Current Server Time: 20:21 (Central Europe)

#apertus IRC Channel Logs

2015/02/13

Timezone: UTC


00:22
fsteinel_
joined the channel
00:25
fsteinel
left the channel
01:11
intracube
left the channel
03:43
fsteinel_
changed nick to: fsteinel
04:28
Bertl
off to bed now ... have a good one everyone!
04:28
Bertl
changed nick to: Bertl_zZ
07:09
se6astian|away
changed nick to: se6astian
07:09
se6astian
good mroning
07:09
se6astian
morning
08:20
Andrej74
joined the channel
08:24
Andrej74
left the channel
08:39
Bertl_zZ
changed nick to: Bertl
08:39
Bertl
morning folks!
11:16
jucar
left the channel
11:24
intracube
joined the channel
12:31
ItsMeLenny
joined the channel
13:39
rhavan
left the channel
13:40
rhavan
joined the channel
14:01
ItsMeLenny
left the channel
14:06
Bertl
left the channel
14:14
Bertl
joined the channel
15:03
se6astian
changed nick to: se6astian|away
15:18
Juicyfruit
left the channel
15:49
Bertl
off for a nap ... bbl
15:49
Bertl
changed nick to: Bertl_zZ
15:57
se6astian|away
changed nick to: se6astian
16:40
danieel
left the channel
17:01
aombk2
changed nick to: aombk
17:03
aombk
sometimes this channel turns into apertus team bed visiting hours announcing board
17:25
danieel
joined the channel
17:48
g3gg0
joined the channel
17:54
Juicyfruit
joined the channel
19:01
se6astian
aombk: true :)
19:01
se6astian
I ll try to slip out more silently from now on
19:36
aombk
no no, dont get me wrong, its great!
19:56
se6astian
ah, then I ll also report about breakfast, lunch, and teeth brush events :D
20:06
aombk
even better!
20:10
lab-bot
sebastian closed T278: Native and maximum ISO as "Resolved". http://lab.apertus.org/T278
20:11
lab-bot
sebastian closed T279: Slow Motion as "Resolved". http://lab.apertus.org/T279
20:16
hozer
left the channel
20:17
hozer
joined the channel
20:24
g3gg0_
joined the channel
20:24
g3gg0_
left the channel
20:25
troy_s
se6astian: Thank goodness there is a Q&A section now.
20:27
se6astian
yes, the "ponder" module in phabricator works better than expected :)
20:27
g3gg0
left the channel
20:29
g3gg0
joined the channel
20:29
g3gg0
left the channel
20:31
troy_s
se6astian: Thank goodness. Thank goodness. Now maybe some of the ... uh ... chunderers can move there.
20:32
g3gg0_
joined the channel
20:34
g3gg0_
left the channel
20:38
g3gg0
joined the channel
20:47
jucar
joined the channel
20:50
jucar
left the channel
20:55
jucar
joined the channel
22:07
se6astian
time for the proper teeth brush
22:07
se6astian
and then right into bed
22:07
se6astian
2 pillows
22:08
se6astian
2 blankets
22:08
se6astian
good night
22:08
se6astian
changed nick to: se6astian|away
22:12
Bertl_zZ
changed nick to: Bertl
22:12
Bertl
back now ...
22:12
Bertl
is making some tea ...
22:12
troy_s
Bertl: aombk wants more status updates.
22:12
Bertl
is working on that ...
22:14
troy_s
Bertl: struct Tea {enum type, float minutes_steeped, float milliliters, float mug_color[3] }
22:14
Bertl
hmm, that needs some improvements, I use a clear glass :)
22:15
troy_s
Bertl: float mug_color rgba[4]
22:15
Bertl
much better, premultiplied I presume?
22:15
troy_s
Bertl: And you could get really fancy with something like [6.0, 0.0, 0.0, 0.0]
22:15
troy_s
Yes of course.
22:15
troy_s
Otherwise you couldn't have a mug that emits lovely red light.
22:16
troy_s
(maybe when hot.)
22:16
troy_s
Damn.
22:16
Bertl
I would suggest rgbia[5] then
22:16
troy_s
Amend the struct for a float temperature_Celsius
22:16
troy_s
???
22:16
troy_s
What do we need the fifth for?
22:16
troy_s
Sheesh.
22:16
Bertl
infrared :)
22:17
troy_s
Just use a different colorspace and leverage the imaginaries.
22:17
troy_s
:P
22:17
intracube
filtered vs tap water?
