Current Server Time: 03:12 (Central Europe)

#apertus IRC Channel Logs

2013/11/05

Timezone: UTC


23:17
dmj_nova1
Wouldn't editorial ideally be independent of the stage of color/vfx processing?
23:18
dmj_nova1
troy_s:
23:19
dmj_nova1
by reconstruction from image, do you mean breaking shots apart for grading based on image content?
23:23
FergusL
yes, these two parts had to be separated, only a one-pass grading, something "light" might be applied, just to be watchable
23:27
dmj_nova1
Yeah, ideally I would imagine vfx/color should feed back into editorial as they progress.
23:31
dmj_nova1
left the channel
23:31
dmj_nova
joined the channel
00:02
dmj_nova1
joined the channel
00:04
dmj_nova
left the channel
00:20
mikkael
left the channel
00:31
FergusL
On all small projects I've been on, grading only started after editing.
00:36
dmj_nova1
FergusL: How did the editing -> grading pipeline work on your small projects?
00:54
dmj_nova1
left the channel
01:01
FergusL
joined the channel
01:37
Bertl
got a nice color/grey chart today from sebastian
01:37
FergusL
ha cool !
01:37
FergusL
which one is it ?
01:38
Bertl
will see, maybe I can make some pictures tonight, sec, let me check
01:39
Bertl
it says: IT8.7/2-1993
01:40
Bertl
and below that 2010.06
01:40
Bertl
on the other side Charge: R100604
01:41
FergusL
well I think that's the very same troy_s suggested we get and offered to source
01:41
Bertl
perfect then
01:41
FergusL
great news
01:41
Bertl
I consider it perfect because it seems to have the perfect size for my setup
01:42
FergusL
if I can focus on it, maybe i'll have the first resule of VNG debayer on your raw captures
01:42
Bertl
still have to find a solution for lighting
01:42
FergusL
maybe a well-balanced high CRI CFL bulb could do
01:45
Bertl
I can probably reverse the neutron flow :)
01:45
FergusL
wuh ?!
01:46
Bertl
CFL = compact fluorescent lamp
01:46
Bertl
CRI = color rendering index?
01:46
FergusL
yes
01:46
FergusL
yes
01:47
Bertl
*rendition
01:47
FergusL
CFL bulbs sold for growing plants are full spectrum aka high CRI
01:47
Bertl
so where would I find a well-balanced high CRI CFL bulb?
01:47
FergusL
but it's still a hacked solution
01:47
Bertl
ah, okay, that is something I can probably find
01:48
FergusL
better wait from someone else approval !
01:48
Bertl
so we want 'sun-light' like bulbs?
01:48
FergusL
I'm assuming we want the purest source to light the chart
01:49
Bertl
well, pure is vague here
01:49
Bertl
http://en.wikipedia.org/wiki/File:Solar_Spectrum.png
01:50
FergusL
CRI charts generally include only the visible range hehe
01:51
FergusL
450 - 700 or around, but I guess you know about all that ?
01:53
Bertl
I have no idea about the charts (yet) but I know about the physics behind it
01:54
Bertl
and it makes sense to assume that calibration charts do only address the visible spectrum (at least for normal imaging and moving pictures :)
01:55
Bertl
my point with the solar spectrum is that the spectrum of the sun is not really uniform and while it is special in its own way, it is not the only light source we have
01:56
Bertl
so the main question is probably, do we aim to recreate sun light or do we aim to have a certain 'color temperature' white
01:56
FergusL
yes
01:57
FergusL
and especially with normal imaging, the actual source isn't that much what matters, what matters is how we calibrate our sensing device !
01:57
[1]Sasha
joined the channel
01:57
FergusL
yes, color temperature
01:58
FergusL
well I guess we're trying to get 1) A source that would illuminate as evenly as possible in the 450-700nm range 2) Calibrate our sensor (actually not the sensor itself but well you know that) so that its "base color" matches that of the light source
02:00
Bertl
so all we need is a furnace which can do 5780K :)
02:00
Bertl
or if we want daylight, 6500K
02:01
Bertl
I wonder if lamp manufacturers specify the color temperature
02:01
Bertl
(for normal room lighting I mean)
02:01
FergusL
http://www.narva-bel.de/ressourcen/spektrum/popup/SP_958_NARVA_BIO_vital_T5_RGB.jpg.png that's what I called "CRI chart", I meant a graph actually, it's as stupid as that... just a graph of how the source performs accross the range
02:02
FergusL
now with CFL it's different I think, but it's on the packaging most of the time
02:02
FergusL
can we assume it's betweel 3200 and 4000 ?
02:02
Bertl
I see, but that has no color temperature associated (the chart)?
02:03
Bertl
does the 958 NARVA BIO specify that?
