Current Server Time: 01:44 (Central Europe)

#apertus IRC Channel Logs

2013/12/17

Timezone: UTC


03:07
jucar
joined the channel
04:08
jucar
left the channel
05:06
aombk
left the channel
06:36
se6astian
joined the channel
06:36
Bertl
morning se6astian!
06:37
se6astian
good morning!
06:38
se6astian
I rendered a new banner image over night
06:38
se6astian
took almost 9 hours
06:38
se6astian
but its super smooth now
06:39
Bertl
sounds interesting ...
06:41
se6astian
I will send an email right away
06:41
se6astian
with NO inline image of course :)
06:47
Bertl
appreciated :)
06:55
sasha_c
joined the channel
06:59
Bertl
evening sasha_c!
07:00
sasha_c
hey Bertl. How are you going?
07:00
Bertl
fine, fine, thanks for asking! and you?
07:01
sasha_c
Very well thanks! Waiting to hear back from Sebastian re: his views on the news article we're all waiting to publish.
07:01
Bertl
he just started his day, I expect feedback soon :)
07:01
sasha_c
Those were really nice images you guys captured :D
07:02
sasha_c
cool
07:03
se6astian
hi sasha
07:03
se6astian
just sent email
07:03
se6astian
and will look at the article next now
07:03
sasha_c
great, can't wait, as it's looking pretty nice imho
07:04
sasha_c
Let me know what your thoughts are re: the images that I've attempted to white balance and colour correct. I'm more than happy for you to correct the originals and re-upload
07:05
Bertl
se6astian: yup, looks indeed nice
07:05
Bertl
sasha_c: they look reasonably well for me, considering that we still have no proper profile
07:06
se6astian
very nice
07:06
se6astian
we should give them a bit of context though
07:06
sasha_c
Awesome, and yes, context is good :D
07:07
Bertl
yes, an interesting detail is that the fan (red) is turning at 1300 RPM
07:07
sasha_c
I take it you're referring to stating that one was taken under halogen lights and the other under natural lighting?
07:08
se6astian
I ll add a text draft
07:17
sasha_c
Great. Now I gotta go out and buy some food supplies before my local shops close. I'll be back in 30 minutes
07:23
se6astian
updated
07:24
Bertl
okay, definitely needs some line break/structure, it falls apart here at smaller widths
07:25
Bertl
i.e. when the third alpha2 image wraps
07:28
se6astian
fixed
07:28
Bertl
yep, looks good now
07:28
se6astian
latest elphel blog post: http://blog.elphel.com/2013/12/nc393-development-progress-the-initial-software/
07:29
Bertl
maybe change the '180deg connector for PMOD debug modules (on back)' to Two PMOD debug modules on the back
07:40
se6astian
done
07:49
Bertl
new lvds code works beautifully now ... will try to integrate it into the snapshot code tonight
07:55
sasha_c
back
07:57
sasha_c
left the channel
08:06
Bertl
off to bed now ... have a good one everyone!
08:09
sashacohen
joined the channel
08:11
sashacohen
changed nick to: sasha_c
08:48
se6astian
sasha_c, article ready for release?
08:48
sasha_c
yep, it is
08:49
se6astian
great
09:00
se6astian
released, and posted
09:00
se6astian
https://www.apertus.org/second-alpha-prototype-ready
09:04
sasha_c
Awesome! I'll help spread the word :)
09:04
se6astian
great
09:05
PhilippeJ
joined the channel
09:14
guesst
se6astian: are there any details on exposure time/gain for those pics? seems noisy :)
09:15
guesst
(especially when they are downscaled pictures - or are they 1:1 crops?)
09:20
sasha_c
guesst: Please remember that the quality of the images in our current post is not an indication of what the final quality will look like when Axiom is ready and we have refined our image processing algorithms.
09:23
guesst
hard to believe that
09:36
mars_
i'd like to see the full picture, if that is possible
09:37
mars_
(and i really like the fan image)
09:42
sasha_c
guesst: Why is that hard to believe?
09:43
guesst
if you got noise, you got noise. no miracle algo can help you in that
09:47
sasha_c
guesst: http://www.google.com.au/patents?hl=en&lr=&vid=USPAT6128039&id=Tk0GAAAAEBAJ&oi=fnd&dq=reducing+sensor+fixed+pattern+noise&printsec=abstract#v=onepage&q=reducing%20sensor%20fixed%20pattern%20noise&f=false
09:47
sasha_c
guest: http://www.darktable.org/2012/12/profiling-sensor-and-photon-noise/
09:47
sasha_c
guesst: http://www.lix.polytechnique.fr/~nielsen/pdf/2007-NoiseReductionCMOS-GVIP.pdf
09:57
[1]se6astian
joined the channel
09:59
se6astian
left the channel
09:59
[1]se6astian
changed nick to: se6astian
10:17
sasha_c