22:17
troy_s
Oh I like it.
22:17
troy_s
Bool or an enum for filter type?
22:17
intracube
yus
22:17
intracube
:)
22:17
Bertl
we have good tap water here in europe, no need for filtered or bottled water
22:18
intracube
Bertl: speak for yourselves! the water in the UK can be... variable
22:18
Bertl
okay, let's say austria then :)
22:18
Bertl
benefit from the alps I guess
22:19
intracube
heh
22:19
Bertl
which is a kind of filtration system :)
22:20
troy_s
I'll take your tap water and raise it four hundred. (British Columbia. :P )
22:20
troy_s
intracube: UGH. How the hell do you begin to deal with "fixing" things when you have folks posting this sort of rubbish? https://developer.blender.org/T43405
22:21
troy_s
coffee.
22:21
Bertl
splendor sine occasu
22:22
troy_s
intracube: I'll give you an A in colorknowledge if you can spot his two errors.
22:28
comradekingu
i guess colours dont saturate evenly, and that they stretch unevenly on the colourscale too
22:32
comradekingu
id say brightness is another scale entirelly, i dont understand the question
22:32
troy_s
comradekingu: More about his idea of values in RGB in relation to what they mean.
22:32
troy_s
comradekingu: That's one of the issues. He makes two key errors:
22:33
troy_s
1) That RGB is somehow a meaningful model. It isn't. It isn't even a colorspace. It's a color model that holds arbitrary and infinite color spaces.
22:33
troy_s
2) That somehow a value of RGB relates to "What someone wants as an artist."
22:33
troy_s
and a third
22:34
comradekingu
there is a summation error going on
22:34
troy_s
3) That 0.6 means _anything_ on either side of the "What a human is thinking" and "What it means in RGB". For example, the luminance weights for sRGB are about 21-ish% red, 72-ish% green, and 7%ish blue
22:35
troy_s
So his assumption that 'brightness' of 80% should be 0.8 is... well whack.
22:35
troy_s
*sigh*
22:35
troy_s
comradekingu: I'm getting more and more weary tackling color issues. It's just a cascading mess of bad knowledge compounded under the weight of bad knowledge and bad education.
22:35
intracube
troy_s: yep, it's not such a good idea to get drawn into specific pixel values
22:36
troy_s
(and broken applications that have legacy / historical / nasty ass hacks / wrongs / etc.)
22:36
intracube
I wonder if he's noticed that perceptual mid-grey isn't at 0.5
22:36
troy_s
intracube: It's really hard to even get into the bug discussion. I chimed in on the color wheel bug only to have a crapload of folks speak first.
22:36
troy_s
LOL
22:36
troy_s
I hereby coin the phrase MKF discussions. Mash Keyboard First.
22:38
comradekingu
if H=0 S=0 V=0.604 relatess to 0.8,0.8,0.8 its only fair that if H=0 S=0 V=0.8 is 0.9,0.9,0.9. it scales
22:38
troy_s
comradekingu: The hard part is that as "Simple" as that person's insight was (and as Simple as their Solution was) is actually about three layers of complexity, and that doesn't even cover the surface of the issues.
22:38
troy_s
LOL
22:38
intracube
but there are still rough areas in the way blender translates the internal 'true linear' values for the end user
22:39
intracube
through the UI colorpickers, etc
22:39
intracube
troy_s: and RGB curves in the comp are still not really fit for purpose, IMO
22:39
troy_s
Technically HSV wheels are a nightmare from a color management position as the V is really never going to be colorspace accurate luminance via OCIO.
22:39
troy_s
intracube: I've worked (and am working on the wheel) and there's no real "right" way. Right now my goal is:
22:40
comradekingu
i usually dont have the audacity in me to question the software i use with "this is for sure error", gotta admire that
22:40
troy_s
1) Have an artist selectable transfer curve via groupbox (Log, sRGB, 709, linear, CineonLog, etc.)
22:40
comradekingu
nono "it is for sure error"
22:40
troy_s
2) Have the circular part and the gradient display the transfer curve selected.
22:40
comradekingu
nice idea
22:40
troy_s
Of course, I'm sure there are a godawful number of peeps that will go "WHAT IS ALL THIS WHY THE COMPLEXITY WHY WHY WHY WHY"
22:41
troy_s
(see this nasty thread of nightmare... https://developer.blender.org/T41287)
22:41
comradekingu
but then you have just brought it down to selection bias, those people will discredit said artist for not picking evenly
22:41
troy_s
comradekingu: The real issue is that the personal bias is all over color bug reports.
