Current Server Time: 07:38 (Central Europe)

#apertus IRC Channel Logs

2014/01/13

Timezone: UTC


01:53
Bertl
off to bed now ... have a good one everyone!
05:35
jucar
joined the channel
06:15
jucar
left the channel
08:32
se6astian
joined the channel
08:32
se6astian
good morning
09:28
guesst
hi
09:28
guesst
the it8 calibration will be tricky :)
09:29
guesst
one has to decide which patches are correct and which are out of the sensor gamut, therefore corrupted
09:33
se6astian
hi guesst, did you notice your sensor can not cover the entire gamut?
09:34
guesst
i have noticet that the color target has colors which are outside of the sRGB gamut
09:34
se6astian
I see
09:34
guesst
i guess that there are colors which are not sensible by the sensor
09:34
guesst
(not saying that the sensor is sRGB, but it can happen)
10:15
jucar
joined the channel
10:19
jucar
left the channel
11:15
jucar
joined the channel
12:59
ThatCantBe
left the channel
13:45
FergusL
guesst: hm... I don't understand why to do that
13:46
FergusL
I mean, why ruling out colors from the calibration
13:46
FergusL
if that's what you mean
13:59
troy_s_
joined the channel
14:00
troy_s
left the channel
14:16
guesst
you have two colors: A and B, A falls to gamut of sensor, B falls outside the gamut of the sensor. Both of these are seen by the sensor as color C. You want to do a C->A conversion, rather than a C->B, since A is the correct color, B is imaginary and seen falsely
14:17
guesst
simply keep the valid values correct and the falses leave corrupted
14:49
FergusL
I think I see guesst
14:49
FergusL
but isn't that common ?
14:50
FergusL
every sensor has this kind of spot I guess where it blends two colors into one, actually it blends every out-of-gamut color with one that is on the outside limit of the gamut ?
14:52
guesst
not sure how it happens in reality
14:54
FergusL
actually... there are patches of the it8 that are out of gamut ?
14:54
guesst
depends what color space you refer to :)
14:56
FergusL
haha true yes, what I said doesn't make sense...
14:56
FergusL
I mean, patches that are rendered as the same color ?
16:25
FergusL
left the channel
16:31
Bertl
morning everyone!
16:35
FergusL
joined the channel
16:44
philippejadin
joined the channel
16:53
se6astian
good evening :)
16:58
philippejadin
left the channel
17:30
FergusL
hi se6astian, do you have the cmv datasheet handy ?
17:31
FergusL
you left a todo in the C code and a ? about the exp time in the comments (in the metadata reader)
17:31
FergusL
and I found a reference in the datasheet but you may know it already
17:37
FergusL
5.1 page 30
17:38
se6astian
its on github
18:04
FergusL
what's on github ? the datasheet ? yes
18:11
philippejadin
joined the channel
18:12
philippejadin
left the channel
18:13
se6astian
yes
18:26
FergusL
yes I know, just saying I found the formula for calculating exposure time because there's a TODO in the C code mentioning it's missing
18:28
Bertl
which means 'go ahead, and implement it correctly!' :)
18:32
FergusL
certainly sir
18:33
FergusL
if it were just me though I'll reimplement the reader as a .py one :p
20:34
se6astian
What would be your primary application for Axiom? Vote now: https://www.apertus.org/node/272
20:54
FergusL
A voté !
21:02
guesst
so much choices naturally mean - we do not know shit about what our product is good at :)
21:02
guesst
sorry for the language
21:04
Bertl
like smart phones, which can only be used to call somebody :)
21:05
guesst
not it seems like we can play angry birds on axiom :)
21:06
Bertl
why not, should be doable ...
21:07
guesst
well.. usually it goes like... can we install linux on it? :) but that will be, right? (or have you decided to go simpler mcu power?)
21:09
Bertl
you will be able to run whatever you want on the hardware (as long as you can get it running)
21:10
Bertl
naturally that includes linux, bsd, and other OS or variations like android
21:10
guesst
not sure if I would care about androit.. it is overrated
21:45
FergusL
Bertl: does that ring a bell ? LVDS_CLK_P/N
21:46
Bertl
it is the lvds clock input, in our case 300MHz
21:46
FergusL
won't change in any case ?
21:47
Bertl
it might change if we decide to do slower transfers, so a #define should be fine
21:47
FergusL
Good, thanks
21:48
FergusL
Exposure time (ns) = [(Exp_time-1) x (reg85+1) + 1701] x LVDS_CLK_P/N-period x #bits
21:49
Bertl
correct
21:49
se6astian
time for bed
21:49
se6astian
good night!
21:49
se6astian
left the channel
21:49
FergusL
use of x instead of *, so pro-ish !
21:53
FergusL
ha... period there means one period of the said clock...
22:35
ThatCantBe
joined the channel
22:59
jucar
left the channel