02:03
FergusL
I don't know, it's just a random picture I found, but it's clearly refering a light source so probably yes
02:03
Bertl
ah, I get it, this _is_ the spectral output of this specific light source
02:04
Bertl
which gives me an idea
02:05
FergusL
http://www.amazon.com/Hydroponic-Full-Spectrum-Light-5500K/dp/B005P29K1S I was thinking something like that
02:05
Bertl
we could use a small slit or a prism (or maybe both) to break down the spectrum of whatever light source we have and capture the image with the sensor
02:07
Bertl
now if we knew the spectral response of the sensor (CMOSIS hasn't disclosed that yet) we could calculate the spectrum of the light source :)
02:20
FergusL
interesting fact I just read
02:20
FergusL
CRI is to be read as a "distance from 100"
02:20
troy_s
Bertl / FergusL NOT CFL.
02:20
Bertl
http://housecraft.ca/wp-content/uploads/2012/09/spectral_responses2.png
02:21
FergusL
Ok, no CFL
02:21
Bertl
yeah, it looks like it has the 'worst' spectral distribution
02:21
Bertl
atm, probably a simple incandescent bulb would be the best
02:22
troy_s
Bertl: You want a D65 at best. A tungsten 3200. Or at worst, a Kino tungsten or daylight.
02:22
FergusL
of course tungsten if possible
02:22
FergusL
but isn't that really harder to find now ?
02:22
FergusL
in France at least it is
02:22
Bertl
sounds good, but I guess they will be expensive
02:22
troy_s
(CF fixtures spike erratically. Flip over a CD and see the unique refraction to see how non-uniform they are
02:23
FergusL
ha ok
02:23
troy_s
Compare against a tungsten source which will appear as a smear.
02:23
Bertl
that's what I suggested with the slit/prism
02:23
troy_s
Ideally you want a lamp close to the white balance of your destination. D65 in this case is a prudent choice.
02:24
FergusL
why is D65 the destination ? the sensor ? the screen ?
02:24
troy_s
Bertl: So nothing fancy. A daylight (5600 or 6500) film light will suffice.
02:24
FergusL
just an habit ?
02:24
troy_s
If not an HMI, then tungsten 3200 (or color temp up via CTB)
02:25
Bertl
okay, where would I get that?
02:25
FergusL
haven't had a lighting talk in monthgs
02:25
troy_s
FergusL: Because stretching the white point adversely impacts ranges of colors via metamerism.
02:26
troy_s
Bertl: Do you have an equipment rental house nearby?
02:26
troy_s
They will likely give you one for free.
02:26
troy_s
Ask them to color temp test a lamp and use it.
02:27
Bertl
equipment means film/photo equipment?
02:27
Bertl
note: I'm not a photo/camera guy :)
02:28
FergusL
yes, a cinema/video gear rental jouse
02:28
troy_s
You should be able to find a film rental house. Talk kind and I am 99% certain they will loan you a light for an eve.
02:29
troy_s
You do not need a large light. A pocket par would suffice.
02:29
Bertl
okay, so probably sebastian will know where to get one, I'll ask him when he is around tomorrow
02:29
troy_s
Just make sure the color temp of the light is noted.
02:29
FergusL
(and check for a replacement bulb)
02:29
Bertl
so the store should know the exact color temperature
02:30
troy_s
(not just blue / red kelvin either. Make sure you aren't getting bad magenta / green.)
02:30
troy_s
But even a ballpark set of test images is ideal.
02:30
troy_s
If you take a tungsten fixture (a redhead is 1k watts)
02:31
troy_s
Ask for some small shreds of 1/2 and full CTB.
02:31
Bertl
I knew redheads are hot, but 1k watt? :)
02:31
troy_s
(they will burn out quick so you will need to work briskly once you put it on the barn doors)
02:31
FergusL
( http://www.linkdelight.com/components/com_virtuemart/shop_image/product/98699dbc7e99c58ad9e1ac2bd7f7ceb8.jpg that's a redhead)
02:31
FergusL
but for the size of the setup 800W is huge overkill
02:32
troy_s
3200 (tungsten) plus half ctb should be abput 4300
02:32
Bertl
I'd say the 120W halogen lamp is already overkill
02:32
Bertl
but I think I will do the following for now:
02:33
FergusL
(HA, time for 3:30am breakfast !)
02:33
Bertl
I have a 100W incadescent bulb somewhere, and I have the 120W halogen lamp
02:33
troy_s
You need a movie tungsten fixture to assert 3200~
02:33
Bertl
I will simply try to distribute the light evenly for both, and take two snapshots
02:34
Bertl
that will give a first idea till we have proper lighting
02:34
troy_s
Full CTB on a tungsten fixture is ideally about 5600.
02:34
troy_s
Just use _one_ lamp
02:34
troy_s
near camer
02:34
troy_s
camera
02:34
Bertl
yes, of course
02:34
troy_s
small piece of diffusion and color on doors
02:34
troy_s
done
02:34
troy_s
super simple
02:34
Bertl
doors?
