Current Server Time: 06:16 (Central Europe)

#apertus IRC Channel Logs

2015/12/20

Timezone: UTC


00:45
pozitrono
joined the channel
00:47
pozitron
left the channel
01:04
danieeeel
changed nick to: danieel
01:05
troy_s
alexML: So... dE is 4.05
01:05
troy_s
Which might be an improvement.
01:12
troy_s
Boo.
01:12
troy_s
alexML: Sadly, it increased.
01:13
troy_s
alexML: So after much futzing and twiddling and correcting, I am getting peak err = 14.417484, avg err = 4.054184 on yours
01:14
troy_s
alexML: And peak err = 9.316373, avg err = 3.964185 off the original. So your tweaks made it worse.
01:14
troy_s
The peak error is noticeably the more worse, which is quite a large margin.
01:15
troy_s
Granted, se6astian|away really needs to _please_ give me some direct sun with direct sky 5000k shots of the Hutch, with nicely bracketed range from blown out to dark would be great.
01:15
troy_s
se6astian|away: IT8 and the HCT are idea. The ColorChecker Passport, while a great little baseline, doesn't have spectro values and as a result, our guesswork is worse.
01:16
troy_s
That said, all of irieger's work appears vastly solid in terms of twiddling the registers. A delta e of 3.9 on less-than-optimal sources is so much improved over the previous Alpha's results.
01:20
troy_s
So, my gut says that the next working environment would be to see if irieger can offer up another set of tweaked values, and then gather the test chart data to A / B test them.
01:36
Bertl_oO
yes
08:35
Bertl_oO
off to bed now ... have a good one everyone!
08:35
Bertl_oO
changed nick to: Bertl_zZ
09:29
pozitrono
left the channel
09:55
cbohnens|away
changed nick to: cbohnens
10:45
se6astian|away
changed nick to: se6astian
10:58
se6astian
good morning
11:02
se6astian
troy_s: added to TODO list: https://wiki.apertus.org/index.php/AXIOM_Beta_TestsTODO
11:03
se6astian
direct sunlight will be a rarity in the next couple of wheeks though I am afraid, but we will see what we can do :=
11:18
alexML
right now it's sunny here :P
11:19
troy_s
Even cloudy should be somewhere close.
11:20
troy_s
Probably more on the high side due to catching blue sky and diffusing.
11:20
alexML
yeah, and it's probably more likely to get an even illumination
11:20
alexML
so I'd say try both
11:22
troy_s
Direct will be even.
11:22
troy_s
It is a ways away :)
11:22
troy_s
But it also assures even distribution of spectrum, which will help tremendously as well.
11:23
alexML
btw, we have some spectral response curves in the datasheet - would it help to calibrate from those curves?
11:23
troy_s
Why am I up. Ugh.
11:23
troy_s
Nope.
11:24
troy_s
You are welcome to try, but they aren't a terribly reliable method.
11:25
troy_s
(Hence why there are endless streams of threads on characterizing cameras out there. And by endless, I mean endless.)
11:25
se6astian
we have the detailed response curve data from cmosis if that helps
11:32
irieger
troy_s: As I'm done playing with the sensor registers for now I will shoot some color tests the next days. Will upload some stuff soon
11:34
irieger
troy_s: Have to look at row noise but should hopefully be not that much of a problem
11:34
troy_s
I think the averaging on the swatches settles most of the noise issues from the colorimetry side.
11:35
troy_s
Fitting too.
11:35
troy_s
The last batch puts us at 3.9 de on a matrix, which is pretty darn decent.
11:35
irieger
jep. But it looked ugly on some of my last tests. Maybe it was just my FPN profile that went wrong.
11:35
troy_s
Ah.
11:35
troy_s
What were you using? A chart?
11:36
troy_s
Also, I am more interested in those nether regions of non-linearity.
11:36
troy_s
Because those are the ones that are going to cause the most problems shooting anyways.
11:36
troy_s
(Ala rat urine colours in highlights on flesh.)
11:36
irieger
for FPN? Dark frames and a (not ideal, small) grey card. Will shoot a exposure braket of the exposure chart with 1/6th of a stop steps
11:37
irieger
1/6th exposure steps are on the chart
11:37
troy_s
Also, that wiki page isn't great advice.
11:41
troy_s
(Typically the charts want to be shot hotter.)
11:44
irieger
I would do some exposure brakets anyways
11:46
troy_s
Yeppers.
11:47
troy_s
I think that the matrix is likely darn close on the linear region. The only outliers will be in the non-linear, of which that higher light will probably be the most important. Possibly even consistent.
11:50
irieger
Yes, I'm looking forward to see how it will go in all regions of the image.
11:52
irieger
Would prefer not to have colored highlights too ;-)
12:16
arpu
joined the channel
14:04
se6astian
changed nick to: se6astian|away
14:14
pozitron
joined the channel
14:14
pozitron
left the channel
14:26
pozitron
joined the channel
14:53
Bertl_zZ
changed nick to: Bertl
14:53
Bertl
morning folks!
15:16
arpu
left the channel
15:38
se6astian|away
changed nick to: se6astian
18:16
tezburma
joined the channel
19:52
alexML
left the channel
20:18
pozitron
left the channel
21:24
tezburma
left the channel
22:00
Oscar_Apertus
joined the channel
22:01
oscarspierenburg
joined the channel
22:04
hozer
What's going on with Apertus these days, how close are you to shipping hardware?
22:04
Oscar_Apertus
left the channel
22:05
oscarspierenburg
left the channel
22:05
Bertl
hey hozer!
22:05
Bertl
Early Beta hardware is finished, we are currently building the first devices
22:07
Bertl
some parts depend on phase 1 (volume orders) and we also need to test each Early Beta, but then it's time to ship them
22:25
hozer
cool
22:31
se6astian
time for bed for me!
22:31
se6astian
good night
22:32
se6astian
changed nick to: se6astian|away
22:46
pozitrono
joined the channel
22:49
Bertl
off for a nap as well, but probably bbl
22:49
Bertl
changed nick to: Bertl_zZ
23:29
intracube
left the channel
23:31
intracube
joined the channel
23:42
arpu
joined the channel
23:53
arpu
left the channel
23:54
arpu
joined the channel