Current Server Time: 08:18 (Central Europe)

#apertus IRC Channel Logs

2014/01/22

Timezone: UTC


23:40
gcolburn
left the channel
23:59
gwelkind
joined the channel
00:38
FergusL
Hi gwelkind
00:38
FergusL
I' have a few questions about using C in Python
00:44
gwelkind
woah, alright, I am no expert, but shoot
00:44
gwelkind
I'll do my darndest (and probably end up googling things)
00:48
FergusL
gwelkind: didn't you say you did this ?
01:01
gwelkind
Hmm, I think I was just praising it's ease of use and presence, but I know C++ alright and I have integrated them in the past
01:01
gwelkind
what's your question, I'll help if I can!
01:01
gwelkind
^?
01:06
FergusL
well just wondering if you were doing it with Pyrex or Cython or something else gwelkind ?
01:17
gwelkind
Nah, start here: http://docs.python.org/3/extending/extending.html
01:18
FergusL
ha !
01:18
FergusL
thanks !
01:18
gwelkind
np, dude, let me know how it goes :)
02:05
gwelkind
left the channel
02:11
shinokubo
left the channel
02:25
gwelkind
joined the channel
02:37
gwelkind
left the channel
03:04
gwelkind
joined the channel
05:46
gwelkind
left the channel
06:27
gwelkind
joined the channel
08:30
Bertl
off to bed now ... have a good one everyone!
10:43
shinokubo
joined the channel
11:04
shinokubo
left the channel
11:08
shinokubo
joined the channel
11:37
S3bastian
joined the channel
11:41
S3bastian
good afternoon
11:53
FergusL
Hi here
11:53
FergusL
or rather, morning everyone !
11:55
S3bastian
did you switch to BUT ?
11:55
S3bastian
;P
12:00
FergusL
no I'm on FUT, my own timezone
12:05
S3bastian
I see :)
12:58
shinokubo
left the channel
13:18
opusprod
joined the channel
13:46
opusprod
hello everyone
13:51
FergusL
hi opusprod
14:20
S3bastian
hey there!
14:33
philippej
joined the channel
14:52
shinokubo
joined the channel
14:55
S3bastian
time to go home for me
14:55
S3bastian
see you
14:55
S3bastian
left the channel
15:01
shinokubo
left the channel
15:09
philippej
left the channel
15:09
shinokubo
joined the channel
15:20
Bertl
morning everyone!
15:35
gwelkind
left the channel
15:36
philippej
joined the channel
15:58
se6astian
joined the channel
15:58
se6astian
back
16:00
Bertl
wb
16:02
philippej
very in teresting discussion on cml about testing red's sensor response to colour
16:03
FergusL
what list is it on ?
16:03
FergusL
or the thread name ?
16:04
philippej
digital raw log
16:04
philippej
thread name dear oh dear (dn't ask me why, it started with someone saying that all nominated movies were shot on arri)
16:04
philippej
but it all boils down this (from Noel sterrett) : The R, G, and B filters each pass a range of wavelengths. The wider the range, the more difficult it
16:04
philippej
is to nail down a particular color. The narrower the range, the more exact the color, but the less
16:04
philippej
light available to the sensor. Accuracy vs. sensitivity.
16:04
philippej
If the filters are so wide that each nearly overlaps the others, you have crosstalk that can't be
16:04
philippej
fixed, i.e., monochrome.
16:06
philippej
Earlier in the discussion, someone tested the red epic, and in red's case, it seems they choosed a colour array that have a quite wide range, so colour is harder to recover but sensibility is better
16:07
Bertl
as all the pseudo technical articles/threads it should be taken with a lot^Hgrain of salt :)
16:09
philippej
yes
16:09
Bertl
JFYI, if you make the filter ranges really small, let's say red filter 620-660nm, green filter 520-560nm
16:10
philippej
what is interesting is that no one knows what is done inside each camera, and it seems a lot of people believe that raw output = unprocessed
16:10
se6astian
page 60 holds the spectral response curve of the CMV12000 datasheet in case anyone wants to see the overlap
16:10
Bertl
then the sensor will see one yellow (the result of mixing red and green light) but not the other yellow (i.e. the range between 560 and 620nm)
16:11
FergusL
se6astian: wasn't that page blank for a time ?
16:11
se6astian
yes they added it in a recent revision of the datasheet
16:11
FergusL
oh great !
16:11
se6astian
let me update github
16:11
FergusL
Troy was interested in seeing it
16:11
Bertl
bottomline, if you want to avoid blind color spots, you have to have overlaping filters
16:12
Bertl
and they have to overlap in a way which somewhat matches the color hue, otherwise it will go from red to green without passing yellow
16:13
se6astian
done
16:13
se6astian
https://github.com/apertus-open-source-cinema/alpha-hardware/tree/master/Datasheets
