Current Server Time: 19:49 (Central Europe)

#apertus IRC Channel Logs

2013/11/16

Timezone: UTC


23:05
Sasha_C
joined the channel
23:46
gcolburn
joined the channel
23:49
FergusL
rexbron: the debayering script ?
23:49
FergusL
It's utterly slow... Has yet to fully debayer a 4k raw picture
00:39
gcolburn
left the channel
02:11
dmj_nova
Bertl: think the prototype would be usable by say April?
02:12
dmj_nova
FergusL: How long has the debayering script taken?
02:16
Bertl
dmj_nova: you mean the axiom prototype? or axiom alpha?
02:17
dmj_nova
actually I'm not entirely particular on that
02:17
dmj_nova
But something functional and fairly compelling.
02:17
dmj_nova
that can act like a camera and output quality images to a PC
02:18
Bertl
well, the axiom alpha should be able to work as camera and output HDMI next month
02:19
Bertl
it will be limited to full HD, because there is no interface with more bandwidth available on the zedboard
02:19
dmj_nova
ah
02:20
Bertl
(but you should know that, IIRC, you have one, no?)
02:20
dmj_nova
Yeah, I've got a zedboard
02:21
dmj_nova
Do you think DNG at 24fps is realistic by april?
02:22
Bertl
I doubt that we will have a working axiom prototype in april
02:22
dmj_nova
ok
02:22
Bertl
the crowd funding itself will take some time
02:22
Bertl
then the hardware development and testing
02:23
Bertl
I'd say june is more realistic
02:24
dmj_nova
hmm
02:27
FergusL
dmj_nova: how long what ? It took me like 5 days, but I'm slow. And a full 4k debayering from raw to any format you like should be like 1h and 10 minute
02:27
Bertl
wow, that is quite some time :)
02:30
dmj_nova
FergusL: this is with python?
02:32
FergusL
Yes but it's not python's fault that it is that slow
02:33
FergusL
It's terribly stupid, it's doing the same calculation up to 4 time
02:33
FergusL
Totally wasting time
02:33
FergusL
Is it of any interest to people here that I improve it ?
02:33
FergusL
I assumed the dng converter was doing the job well
02:34
Bertl
unrelated, the dng writer does not do debayering
02:38
FergusL
Yes, I assumed that was better for you lot
02:38
Bertl
well, investigating different debayering algorithms is something which will come handy at some point
02:38
FergusL
Yes
02:39
FergusL
But dcraw has the one I implemented
02:39
FergusL
Though I realise, dcraw needs a profile for the camera which we don't habd
02:39
FergusL
Have
02:39
Bertl
working on that would probably be a good idea as well :)
02:40
FergusL
Yes, and to profile we need a debayered image
02:41
Bertl
which you can get in at least three different ways by now
02:41
Bertl
1) with your python script
02:42
Bertl
2) with a tool called bayer2rgb
02:42
Bertl
3) with display by splitting up the planes and simply combining them
02:42
dmj_nova
FergusL: why is it doing the same calculation 4 times, why not avoid that by altering the algorithm?
02:42
Bertl
all three methods should be more than sufficient for the profile adjustments
02:57
FergusL
dmj_nova: because I was just a bit stupid and overlooked how it could make my script that slower! It's actually unusable indeed atm so yes I need to rewrite some part of it
02:58
Bertl
doesn't matter if it runs two hours once, if you get the properly debayered image for the IT8, does it?
02:58
FergusL
I need a maths tool called a graph
02:58
FergusL
Absolutely, I can try to do that that night, or tomorrow morning
04:56
Bertl
off to bed now ... have a good one everyone!
06:53
Sasha_C
left the channel
09:49
se6astian
joined the channel
10:09
[1]se6astian
joined the channel
10:10
se6astian
left the channel
10:10
[1]se6astian
changed nick to: se6astian
12:54
se6astian
left the channel
13:14
se6astian
joined the channel
13:24
rexbron
left the channel
13:25
rexbron
joined the channel
13:25
troy_s
left the channel
13:27
troy_s
joined the channel
13:27
Bertl
morning everyone!
14:02
mars_
i have a bunch of prime and fast telelenses for the nikon f mount
14:10
Bertl
cool
14:18
se6astian
hello
14:18
