Current Server Time: 19:27 (Central Europe)

#apertus IRC Channel Logs

2016/01/19

Timezone: UTC


00:01
bw
left the channel
00:02
eliemichel
left the channel
00:06
intracube
http://i.imgur.com/j23B4rt.png
00:12
troy_s
intracube: Pink is probably because the bins are pinned.
00:13
intracube
pinned?
00:13
troy_s
FUll.
00:13
troy_s
Full bins.
00:13
troy_s
Also, that thing is noisy as hell which implies that the matrix was based off of a not quite exposed chart.
00:14
intracube
I mean just the bottom right corner
00:14
intracube
it looks a bit like the false colour you sometimes get with long exposures
00:16
intracube
amp noise?
00:17
intracube
half way down: http://www.bythom.com/d80review.htm
01:02
bw
joined the channel
01:15
troy_s
intracube: That pink is almost certainly pinned bins.
01:15
troy_s
;)
01:15
troy_s
intracube: Full wells that are failing to represent their quantities, while the other one still has data left.
01:15
troy_s
(aka non-data in that region)
01:16
intracube
hmm, but as I say it doesn't seem related to the scene lighting
01:17
intracube
troy_s: I mean, the wells might be full making it pink
01:17
intracube
the question is... why just that corner
01:18
troy_s
It's just "overexposed' there
01:18
troy_s
That's a case where a typical camera sets all wells to full if one well is pinned.
01:18
troy_s
(Again, rightfully)
01:18
troy_s
At least I think it is.
01:18
intracube
troy_s: only red and blue channels are affected
01:18
intracube
look at the green channel
01:19
intracube
not even a hint of it getting brighter in that corner
01:19
troy_s
That's probably because that image has been massaged.
01:19
troy_s
Judging from the ridiculous sensor noise etc. from the matrix.
01:19
troy_s
So the original is probably underexposed, and non-uniform.
01:19
troy_s
Then someone tried to "fix" that before profiling (which of course will adversely impact the resultant matrix as a whole)
01:20
troy_s
Pure speculation, but that's where I'll put it.
01:27
intracube
some unscientific colour meddling by me: http://i.imgur.com/HrczJ4C.png (right image)
01:28
intracube
pulling back on the corner pink and some other stuff
05:46
intracube
left the channel
05:54
intracube
joined the channel
06:09
alexML
intracube: when taking the overexposed reference frame, that corner was a little darker
06:16
intracube
which reference frame was used?
06:16
intracube
alexML: ^
06:18
intracube
this one: http://files.apertus.org/AXIOM-Beta/snapshots/IT8%20Charts%20Nikon-08.01.2016/it8-gainx1-offset2047-80ms-02.raw12 ?
06:22
intracube
is looking for the HTC-gainx1-offset2047-80ms-02 version
06:26
intracube
ah, found them
06:29
intracube
wow, that's one heck of a roll-off to that image
06:41
intracube
changed nick to: intracube_afk
06:54
alexML
intracube_afk: http://files.apertus.org/AXIOM-Beta/snapshots/HCT%20Nikon-08.01.2106/
06:55
jucar
joined the channel
06:56
alexML
as reference frames, I used nonlinearity-gainx1-offset2047-75ms-01.raw12 and nonlinearity-gainx1-offset2047-81ms-01.raw12 as clip frames, and darkframe_gainx1_5ms_01.raw12 and darkframe_gainx1_50ms_01.raw12 as dark frames
06:56
alexML
you can create those with e.g. raw2dng darkframe*x1*.raw12 --swap-lines --calc-darkframe
06:57
alexML
or raw2dng nonlin*x1*raw12 --swap-lines --calc-clipframe
06:57
alexML
these two commands will create darkframe-x1.pgm and clipframe-x1.pgm
06:58
alexML
once you have those files in current dir, raw2dng will use them to correct all images you develop
06:58
alexML
now you can run highlight_test from here: https://github.com/apertus-open-source-cinema/misc-tools-utilities/tree/master/nonlinear-highlights
06:58
alexML
and it will give you 4 LUTs
06:59
alexML
copy those into raw2dng, recompile, then: raw2dng *80ms*.raw12 --swap-lines --fixrn --lut
06:59
alexML
and that's how you can reproduce my experiment
07:15
jucar
left the channel
07:29
tezburma
joined the channel
07:39
intracube_afk
alexML: thanks. will have a look tomorrow
07:41
Bertl_zZ
changed nick to: Bertl
07:41
Bertl
morning folks!
08:16
se6astian|away
changed nick to: se6astian
08:18
tezburma
left the channel
08:30
tezburma
joined the channel
08:33
se6astian
good morning
09:45
Walid
joined the channel
10:14
Walid
left the channel
13:42
jucar
joined the channel
13:57
cbohnens
changed nick to: cbohnens|away
14:45
troy_s
