Current Server Time: 20:28 (Central Europe)

#apertus IRC Channel Logs

2016/05/22

Timezone: UTC


00:27
Spirit532
left the channel
01:17
Bertl
off to bed now ... have a good one everyone!
01:17
Bertl
changed nick to: Bertl_zZ
01:58
ItsMeLenny
joined the channel
02:19
jucar
joined the channel
02:42
Gabe
left the channel
02:47
John-K
joined the channel
02:53
John-K
left the channel
04:21
mithro
Bertl_zZ / se6astian|away: How big is the AXIOM firmware?
04:45
jucar
left the channel
07:06
fredy
left the channel
07:09
fredy
joined the channel
07:38
toxitobi
joined the channel
07:58
niemand
joined the channel
08:10
Bertl_zZ
changed nick to: Bertl
08:10
Bertl
morning folks!
08:11
Bertl
mithro: define AXIOM firmware
08:17
Bertl
btw, I finally finished the triple mini DP plugin .. how is your progress on the opsis side?
09:12
Guillaume_Chooks
joined the channel
09:40
se6astian|away
changed nick to: se6astian
09:42
se6astian
good morning
09:42
Bertl
morning se6astian!
09:42
se6astian
mithro: we use 4 or 8gb micro sd cards currently
09:43
se6astian
how full they are with arch linux herbert will be able to tell
10:43
Spirit532
joined the channel
11:16
mithro
Bertl: please keep bugging me
11:17
danieel
joined the channel
11:19
Bertl
buggs mithro
11:20
Bertl
we will test the 3xmDP module in the next few days and if it works as expected, we'll probably send you one as well
11:20
Bertl
(or maybe two, if you get the AXIOM Beta adapter done before the 3xDP module :)
11:22
mithro
The axiom beta adapter is on the list first
11:23
Bertl
okay, so preferably two modules then
11:24
mithro
That should be ready, but I haven't had the time to finish the verification
11:24
Bertl
I can probably skip the power supply for those, as it won't be needed for the link
11:26
mithro
I still need to find a board designer that is willing to work on tofe boards for me.
11:27
Bertl
what do you pay per design or hour? :)
11:28
mithro
Not enough :-P
11:28
Bertl
maybe that's the reason ...
11:30
mithro
Well, work pays me pretty well but I need to fund my fpga developer and buy expensive axiom stuff too ;-)
11:30
Bertl
i.c.
11:31
toxitobi
left the channel
12:52
se6astian
gotta go
12:52
se6astian
changed nick to: se6astian|away
13:48
Bertl
John_K: regarding the bayer order, you probably want to make that an option and not simply 'change' it, because when you flip the sensor image vertically (via sensor reg) then you will get the wrong pattern again
14:12
Bertl
off for now ... bbl
14:12
Bertl
changed nick to: Bertl_oO
14:54
jucar
joined the channel
15:43
jucar
left the channel
16:29
ItsMeLenny
left the channel
17:20
se6astian|away
changed nick to: se6astian
18:31
RexOrCine
left the channel
18:46
niemand
left the channel
20:52
troy_s
joined the channel
20:52
troy_s
Greets @Bertl_oO
20:52
troy_s
Hope things have been well for you.
20:59
Bertl_oO
hello troy_s! LTNS!
21:00
Bertl_oO
things have been mostly busy, but I can't complain ...
21:00
Bertl_oO
eveything fine on your end?
21:01
troy_s
All good. Busy.
21:01
troy_s
Been making progress?
21:03
Bertl_oO
I guess so, the first bunch of Early Betas have arrived at their new owners
21:03
Bertl_oO
the experimental 4k raw got good publicity
21:05
troy_s
Bertl_oO: How has that been going?
21:07
Bertl_oO
the 4k raw?
21:10
troy_s
Yes.
21:11
Bertl_oO
there is an article describing the process
21:11
Bertl_oO
https://www.apertus.org/axiom-beta-uhd-raw-mode-explained-article-may-2016
21:11
troy_s
I read someone talking about a "magenta cast" not too long ago and I cringed at the sadly missing knowledge blocks.
21:12
se6astian
hi troy_s
21:13
se6astian
good to see you back here!
21:13
se6astian
have to leave soon though
21:13
troy_s
Greets se6astian, hope you are well.
21:13
troy_s
Any luck in giving up on DNG yet?
21:13
troy_s
:)
21:13
troy_s
KIDDING!
21:14
troy_s
Bertl_oO: I've found quite a few interesting bits of research that will interest you very likely.
21:14
troy_s
First, DolbyPQ, which is a technique for quantization across bits. Well worth researching. Implementation in the ACES development tree with the CTLs.
21:15
troy_s
A very interesting PDF on scene referred versus display referred transforms in video games of all things. Excellent material covering desaturation / crosstalk as well. http://gpuopen.com/wp-content/uploads/2016/03/GdcVdrLottes.pdf
21:15
troy_s
