(it's enough to make all the RCN and darkframe calibration)

or add something like prerequisites section

sounds good

bye!

what kind of information do we get from this step ?

it checks probably, if the frame is not reaching up to maximum, 12bit = max 4095 and minimum, like 0

yep, but what do we do if it's reaching 4095 ?

then it reaches the highest possible value of the sensor value for AD conversion

off for today, good night

bye