Current Server Time: 13:00 (Central Europe)

#apertus IRC Channel Logs

2015/12/18

Timezone: UTC


00:00
pozitrono
joined the channel
00:08
pozitron
joined the channel
00:10
pozitrono
left the channel
03:09
slikdigit
left the channel
07:08
tezburma
joined the channel
08:00
se6astian|away
changed nick to: se6astian
08:00
Bertl
morning se6astian!
08:12
cbohnens_afk
changed nick to: cbohnens
08:42
se6astian
good morning!
08:44
arpu
joined the channel
09:11
pozitron
left the channel
09:57
Bertl
off to bed now ... have a good one everyone!
09:57
Bertl
changed nick to: Bertl_zZ
10:09
arpu
left the channel
10:56
pozitrono
joined the channel
11:10
pozitrono
left the channel
11:11
arpu
joined the channel
11:18
tezburma
left the channel
11:30
tezburma
joined the channel
14:30
arpu
left the channel
15:22
arpu
joined the channel
16:08
Bertl_zZ
changed nick to: Bertl
16:08
Bertl
morning folks!
16:09
alexML
hi Bertl
16:09
alexML
may I ask you what's the current LVDS clock value?
16:11
Bertl
currently we have 250MHz and 280MHz in use
16:13
alexML
do you remember the value used in these samples? http://files.apertus.org/AXIOM-Beta/snapshots/colorcharts/
16:13
Bertl
most likely 250MHz
16:14
alexML
why I ask: I'm trying to decode the temperature sensor, and they say at 300 MHz, offset at 0 degrees C is 825, and it's proportional to CLK_IN (which I guess it's LVDS clock)
16:15
alexML
with 280 MHz, the offset would be 687
16:15
alexML
but... reg[127] is 635, so I get -18 degrees C :P
16:15
alexML
the camera wasn't in a freezer, I guess
16:15
Bertl
unlikely :)
16:16
Bertl
as far as I understood the datasheet, the offset varies greatly with the sensor
16:17
Bertl
and that also matches our experience with different sensors, so to make sense of that value, you have to do a calibration
16:27
alexML
btw, do you save the lvds clock anywhere in the metadata?
16:29
Bertl
nope, it is not directly available to userspace yet
16:31
alexML
okay, then I'll assume it's 250 in raw2dng when interpreting the metadata, for now
16:42
alexML
would be nice to have it in metadata though, if it's changed often
16:43
alexML
(a quick and dirty way that won't change the file format would be to put it instead of reg[0], which is unused anyway)
16:47
Bertl
it can only change when the FPGA firmware changes (for now) so the firmware identifier, which is available in userspace, could identify it
16:47
Bertl
we can add that to the raw dump (with an option)
16:56
cbohnens
changed nick to: cbohnens_afk
17:08
arpu
left the channel
17:36
se6astian
gotta go
17:36
se6astian
changed nick to: se6astian|away
17:42
irieger
alexML: is the fixed pattern noise correction currently in misc-tools your current state?
17:43
irieger
I want to read through the code the next days so I'd like to know if it is the right code I read ;-)
17:43
irieger
my correction currently seems to not work out well
17:44
alexML
yes, it is (--fixpn option)
17:45
alexML
you may also want to try RawTherapee's version (it's faster)
17:46
alexML
in RT, it's in the Preprocessing tab, under Demosaic (Line Filter)
17:47
alexML
my fix is basically a strong denoise, then I decompose the noise into constant row (or column) offsets + residuals, then apply just the offsets
17:48
alexML
however, I smell that pattern noise intensity increases in bright areas (this is not yet handled in my algoroithm)
18:02
irieger
Ok. I currently have more row noise in the dark parts. But maybe I just had the idea where it may come from. Have to check on the data
18:05
alexML
do you have application note AN01 v6? there they describe quite a few things to try
19:12
alexML
my first attempt at linearizing the sensor response: https://dl.dropboxusercontent.com/u/4124919/bleeding-edge/apertus/linearized/colorchecker_gainx2_15ms_01.dng
19:15
alexML
(you can do white balance on the gray patches, with results a little more consistent than before)
19:15
alexML
(also, the shadows look a bit less crushed)
19:16
alexML
(original sample is http://files.apertus.org/AXIOM-Beta/snapshots/colorcharts/ColorcheckerPassport/colorchecker_gainx2_15ms_01.raw12 )
19:23
pozitron
joined the channel
19:24
pozitron
left the channel
19:41
tezburma
left the channel
19:49
Bertl
off for now ... bbl
19:49
Bertl
changed nick to: Bertl_oO
20:16
slikdigit
joined the channel
20:16
slikdigit
left the channel
20:16
slikdigit
joined the channel
20:32
pozitron
joined the channel