And some other interesting Tweets recently that are of relevance and interest:
21:16
troy_s
https://twitter.com/NateOrloff/status/734183388727414785
21:16
troy_s
(On the subject of EXR conforming.)
21:16
troy_s
and
21:16
troy_s
https://twitter.com/hpduiker/status/734460111935012865
21:17
Bertl_oO
troy_s: any URL for the DolbyPQ?
21:17
troy_s
(HPD is a big brainer that designed the ACES transforms package. What is interesting here is that the Academy's stance is to keep colour grades in the scene referred domain, and the tonemaps on the view transforms.
21:17
troy_s
Bertl_oO: There's two sources I've seen in code, the ACES being the most reliable.
21:17
Guillaume_Chooks
left the channel
21:18
troy_s
It's fascinating stuff though. Covers up to 10000 nit range, which in the grand schemes of colour and its management, the elephant in the room has always been emission levels of the display.
21:19
troy_s
Quite an interesting post from HPD as it makes perfect sense.
21:20
troy_s
Also interesting in terms of pipelines for software, as it uniquely splits some of the grading decisions between the scenic data and the view.
21:20
se6astian
very nice to see things moving so well
21:20
se6astian
off to bed now though
21:20
se6astian
good night
21:20
troy_s
Night se6astian
21:20
se6astian
changed nick to: se6astian|away
21:21
troy_s
Bertl_oO: The ACES view transforms have DolbyPQ variants which are the DolbyPQ adjustments with a bezier weighted knot set of curves for 48 nit, 1000 nit, and 2000 nit.
21:23
troy_s
You might also dig this presentation by Alex Fry on dealing with scene referred manipulations in post production. It specifically deals with ACES, but the concepts are all of the same issues faced with cameras and such.
21:24
troy_s
https://www.youtube.com/watch?v=vKtF2S7WEv0
21:24
troy_s
(Although the more I read about ACES and work through the CTLs, the more I'm blown away by how much experience and science is baked into it.)
21:25
troy_s
(Smartest thing for Apertus would be to offer an ACES path immediately.)
21:25
troy_s
(Very easy to implement as an IDT once you have decent characterizations down. If you crawl through the Arri vendor supplied bits in the repository for 1.0.2, you will see a per-EI set of tables for the LogC encodes.)
21:33
troy_s
Egads. Looks like a bit of chuck numbers at wall and hope approach.
21:36
Bertl_oO
I have no idea about ACES
21:36
troy_s
It's not complex.
21:36
troy_s
Just a standard set of primaries, a protocol, a set of naming conventions, file formats etc.
21:36
troy_s
All based on existing standards (the file format is ADX which is a DPX with very strict details on what is in it and what it means)
21:37
Bertl_oO
is there something to work with on Linux?
21:37
troy_s
All of it works on any platform.
21:37
Bertl_oO
urls?
21:37
troy_s
The entire official ACES LUTs etc., are all OCIO.
21:37
troy_s
Here's the official 1.0.2, which gets pulled into the OpenColorIO-Configs repository.
21:37
troy_s
https://github.com/hpd/OpenColorIO-Configs/tree/master/aces_1.0.2
21:37
troy_s
That's the "end user" LUT set.
21:38
troy_s
The generating bits are also in there, including Python.
21:38
troy_s
The aces-ctl transforms and such are all here:
21:38
troy_s
https://github.com/ampas/aces-dev
21:38
troy_s
The specifications and such are all SMPTE and such and they can be found here:
21:39
troy_s
https://github.com/ampas/aces-dev/tree/master/documents
21:39
arpu
left the channel
21:39
troy_s
Bottom line, is that when you don't have a clue, ACES takes all the guesswork out of things.
21:39
troy_s
It provides both an archival full spectral locus container (ACES proper)
21:40
troy_s
and a grading / cg set of constrained primaries (long story as to why you want that) ACEScg / ACEScc which are the same, but the ACEScc has a transfer curve specification.
21:45
Bertl_oO
sounds good, maybe our color science folks can take a closer look at it ...
21:53
arpu
joined the channel
22:14
arpu
left the channel
22:16
arpu
joined the channel
22:27
troy_s
left the channel
22:57
John-K
joined the channel
22:58
John-K
@Bertl_oO: Hrmm, it seems to be happening a lot. When you say there are mechanisms in firmware - do you mean in VHDL or in cmv_snap3?
22:59
Bertl_oO
what happens a lot?
23:10
Bertl_oO
anyway, off to bed now ... have to get up early ...
23:10
Bertl_oO
changed nick to: Bertl_zZ
23:23
Spirit532
left the channel