16:13
Bertl
maybe add the ANs as well, and I think I have a bunch of other documents which might we worth adding
16:13
se6astian
I will need to ask for permission to publish them first
16:14
Bertl
(but double check with cmosis, not that they have a problem)
16:14
philippej
yes, but as a guy tested it, he said that in case of red camera, the overlap was so huge that it would have poor dr on some colors.
16:14
se6astian
firstly I would like to understand what "dynamic range of colors" means exactly?
16:15
se6astian
latest facebook stats: Weekly Total Reach: 30,622 +2,262.8%
16:16
Bertl
se6astian: out of context it sounds like a PR term :)
16:18
philippej
the guy poster this: http://achtel.com/Downloads/Canon%2050D%20vs%20RED%20Epic.jpg
16:18
philippej
(posted)
16:18
gwelkind
joined the channel
16:19
se6astian
I am subscribed to the list as well and read a mail from time to time but was wondering if they really know what they are talking about...
16:19
Bertl
considering that they post color squares in jpeg, I'd say they know little about anything :)
16:20
philippej
well, unfortunately the whole thread is not online afaik, but it is interesting
16:21
se6astian
maybe they do know a lot but are not able to articulate their thoughts in a way that is clearly comprehensible to others, it can happen, I have seen a lot of brilliant scientists who are just not able to communicate their ideas properly
16:28
FergusL
there are a lot of people on this huge list, primarly made for DOPs, some do know what they are talking about, you can easily notice them, they pop in the right threads generallu
16:35
se6astian
re-reading part of this discussion now it gets a bit more clear for me
16:35
se6astian
obviously I also was part of the problem as I just didnt pay enough attention
16:36
gwelkind
left the channel
16:36
se6astian
there seems to be the idea that the RED MX sensor could be using a different color filter array than the classic bayer pattern
16:36
se6astian
like CYYM filter or another variant: https://en.wikipedia.org/wiki/Color_filter_array
16:37
se6astian
"CYGM gives more accurate luminance information than the Bayer filter, hence a wider dynamic range, but at the expense of color accuracy. "
16:39
se6astian
each color filter allows only a specific band of wavelengths to pass, if you make that band wider you obviously let more photons pass and thus are more sensitive to the same amount of light
16:39
se6astian
though you can not as accurately say of which wavelength the photon originally was (using the reading from a 2x2 photosite block)
16:40
Bertl
first, that really depends on the light
16:40
philippej
I just wanted to point those discussions among people who (at least some) look like they know what they are doing, and it's fascinating to see how much reverse engineering is being done to understand what the hell is doing a particular camera. So, an interesting use case for us being opensource.
16:41
Bertl
if your red filter is centered around 680nm, and your light source is 680nm, then it won't matter if you make the filter band wider or smaller
16:45
FergusL
agreed philippej, I like how CML is a melting pot of all this
16:45
FergusL
and there are some very solid threads particularly about VFX, grip and lighting stuff
16:54
gwelkind
joined the channel
16:57
opusprod
I sebastian , will you give me please the renders you 've done
17:01
se6astian
sure
17:02
guesst
philippej: 400nm is nearly UV :)
17:05
guesst
comparing the cmv- to kac-, the cmv would be more like red (violet instead of blue on 400nm)
17:10
philippej
left the channel
17:25
philippej
joined the channel
17:44
gwelkind
left the channel
18:20
philippej
left the channel
18:24
gwelkind
joined the channel
18:33
shinokubo
left the channel
18:44
gwelkind
left the channel
19:01
gwelkind
joined the channel
19:13
philippej
joined the channel
19:13
philippej
left the channel
19:14
philippej
joined the channel
19:15
philippej
left the channel
19:36
opusprod
left the channel
19:55
philippej
joined the channel
20:01
gwelkind
left the channel
20:03
philippej
left the channel
20:37
gwelkind
joined the channel
20:57
gwelkind
left the channel
21:17
mikkael
joined the channel
21:19
shinokubo
joined the channel
21:22
se6astian
time for bed
21:22
se6astian
good night
21:23
se6astian
left the channel
21:27
carsonau
joined the channel
21:28
carsonau
left the channel
22:07
shinokubo
left the channel
22:09
mikkael
left the channel