#apertus IRC Channel Logs

exciting backlog about profiling

at least I'm willing too

aren't the raw files enough right now ? that simply need to be dumped as greyscale images in "precise" formats

greetings!

hello

I'm trying to work on the camera calibration

on the left side of the sensor

http://vserver.13thfloor.at/Stuff/AXIOM/RAW/IT8_incand.png

i.e. if they confuse the software, a copy/paste should do the trick

there is a file that tells the calibrations software where the patches are on an IT8 chart. so if it looks at the overlapping defects the average would come out wrong

that could work

I'm getting stuck with an error from Argyll and haven't figured it out yet

it looks like its finding the patches okay

the error I get later is: "colprof: Error - Input file doesn't contain keyword COLOR_REPS"

i think I'll have to search through their source code to figure out what's missing. searching hasn't helped too much

Hi here

morning

Good morning Bertl

hi folks

they are very interesting, about camera profiling

on a related note, can someone point me to a doc which explains the format of the .raw8/.raw16 files here http://vserver.13thfloor.at/Stuff/AXIOM/RAW/ ?

just curious :)

it's a binary raw file

for each pixel

and as this is the raw data, it would follow the pixel layout of the sensor?

yes

so twice as many green photosites as R, B ?

yep

recent .raw8/.raw16 append the sensor registers at the end

has anyone uploaded a good quality debayered 16bits tiff from the IT8 shot ?

hm hm

the sensor has 128 registers to configure how an image is captured

ah ok

you're welcome!

reading them can be done with dd, interpreting them would be useful

I could volunteer for this task

but I thought there might be more exciting things to do right now, to help

in case somebody wonders: 'dd bs=256 skip=98303 if=<file.raw16>' will extract the sensor registers

seek(width*height*2,0) and then read(256) should do the same in python on the file

i.e. doesn't work the unix way with pipes

yes, doesn't work with pipes, unless it's wrapped

so you would need to create a temporary file which is ugly and unneccesary

time to leave the office

FergusL: does the padding you mentioned only apply to the .raw16 files?

or are the 8bit variants padded somehow

i.e. the 4 LSBs

and the bits are just truncated to fit the .raw8 format?

ah cool

np

back now ...

it's indeed walnut, found a few reference for that