02:34
troy_s
Redhead is as small as it gets
02:35
troy_s
(open face)
02:35
troy_s
you can get smaller with a fresnel based fixture
02:35
Bertl
okay, please slow down :)
02:35
troy_s
but I am certain they will loan a redhead out for nothing.
02:35
Bertl
what does 'color on doors' mean?
02:35
troy_s
barn doors = metal doors that slide onto fixture to control the light spill
02:36
Bertl
ah, the flaps on the side of the lamp?
02:36
troy_s
you normally clip (wooden clothespegs) gel / diffusion to the doors
02:36
troy_s
yes
02:36
Bertl
okay, and 'color' means?
02:36
troy_s
it won't burn as fast with a distance
02:36
troy_s
hence "on the doors"
02:36
troy_s
color means transparent gels of color
02:37
troy_s
CTB = reference blue gel to cool down (increase color temp)
02:37
troy_s
each of reference level - 1/2 CTB takes 3200 to 4300 (off top of head)
02:37
Bertl
ah, I see, so you use that to 'condition' the spectrum, yes?
02:37
troy_s
full CTB takes 3200 to 5600
02:38
troy_s
they used to be for migrating tungsten to daylight or back from daylight (hmi) to tungsten
02:38
troy_s
to balance for various lights
02:38
troy_s
Sense?
02:38
Bertl
migrating means 'moving from one setting to the other'?
02:39
troy_s
Yes. In the old days there were only tungsten balanced stocks and daylight.
02:39
FergusL
yes, 3200K light source through a full CTB give 5600K
02:39
troy_s
So you had to balance for only one if you wanted an achromatic appearing source.
02:39
Bertl
tungsten is considered in-door light?
02:39
FergusL
(yes)
02:39
troy_s
Now with digital you can dial several whites.
02:40
troy_s
indoor lighting is tungsten generally. Night as well.
02:40
troy_s
(because you are dealing with primarily artificial light that normally ends up in the 3000~ range)
02:40
troy_s
anyways... it doesn' t matter too much.
02:41
troy_s
key is to know white point
02:41
troy_s
and get equal illumination
02:42
troy_s
Then you can profile that camera and figure out what the primaries are of the RGB and from that, the latitude and gamut.
02:42
Bertl
how well calibrated is a typical picture camera?
02:42
troy_s
Extremely.
02:42
Bertl
e.g. an EOS 10D, do we know the spectral response of that one for example?
02:42
troy_s
A device will always be somewhat unique
02:43
troy_s
but the transform is often baked in low level
02:43
Bertl
because it would be quite easy for me to take a picture of the chart with the camera under similar (maybe identical) lighting conditions
02:43
troy_s
Keeners profile their own camera and create either a matrix (again most of the range is linear so easy transform 3x3 matrix) or
02:44
troy_s
a 3D LUT
02:44
troy_s
Yep
02:44
troy_s
The net sum result is that you take the known values of the IT8 and figure out what the camera values are
02:44
troy_s
then calculate the transform that gets from A to B
02:45
troy_s
and in that transform you also learn how far out the primaries are.
02:45
FergusL
hahaha, my raw to EXR script has been taking a full CPU thread for about 2 minutes now... still working
02:45
Bertl
yes, but for a known light source
02:45
Bertl
so basically I need the spectral output of the light source
02:45
troy_s
which gives you an idea of the volume of color a device can render.
02:46
troy_s
Bertl: Movie lights are pretty damn close
02:46
troy_s
if you stick to HMI or tungsten
02:46
troy_s
(again, they _do_ drift (tungsten shifts as the bulb ages)
02:46
Bertl
pretty damn close to what? eachother? the perfect XXXK source?
02:46
troy_s
but by and large you can figure out where they are
02:47
troy_s
most houses have color meters and can tell you the kelvin and when they were tested.
02:47
troy_s
Yes
02:47
troy_s
HMI is always close to daylight (they test the heads)
02:47
troy_s
and tungsten is well... tungsten. Physical property of getting it hot (age caveat aside)
02:48
troy_s
And with that, I must bed.
02:49
troy_s
Bertl: Feel free to hit me via email.
02:49
Bertl
HMI is mercury arc, yes?
02:49
Bertl
troy_s: okay, thanks for the information, I appreciate it!
02:51
troy_s
Bertl: Yes. http://en.m.wikipedia.org/wiki/Hydrargyrum_medium-arc_iodide_lamp
02:52
troy_s
Bertl: But to be honest tungsten is probably a better choice if you can snag some CTB slices
02:52
Bertl
I just wonder because mercury lamps have a really weird spectral profile
02:52
troy_s
(having stills at various color temps is great)
02:52
troy_s
(ask the house if they can give you the kelvin values with half ctb and full ctb, then use the two)
02:52
Bertl
okidoki, will see what I can do
02:53
troy_s
Bertl: They are relatively smooth, not nearly as smooth as tungsten.