guesst: Please also note that the images we've posted have been compressed from 4k raw captures down to 1024x768 jpg files (less than 1Mb) for web distribution.
10:27
intracube
joined the channel
10:28
[1]se6astian
joined the channel
10:31
se6astian
left the channel
10:31
[1]se6astian
changed nick to: se6astian
10:39
sasha_c
left the channel
10:45
PhilippeJ
left the channel
10:54
se6astian
time to go shopping
10:55
se6astian
see you
10:55
se6astian
left the channel
13:11
se6astian
joined the channel
13:18
dmj_nova
joined the channel
13:30
se6astian
4chan has found us :)
13:30
se6astian
" Excuse for being retarded about cameras.. But, why is there a case fan on it?"
13:30
se6astian
very entertaining: https://boards.4chan.org/g/res/38830103
13:31
se6astian
"All /g/ would know is moar megapicklesis better."
13:31
se6astian
"It's a camera for Richard Stallman so he can make dance videos"
13:32
se6astian
">several thousand dollar security camera
13:32
se6astian
u fuckin wot m8?
13:32
se6astian
"
13:37
dmj_nova
se6astian: yeah, looks like a handful of people getting it, and a handful being hilariously clueless
13:39
se6astian
>ougly contraption with a fan on it because they designed it to overheat
13:39
se6astian
This sums up all open source endeavors.
13:39
se6astian
Its a prototype you mong
13:45
dmj_nova
I especially love the comment about why we're not just using a smartphone processor because that's 4k and doesn't overheat
13:45
se6astian
classic
13:45
dmj_nova
to be fair though, doing so wouldn't be a bad idea if Hollywood would accept such a camera for production
13:46
dmj_nova
you know, replace the REDs with smartphones
13:46
dmj_nova
:P
13:47
dmj_nova
still, awesome job on the prototype
13:49
se6astian
thanks
14:37
Bertl
back now ...
14:49
se6astian
wb
14:49
se6astian
10 minutes till conference call
14:55
mamobros
joined the channel
14:57
mamobros
left the channel
14:57
[1]se6astian
joined the channel
15:00
se6astian
left the channel
15:00
[1]se6astian
changed nick to: se6astian
15:39
aombk
joined the channel
15:39
aombk
left the channel
15:39
aombk
joined the channel
17:31
se6astian
http://www.redsharknews.com/technology/item/1345-more-progress-with-the-apertus-axiom-open-source-modular-camera
17:37
Bertl
nice
17:40
jucar
joined the channel
17:41
dmj_nova
se6astian: :D
20:09
jucar
left the channel
20:33
kkkkaaaa
joined the channel
20:46
kkkkaaaa
left the channel
20:54
se6astian
time for bed
20:54
se6astian
good night
20:55
se6astian
left the channel
21:04
sasha_c
joined the channel
21:26
aombk
left the channel
21:26
aombk
joined the channel
21:26
aombk
left the channel
21:26
aombk
joined the channel
21:31
FergusL
Bertl: off the top of your head, is scene_halogen_2ms_c2.raw16 flipped ?
21:31
FergusL
trying to have my vng algo to work with it
21:33
Bertl
all the newer images are 'flipped' so that the image looks 'normal'
21:34
Bertl
(the flipping works vertically)
21:35
FergusL
okay... so...
21:36
Bertl
the only difference to the 'colors*' image is the bayer order
21:36
Bertl
i.e. RGGB becomes GBRG
21:36
Bertl
(and vice versa)
21:36
FergusL
yes
21:36
FergusL
but the bayer order is kind off hard coded in my code
21:36
FergusL
yes, it is ugly
21:36
Bertl
so simply flip the image vertically first, do your stuff and flip it again
21:37
FergusL
yes but I'm starting on the raw data
21:37
Bertl
imagemagick can do that in a few seconds
21:37
FergusL
I'm considering first dumping the .raw16 to a non-destructive high bitdepth format
21:37
Bertl
imagemagick doesn't care if it is raw or not
21:37
FergusL
flipping it at the same time
21:37
FergusL
well I guess it doesn't work directly on the .raw16 ?
21:38
Bertl
sure, just specify the format
21:38
FergusL
?!
21:38
Bertl
display -size 4096x3072 -depth 16 gray:some.raw16
21:39
FergusL
but I'm worried it's going to bake something, assume it's an image that'll be displayed
21:39
FergusL
wow I'm speechless, I was far from thinking that would work...
21:41
Bertl
xzcat scene_ir850_211ms.raw16.xz | display -size 4096x3072 -depth 16 gray:-
21:41
Bertl
xzcat scene_ir850_211ms.raw16.xz | display -size 4096x3072 -depth 16 -flip gray:-
21:41
Bertl
and similar
21:42
Bertl
convert -size 4096x3072 -depth 16 -flip gray:input.raw16 gray:output.flipped.raw16
21:43
Bertl
so not really a big deal ...
21:43
FergusL
ok thanks
21:43
Bertl
np
21:43
Bertl
off for a nap now ... have fun!
21:44
FergusL
thanks, later
22:08
sasha_c
left the channel