22:42
troy_s
comradekingu: "I should be able to pick E8E8E8 and get xxxx.
22:42
troy_s
Which is just junk education.
22:42
comradekingu
nvm, i thought you were going to have an actual artist map colours to an arbitrary colourwheel in each colourspace
22:42
troy_s
And then in that report for example, "Have an sRGB picker"
22:42
troy_s
which is like... good god.
22:42
troy_s
comradekingu: That _kind_ of happens in a color managed system.
22:43
troy_s
comradekingu: For example, if you try the marciewhacked.exr - the primaries are rotated. So Marcie looks deadly wrong with good old purple alien looking skin.
22:43
intracube
troy_s: is quite late for my brain to comprehend, but I look forward to the improvements to the picker
22:43
troy_s
comradekingu: So based on your chosen colorspace for working in, the primaries can be arbitrary (and this is especially relevant for cameras where each camera sensor has its own color primaries response.)
22:43
troy_s
intracube: You might be waiting a while.
22:44
troy_s
Even if I patch it, the grumpy Russian would need to commit it.
22:44
intracube
can be patient
22:44
intracube
heh
22:45
troy_s
Biggest problem? My OpenGL-Fu sucks.
22:45
troy_s
I can't think in terms of oldschool triangles and cruft in the OpenGL system.
22:46
troy_s
Looking forward to GLNext.
22:51
intracube
troy_s: are you saying that it should be possible to switch the colour picker to sRGB mode and that R G B all at 0.5 should give perceptual mid grey?
22:51
intracube
or set linear, and have values appear as they do now
22:51
troy_s
yes.
22:52
troy_s
Well not really
22:52
troy_s
because middle grey doesn't live at 0.5 in sRGB but you get the point.
22:53
comradekingu
so if the colour picker is a rectangle the corner cases are interpolated values. Less dynamic range on some parts of the rectangle?
22:53
intracube
well, slightly off 0.5 (not exactly at 128 8but
22:53
intracube
*bit
22:53
comradekingu
and im thinking there is another level where you dont percieve colours evenly
22:54
troy_s
comradekingu: That's another great point that is often missed.
22:54
intracube
troy_s: wouldn't the option be better in the prefs or CM settings?
22:54
troy_s
comradekingu: perception isn't linear etc. so the model you pick, when laid out in a nicely symmetrical linear sort of UI element, wastes portions of values etc.
22:54
troy_s
intracube: So let's see...
22:54
intracube
rather than be on each individual picker
22:54
troy_s
intracube: You are painting a quickie texture and you want to place a middle grey display referred value of reddish somewhere...d one.
22:54
troy_s
intracube: Now you flip to the bump map
22:55
comradekingu
i think if the picker was shaped in the form of a complete colourwheel maybe it would be more obvious what happened when you switched colourspaces
22:55
comradekingu
if that works at all
22:55
troy_s
intracube: And want to paint the height to be about 50%
22:55
troy_s
So... Open up user Preferences.
22:55
troy_s
Find the tab
22:55
intracube
eh, yeah. I see the issue
22:55
troy_s
Flip the colorspace to linear.
22:55
troy_s
flip back
22:55
troy_s
:)
22:55
troy_s
comradekingu: Explain?
22:56
comradekingu
can this be the colour picker https://upload.wikimedia.org/wikipedia/commons/3/37/Colorspace.png ?
22:56
troy_s
comradekingu: There's a pretty massive disconnect with the shapes we see (both linearized and non linearized and more perceptual (See Macadams)) and shapes we can pick from.
22:56
troy_s
So there's no real win.
22:56
troy_s
comradekingu: That doesn't work either for much of things... great for picking values as the thing is a linear layout
22:56
troy_s
But we don't see color uniformly either.
22:57
g3gg0
left the channel
22:57
troy_s
comradekingu: http://upload.wikimedia.org/wikipedia/commons/f/f4/CIExy1931_MacAdam.png
22:57
troy_s
MacAdam's ellipses
22:57
comradekingu
well the user could make their own colourpicker by first doing a test of which colours they can work out
22:58
troy_s
comradekingu: A color within those ellipses has a Just Noticeable Difference of < 1.0 (as in a standard observe can't tell the difference easily without more context.)
22:59
comradekingu
i have enough trouble accepting that pixels are square
22:59
Bertl
they aren't :)
22:59
Bertl
it's a myth :)