02:53
troy_s
A redhead with a bit of diffusion and CTB is for the win.
02:53
troy_s
And likely free.
02:53
troy_s
Most rental house peeps are very cool, and always looking to help out indies.
02:54
troy_s
Often a bottle of wine, some beer, or even a huge thanks can work miracles.
02:55
troy_s
If you treat them well, you can forge a good start to a long relationship (and trying to develop a camera mean you might want to try and keep them on the good side.) :)
02:55
FergusL
I can comment on this
02:55
Bertl
that's probably a job for sebastian then, he will know where to ask, thanks for all the input!
02:55
troy_s
Night folks.
02:55
Bertl
have a good one!
02:55
FergusL
night
02:55
troy_s
Bertl: The gamut is something I am keen to find out.
02:55
troy_s
Bertl: And latitude. Two massive things for shooters.
02:58
FergusL
I wonder what's a decent time for debayering a 4K image with VNG
02:59
FergusL
with my python script it's slightly above one second per line... and there are 3072 lines /D
03:02
Bertl
that sounds a little slow :)
03:02
FergusL
yes
03:02
FergusL
my implementation is silly too
03:05
FergusL
VNG is really high quality though
03:05
FergusL
it's not simple interpolation
03:08
FergusL
and it's all white, meaning something has gone wrong
03:09
Bertl
high quality white :)
03:10
FergusL
yes, if it takes 40 seconds to render a 400x400 white square, it must be high quality white
03:15
FergusL
actually is *is* working
03:16
FergusL
it's just that values aren't normalised
03:16
FergusL
Pixel 6 , 0 done.
03:16
FergusL
B: 321.0
03:16
FergusL
G: 349.375
03:16
FergusL
R: 375.75
03:18
FergusL
can I assume /4095 ?
03:18
FergusL
or /4096 ? as in value = calculated_val / 4095
03:18
FergusL
the other way round, 4095 / calc_val
03:19
Bertl
maybe, the range of the unshifted .raw16 values is 0x0000 - 0xFFF0
03:20
Bertl
when you shift it right by four, the range will be 0x000 - 0xFFF (so 0 - 4095)
03:20
FergusL
after being shifted, we get 12 bits of actual data, right ?
03:20
Bertl
yep
03:21
Bertl
so 1.0 ~ 4095 and 0.0 ~ 0
03:21
Bertl
but it doesn't matter if you take 4095 or 4096 :)
03:21
FergusL
oh god I'm lost, uh, give me a sec
03:22
Bertl
if you want to normalize the range to 0-1.0 you have to divide by 4095 (or 4096)
03:23
FergusL
that's what lost me
03:23
Bertl
depending on how you define the 1.0 value
03:23
FergusL
why did I put it before
03:23
Bertl
i.e. if you divide by 4095, you will reach 1.0
03:24
FergusL
yes, just tested
03:24
Bertl
if you divied by 4096 you will get the 'lower end' of a 1/4096th interval
03:24
FergusL
fuuuh... this kind of error means I'm don't do code enough !
03:24
Bertl
i.e. the center would be at V/4096 + 1/8192 :)
03:26
FergusL
hmmmm... it's certainly not a VNG debayered image
03:26
FergusL
but, it's not completely fucked
03:27
Bertl
I'd suggest to get a 'normal' RGB image with reasonable resolution and simply break it down into an RGGB bayer pattern for testing
03:28
FergusL
yes, I'll have to do that
03:28
Bertl
(by simply removing G and B from the R sites, R and B from the G sites and R and G from the B sites
03:28
Bertl
)
03:29
Bertl
no interpolation or shifting/overlapping and then simply convert the RGB into gray values
03:29
FergusL
HA, self_20ms isn't Y flipped
03:29
FergusL
!
03:29
FergusL
I can't tell because mine is indeed flipped
03:30
FergusL
I can tell*
03:30
Bertl
that's quite possible, the flip should be recorded in the sensor register dump
03:30
FergusL
mars_: ping ?
03:31
FergusL
http://www.pasteall.org/47018/python
03:31
FergusL
(I know it's a huge non-sense mess)
03:32
FergusL
hardcoded at some point where knowledge lacked
03:33
FergusL
I'm going to work with Lenna :D
03:35
FergusL
I should also forget python and do it in C++ now
03:39
FergusL
(http://en.wikipedia.org/wiki/Lenna)
04:25
Bertl
off to bed now ... have a good one everyone!
04:30
[1]Sasha
Night Bertl
04:33
dmj_nova
joined the channel
06:41
[1]Sasha
left the channel
07:38
Sasha_C
joined the channel
09:43
PhilippeJ
joined the channel
12:04
se6astian
joined the channel
12:05
se6astian
hello!