Looks like PWLC is landing!
14:45
troy_s
Exciting.
14:52
se6astian
it is :D
15:06
jucar
left the channel
15:33
se6astian
http://files.apertus.org/AXIOM-Beta/BetaPLR1.jpg for full resolution image
15:35
troy_s
se6astian: Amazing. Looks like the UI basics are in place too eh?
15:35
troy_s
se6astian: Can't wait to see the tweaks.
15:37
se6astian
actually its a direct copy of the webgui from the alpha
15:37
se6astian
we just did the groundwork today to make it work with the beta
15:37
troy_s
se6astian: Where you mucking with the knee points for starting out?
15:37
se6astian
we played around a bit with the kneepoints
15:37
troy_s
Really??? I wasn't aware the Alpha got that far.
15:38
se6astian
but it brought us more confusion than clarification
15:38
troy_s
Oh noes. Explain?
15:38
troy_s
I take it this was yourself and Herb?
15:38
se6astian
I am sure alexML will be able to untie the gordian PLR knot :)
15:39
se6astian
yes we only had one complete alpha camera unit so sharing it was tricky
15:39
se6astian
so it was mainly herbert and my stuff on it
15:39
troy_s
What was the confusion point?
15:39
se6astian
now the Alpha is in belgium
15:39
troy_s
Something odd with the knee response?
15:41
se6astian
yes, it felt like pushing the kneepoint towards recovering overexposed areas (the glowing tungsten filament) actually clipped the darker midtones more
15:41
se6astian
I expected it to do exactly the opposite: push those areas further down on the histogram
15:42
troy_s
Hrm. Was the knee close to the photo site well saturation of the filament for example?
15:43
troy_s
Or of the "darker midtones"
15:43
troy_s
Rather.
15:43
se6astian
hard to say, we will do more scientific image series soon to test that
15:43
troy_s
(IE the hingepoint of the knee landing in the range of those midtones)
15:44
troy_s
The mojo behind the math _at_ the knees might be funny.
15:44
se6astian
I ll grab a few extreme snaps and upload them now before I leave the office
15:44
troy_s
One would expect everything _up to_ the knee to be preserved at exposure A, and everything above to be at B, but perhaps there is feathering or something.
15:44
se6astian
but dont expect any miracles :)
15:44
troy_s
I won't.
15:44
troy_s
But I find it interesting.
15:45
troy_s
When does irieger get it?
15:52
Bertl
off for a nap ... bbl
15:52
Bertl
changed nick to: Bertl_zZ
15:54
AxiomHQ
iriegers beta is configured a bit differently for high speed snaps
15:54
AxiomHQ
but I guess he should be able to use it as well
15:55
AxiomHQ
herbert has to confirm the memory issue though
15:55
AxiomHQ
upload complete
15:55
AxiomHQ
http://files.apertus.org/AXIOM-Beta/snapshots/PLRHDR/
15:55
AxiomHQ
check out: http://files.apertus.org/AXIOM-Beta/snapshots/PLRHDR/PLRHDR-Test-linear-15ms.DNG
15:55
AxiomHQ
and http://files.apertus.org/AXIOM-Beta/snapshots/PLRHDR/PLRHDR-Test-linear-1.5ms.DNG
15:56
AxiomHQ
those are PLR off images
15:56
AxiomHQ
then I tuned in these settings
15:56
AxiomHQ
http://files.apertus.org/AXIOM-Beta/snapshots/PLRHDR/PLRHDR-Test-extremePLR2-15ms.png
15:56
AxiomHQ
using 15ms and 1.5ms as exposure time components
15:56
AxiomHQ
http://files.apertus.org/AXIOM-Beta/snapshots/PLRHDR/PLRHDR-Test-extremePLR2-15ms.DNG is the result
15:58
AxiomHQ
15ms -> 1.5ms = a bit more than 3 stops
16:00
se6astian
but you can also make the first kneepoint higher than the second one...
16:00
se6astian
so lots of things to play around
16:00
se6astian
turning off to go home now
16:01
se6astian
changed nick to: se6astian|away
16:07
jucar
joined the channel
16:16
troy_s
God.
16:30
se6astian|away
changed nick to: se6astian
16:31
se6astian
troy_s: have you seen the light? :)
16:31
troy_s
No. Just shut down when I realized I would have to convert formats.
16:32
troy_s
That was the awareness of the DuNG format.
16:32
troy_s
:)
16:59
aombk3
the hdr plr mode can have 2 kneepoints max?
17:04
aombk3
changed nick to: aombk
17:08
troy_s
aombk: Yes.
17:40
slikdigit
joined the channel
17:41
slikdigit
left the channel
17:41
slikdigit
joined the channel
17:46
arpu
left the channel
18:54
intracube_afk
changed nick to: intracube
20:42
arpu
joined the channel
21:06
arpu
left the channel
21:10
tezburma
left the channel
21:31
se6astian
off to bed
21:31
se6astian
changed nick to: se6astian|away
22:15
jucar
left the channel
22:25
arpu
joined the channel