12:07
mars_
hi
12:19
PhilippeJ
left the channel
12:22
se6astian
how did the rest of the evening go yesterday?
12:54
Sasha_C
hey everyone
12:55
se6astian
hi sasha
12:55
se6astian
we received 4 donations yesterday!
12:55
Sasha_C
I saw. Very nice :)
12:56
Sasha_C
Let's keep building up this momentum
12:56
FergusL
yes I saw that se6astian, impressive !
12:56
FergusL
troy_s Bertl and I talked about the next tests
12:56
FergusL
especially regarding lighting
12:58
FergusL
you sent a color chart to Bertl, right ?
12:58
Sasha_C
Hey FergusL, I was watching that conversation earlier today.
12:58
Sasha_C
I'd never heard of Lenna before ;)
12:58
FergusL
yes
12:59
se6astian
yes I gave Bertl a chart yesterday
12:59
FergusL
you're both living close by ?
13:00
se6astian
yes :)
13:01
FergusL
ha cool
13:02
FergusL
so, troy_s suggested we get a cinema light fixture for a day in order to conduct some more tests
13:03
FergusL
to light the chart for exampel. an HMI/daylight balanced source would be the best but a tungsten-halogen would be fine with CTB (blue gel) on it
13:05
FergusL
do you know some cinema gear rental hosues ?
13:05
FergusL
houses*
13:09
se6astian
I have a kinoflo diva
13:09
se6astian
that should be pretty color accurate
13:10
Bertl
morning folks!
13:10
Sasha_C
morning Bertl
13:15
se6astian
morning!
13:16
se6astian
how long did you stay at the happylab? :)
13:17
Bertl
it was well after midnight, but we didn't get that much done
13:17
se6astian
oh no, why not?
13:17
Bertl
actually we didn't get anything done worth mentioning, but we had a very productive conversation
13:17
se6astian
:)
13:19
Bertl
well, the second milling drill bit died at basically the same location than the previous
13:20
Bertl
and after further investigation, we concluded that the cooling must be insufficient
13:20
Bertl
because it happens exactly when the metal gets really thin and cannot transport the heat away
13:21
Bertl
mars_ will give you the details when he finds the time
13:21
PhilippeJ
joined the channel
13:21
Bertl
and then we waited quite some time for the lasercutter, but around midnight we decided to do the cutting another day
13:22
se6astian
oh no!
13:22
Bertl
but don't worry, I'll take care of that at the end of the week
13:23
se6astian
great!
13:24
se6astian
maybe the cooling liquid was not hitting the mill tip properly? we tuned it down to only spray a little amount of cooling liquid... maybe that was the problem
13:25
Bertl
maybe, I didn't do any further investigations of the setup and/or results, but I can take a picture of the partially milled piece
13:28
se6astian
why not yes, so what was the fruitful converssation about?
13:29
Bertl
everything axiom :)
13:29
se6astian
anything else would have disappointed me ;)
13:30
Bertl
prototype and axiom development wise, from power supply to memory configurations
13:30
se6astian
very nice, I would love to hear a summary sometime soon
13:30
Bertl
(so mostly technical stuff)
13:31
Bertl
rest assured, you will :)
13:31
se6astian
I just contacted a developer from the apertus community outskirts about the bash scripts, he said he will get right to them
13:32
se6astian
and might come to IRC at some point, name: Jehan
13:33
Sasha_C
Sorry to interrupt, but we've just been featured again on Redshark: http://www.redsharknews.com/technology/item/1208-first-images-from-axiom-alpha-prototype-camera-from-apertus?utm_source=www.lwks.com+subscribers&utm_campaign=efa976005a-RSN_Nov05_2013&utm_medium=email&utm_term=0_079aaa3026-efa976005a-74476833
13:35
FergusL
cool se6astian, do you have a color meter ?
13:35
se6astian
great, they covered us in their weekly newsletter you mean right?
13:35
Sasha_C
yep
13:36
Bertl
funny how the wording changes ... now please tell me, who were the folks banging their head against a wall? *G*
13:37
Sasha_C
yeah I know. It's not the best reporting, but it's spreading the word for our cause
13:41
Bertl
yes, of course, but it is a desaster if you look at the details, and I'm not even talking technical details here :)
13:41
Bertl
note that I didn't expect anything else though
13:42
Sasha_C
Do you want us to contact the writer and request a redaction / correction to her article?
13:42
PhilippeJ
Hello guys !
13:43
Sasha_C
Greetings PhilippeJ
13:44
se6astian
unfortunately I have no color meter, I just got a light meter that I will forward to herbert when we meet the next time
13:44
PhilippeJ
Lossy DNG could save 5 times the space, interesting : http://helmersworks.com/journal/2013/03/digital-negatives-dng-with-lossy-compression-is-it-worth-it/
13:46
se6astian
last time I checked the only mode for lossy DNG compression was based on JPEG for bayer pattern and limited to 8bits
13:47
PhilippeJ
some docs here : http://www.pointsinfocus.com/learning/digital-darkroom/first-thoughts-on-lossy-dng/
13:53
Bertl
Sasha_C: nah, not worth it, I guess any news is good news ATM
13:53
Bertl
PhilippeJ: we don't want to go lossy
13:54
Bertl
you can always create a jpeg if you don't care about artefacts
13:54
PhilippeJ
as the guy said, "Whoever said storage is cheap, never had to buy and manage that storage" :-)
13:55
PhilippeJ
just that at some point, it might be interesting to have some support for light compression
13:56
PhilippeJ
what I don't understand is that it seems the compression is applied to debayered images
13:56
PhilippeJ
loosing most of the benefits of raw
14:01
PhilippeJ
Anyway, se6astian
14:01
PhilippeJ
how did the milling end up ?
14:02
FergusL
se6astian: Okay. For the chart test to be useful we need to know the color temperature of the lighting source as precisely as possible
14:04
FergusL
because with this test we're going to calibrate the white balance for the camera
14:04
PhilippeJ
FergusL
14:04
FergusL
(this is what I understood from troy_s explanations yesterday)
14:04
PhilippeJ
(sorry, strange irc client) you could use calibrated lights like kino or any cinema lights?
14:05
FergusL
yes
14:05
PhilippeJ
tungsteen / daylight
14:05
PhilippeJ
I don't really see any practical use for other temperatures, since it can be corrected in post
14:06
se6astian
PhilippeJ, bad the millhead broke twice and then mars and Bertl stopped
14:06
PhilippeJ
se6astian, too bad :-(
14:07
mars_
i gave bertl a 18% gray card yesterday, as reference
14:07
se6astian
great
14:07
FergusL
cool too !
14:07
se6astian
I have a kinoflo with daylight and tungsten tubes
14:07
se6astian
we can do the calibration with these
14:08
FergusL
yes, but will the "5500" or "5600K" marking on the tubes might be enough for such precise calibration ?
14:09
PhilippeJ
I don't think it's a problem
14:09
FergusL
-might
14:10
PhilippeJ
the image must be coherent, meaning, each pixel related to the others has the same calibration
14:10
PhilippeJ
color temperature, if slightly off, on the whole surface, is not a problem
14:10
PhilippeJ
you can adjust it in post
14:11
FergusL
that's not what I meant
14:11
PhilippeJ
sorry, then I misunderstood :-)
14:11
PhilippeJ
what are you trying to calibrate?
14:16
se6astian
I can rent a color temperature meter if we think its necessary
14:16
se6astian
but the kinoflo tubes are the most accurate light sources you can get
14:16
se6astian
in terms of CRI
14:17
FergusL
we're going to look at R G and B values for a known white, the known white is the one on the chart, we look at the R G and B value captured by the sensor. We then shift the values so that R = G = B, this is white balancing, at this point we know "how much" to shift from the values read by the sensor to get to <Temperature>
14:17
FergusL
Temperature being the known temp of the light source used
14:17
PhilippeJ
but you do the same shift for all the pixels, right?
14:18
FergusL
yes, kinos are reliable
14:18
FergusL
yes
14:18
PhilippeJ
so, if you are 1% off, it will be for the whole image, and will be easy to correct
14:19
FergusL
you might be 1% off, but you won't see it
14:19
FergusL
because you will look at white
14:19
PhilippeJ
which imho, means that kino or similar lighting will be good enough
14:19
FergusL
and white is white
14:21
FergusL
it's in the equation that the issue is : we have to start from a well known base. If we assume 5600K as a base colortemp from the kinos but that it is in fact 5587K (or even 5595K) (depending on age of tube or other factors), we will *constantly* have an unwanted 13 Kelvins shift
14:22
FergusL
wouldn't matter much in the end I agree, but I think it's worth the hassle to go that much in detail when we're that much into the heart of the camera :)
14:23
FergusL
(also I might be wrong somewhere, I'm waiting for Bertl's and troy_s's thoughts on this)
14:23
PhilippeJ
could you use another, calibrated camera to shoot under the same condition and use it as a temperature meter?
14:24
FergusL
I think it would work
14:24
FergusL
but we need to access the RAW of the other camera, the *real* raw data, the 10/12bits data for each sensel
14:25
FergusL
off to lunch, brb
14:26
PhilippeJ
:-) good luck! Interesting tutorial on camera calibration : http://staffwww.itn.liu.se/~karlu/div/howto/qpcard201_argyll.php
14:36
Sasha_C
left the channel
15:01
FergusL
Yes argyll is solid!
15:10
se6astian
left the channel
16:11
PhilippeJ
left the channel
16:33
troy_s
Tungsten is probably smoothest. Kino is ill advised. Too many gaps in spectrum.
16:36
FergusL
(hello troy_s)
16:37
troy_s
And no, a color temp meter to keep things simple. Again, the guys and gals at the house can measure the values.
16:41
FergusL
do I have the facts straight about the base colortemp and ?
18:06
troy_s
FergusL: There are basically three points white balance enters into things. 1) When shooting. In this case, tungsten is probably the only decent option due to full spectrum with no spikes.
18:07
troy_s
FergusL: 2) During a PCS conversion. If we use Argyll or any other ICC based software, there is always a D50 implied PCS, so a Bradford color transform happens from source photographed white to PCS.
18:07
troy_s
3) Destination white. Another Bradford here.
18:18
troy_s
Basically it would be along these lines:
18:18
troy_s
1) Show up at house for redhead lamp. Measure color temp using their meter. Note it.
18:19
troy_s
2) Slap half CTB on the doors, measure temp. Should be around 4300k. Note it.
18:19
troy_s
3) Slap full CTB on the doors, measure temp and note.
18:19
troy_s
4) Take lamp home, using diffusion and color as per tests above, shoot:
18:19
troy_s
4.1) One at native tungsten output.
18:20
troy_s
4.2) One with half CTB.
18:20
troy_s
4.3) One with full CTB.
18:21
troy_s
4.4) Another set if there are white balance settings on the sensor. One set per ballpark setting. Likely 3200, 4300, and 5500.
18:32
se6astian
joined the channel
18:40
FergusL
I'm lost during 2) and 3)
18:40
FergusL
you might need what you've missed, pasting log in /query se6astian
18:53
troy_s
FergusL: 2 and 3 are sort of part of an ICC based CM.
18:53
FergusL
what is a PCS conversion ?
18:53
troy_s
FergusL: No avoiding a PCS with ICCs.
18:53
FergusL
and a Bradford ?
18:54
troy_s
Profile Connection Space. Normally a CIE absolute like XYZ.
18:54
troy_s
There are several ways to move the white point through XYZ and transform it. XYZ scaling (worst - aka Wrong Von Kries), Von Kries, and Bradford
18:55
troy_s
Bradford is the accepted best of breed white balance transform currently.
18:55
FergusL
I think I see
18:55
FergusL
is CIE XYZ the one with unbound luminance ?
18:59
troy_s
XYZ is all unbounded. Think of it as an infinitely large cube rotated like a diamond. XY and Z grow out and up
19:00
troy_s
If you scale XY to xy, you have the "color" of the light. Y isolates luminance.
19:00
FergusL
got it
19:05
se6astian
Hi troy_s, I summarized what you said about the calibration process on the apertus wiki, how do you want to be credited? irc nick, full name, etc.?
19:07
se6astian
https://wiki.apertus.org/index.php?title=Camera_Color_Temperature_Calibration_Procedure
19:09
se6astian
Do you mind me summarizing what you said on the wiki?
19:18
FergusL
there's something wrong in my implementation of VNG, it's all greenish !
19:26
se6astian
troy_s, is the goal of the calibration for each color temperature to get 3 factors (one for each primary) or an entire curve for each primary mapped to luminosity?
19:26
philippej
joined the channel
19:27
philippej
changed nick to: philippejadin
19:57
troy_s
se6astian: No credit thanks. I just make it up.
19:57
troy_s
se6astian: A few samplings can help to test the results.
19:57
FergusL
troy_s: I'll just leave my questions here
19:57
FergusL
I'm trying to grasp more precisely the process of calibrating and more generally white balancing
19:58
FergusL
afaik, white balancing is being at "pure white", there's an exact term for this : equi-something white
19:59
troy_s
FergusL: No such thing as whitr
19:59
troy_s
it is merely the origin - a point of achromatic.
20:00
FergusL
so basically R=G=B=W, is W decided before-hand as an absolute value ? or does white-balancing averages ? like R=10, G=8 and B = 6, W = 8
20:01
troy_s
se6astian: Ultimately, after profiling, you get a pretty good idea as to the primaries of the camera - how the triangle base (and the 3D volume) of the color gamut of capture is.
20:01
troy_s
(as well as latitude and things imagers care about like "How many stops does it hold?")
20:02
troy_s
FergusL: Again, white does not exist.
20:02
troy_s
FergusL: Understood that yet?
20:02
troy_s
it is merely a point of reference
20:02
FergusL
are obsessed about*
20:03
troy_s
Think of it as a map. If I say "Hawaii is six miles south" and you don't know your starting point, all of your map is wrong.
20:03
troy_s
Similar to white point.
20:03
troy_s
You must know it precisely.
20:03
[1]se6astian
joined the channel
20:04
FergusL
yes, I just realised that
20:06
se6astian
left the channel
20:06
[1]se6astian
changed nick to: se6astian
20:06
FergusL
the starting point is the "base" temp of the sensor ?
20:06
FergusL
so that 3200K is R = baseR + R, Gbase + R etc... calculated as distances
20:07
FergusL
Gbase + G*
20:07
troy_s
No. But think map again.
20:08
troy_s
If your and I know that a swatch is of xy chromaticity
20:08
troy_s
and we take a camera shot of it
20:08
troy_s
if we don't know where white is (the light illuminating the swatch), how do we know what color the swatch is?
20:16
[1]se6astian
joined the channel
20:18
se6astian
left the channel
20:18
[1]se6astian
changed nick to: se6astian
20:28
philippejadin
left the channel
20:29
[1]se6astian
joined the channel
20:32
se6astian
left the channel
20:32
[1]se6astian
changed nick to: se6astian
20:34
philippejadin
joined the channel
20:38
Bertl
so, finally done with useless stuff for today
20:39
philippejadin
depending on your definition of useless, I guess it's a good news :-)
20:39
se6astian
hurray, hail the tax minister :)
20:41
Bertl
yes, was mostly busy doing my tax declarations today
20:42
Bertl
(so I think it can be considered useless work :)
20:47
philippejadin
:-)
20:47
philippejadin
I'm looking for existing dng writters libraries
20:47
philippejadin
it's a huge mess, or maybe I don't know enough to find my way. Anyone has an idea?
20:49
Bertl
gabe is working on a dng writer and has written a dng reader AFAIK, probably best to ask him
20:50
FergusL
what software do you want to use with DNGs ?
20:51
FergusL
troy_s: we don't, so we need a reference !
20:52
Bertl
troy_s: so after finding the perfect calibration process :) does it make any sense to take a picture of the IT8 chart with incandescent and halogen lamps or should I skip that till we have high quality light sources?
20:53
Bertl
s/skip/postpone/
20:59
FergusL
I'd say simply skipping
21:01
FergusL
troy_s: I remember using ENG cameras at school, there was this WB button, when you pressed it, the white became white
21:01
FergusL
and all other pixels got a color by reference to this color
21:01
philippejadin
In current state the more images we get the better, and as soon as we can play with them, we'll get a better idea
21:02
Bertl
well, the thing is, ATM we have exactly one prototype with a partially defective sensor
21:03
Bertl
and doing a setup to get a proper picture (lighting, position, sharpness) is somewhat complicated
21:03
Bertl
i.e. if the images are of no use, it's better to spend the time on coding stuff :)
21:03
philippejadin
any image will do to judge based on very subjective aspects :-)
21:03
philippejadin
definitely
21:04
FergusL
that's why I'm being helpful: coding
21:04
FergusL
(kidding)
21:04
FergusL
dinner time, later
21:04
Bertl
but if troy_s (or somebody here) says, he can do something useful with the images, I'd consider it worth spending the time
21:05
philippejadin
I think with the latest "self portrait" images, there is already a lot to do
21:05
FergusL
troy_s: I remember using ENG cameras at school, there was this WB button, when you pressed it, the white became white and all other pixels got a color by reference to this color. My question is what happened to the color, was Green Red and Blue changed ? was it left as is but only the other colors changed accordingly ?
21:05
philippejadin
what are the next steps coding-wise?
21:06
Bertl
getting the image from memory to the HDMI output
21:06
philippejadin
FergusL: here is a nice explnation : http://www.cambridgeincolour.com/tutorials/white-balance.htm
21:07
philippejadin
Bertl: I guess not an easy task at all
21:08
Bertl
you can bang your head against the wall while I implement it ... after all, the news people write the truth, don't they?
21:09
philippejadin
:-)
21:11
FergusL
left the channel
21:14
FergusL
joined the channel
21:31
philippejadin
Time to leave for me. See you later everyone !
21:31
Bertl
cya
21:36
dmj_nova1
joined the channel
21:37
philippejadin
left the channel
21:38
dmj_nova
left the channel
21:42
se6astian
there is a distinct increase of traffic from facebook to the apertus website today
21:42
se6astian
I wonder whats going on there
21:44
Bertl
do you have specific pages which transitioned to the site?
21:54
se6astian
not from "inside" facebook
21:55
se6astian
as they mask that out, it always comes from facebook.com
22:02
se6astian
Digitalbolex Joe Rubinstein sees the times are right for small companies building cameras: Iím not saying the big guys wonít be successful, but the Axiom project could very well be successful too.
22:02
se6astian
http://nofilmschool.com/2013/11/first-images-apertus-super-35mm-camera-prototype-axiom-alpha/
22:46
se6astian
time for bed
22:46
se6astian
good night!
22:47
se6astian
left the channel