se6astian
fantastic!
14:21
mars_
i'm on my phone, talk to you later..
14:24
se6astian
see you
15:33
intracube
joined the channel
18:33
aPinky
joined the channel
18:34
Bertl
welcome aPinky
18:38
aPinky
left the channel
18:41
aPinky
joined the channel
18:41
Bertl
wb aPinky!
18:42
aPinky
left the channel
18:43
aPinky
joined the channel
18:45
aPinky
left the channel
18:46
aPinky
joined the channel
18:50
Bertl
off for now ... bbl
18:59
se6astian
ah hello pinky, you must be our new robotic friend to log the channel
20:11
gcolburn
joined the channel
20:14
Bertl
back now ...
20:19
gcolburn
hi Bertl. I just got my HDMI->DVI adapter though I thought I would test out the HDMI on the zedboard
20:20
Bertl
okay, note that you need to adjust the scripts to handle DVI out
20:21
Bertl
(I presume it is a passive adapter)
20:21
gcolburn
yes
20:21
gcolburn
its passive
20:22
gcolburn
the base HDMI test that ships with the zed board and displays a default image worked fine
20:23
Bertl
that's a good start
20:25
troy_s
in
20:26
troy_s
intracube: All Bayers are 422 :)
20:28
Bertl
I disagree, for example the X-Trans pattern is 522
20:28
troy_s
Bertl: You with your nerd knowledge
20:28
Bertl
nah, just google power :)
20:29
troy_s
Bertl: How about “99% of Bayers are 4:2:2”
20:29
troy_s
Bertl: Odd... so it samples from a block of five pixels?
20:29
Bertl
it seems that it has 9 pixels with 5 green, yes
20:29
troy_s
Slightly worse scaled chroma. I cannot imagine that nightmare when it gets into macroblock compression.
20:30
troy_s
(IIRC all frequency based algos are almost universally mod 8)
20:30
Bertl
actually wikipedia says it is 6x6
20:31
Bertl
(because the R and B values switch as well)
20:31
troy_s
Bertl: I intend to actually try and hammer on the image when I get home, but sadly am also on six day weeks currently.
20:31
troy_s
Bertl: Quite sure I can fumble through the profiling aspect as well as maybe generate a debayering tool.
20:32
troy_s
I wish DNG were in OIIO. That factoid makes it hard for me to be behind DNG currently. It makes integration a nightmare.
20:32
Bertl
as I said, for crude debayering, I guess we already have the tools at hand
20:33
Bertl
regarding DNG, it is one option atm, I'm not sure that deep color tiff isn't a better one for test images
20:33
Bertl
(but I haven't tested anything in this regard)
20:34
troy_s
DNG certainly seems almost appropriate for the "one format to rule them all"
20:34
troy_s
and given that BMC does a dump matching realtime, it is plausible
20:34
troy_s
TIFF not so much
20:35
troy_s
I think the choice should be between the canonical and venerable DPX and DNG.
20:35
troy_s
And DNG certainly seems more appropriate for dumping Bayers.
20:36
troy_s
My only regret currently is DNG in OIIO, but perhaps someone can make a post to see if they can get it integrated into OIIO. I suspect that Mr. Gritz would pull it.
20:36
gcolburn
for fast debayering on the CMV12000 images I've just been running dcraw on the DNG
20:36
troy_s
gcolburn: Greets. Any progress?
20:36
gcolburn
sort of
20:37
gcolburn
parts of the tutorial gave errors. I tried directions straight from the documentation and made further progress
20:37
troy_s
gcolburn: You can easily scale and crop that IT8 if need be. Just be sure that whatever app isn't mangling - and that isn't easy.
20:37
troy_s
gcolburn: Ah good.
20:37
troy_s
I worry that the image is slightly underexposed
20:38
Bertl
hmm, dpx seems to be a good choice actually, as it is supported by imagemagick as write format
20:38
troy_s
exposure is important on IT8 work as the saturations of the wider colors is inherently part of luminance - which in simple terms means that the color may not be reached without sufficiently intensity.
20:39
troy_s
Bertl: There are only two real formats in post
20:39
troy_s
Bertl: OpenEXR (intra pipeline) and DPX (between or to print)
20:40
troy_s
Bertl: But DPX isn't quite idealized for Bayers - something DNG seems almost perfect for, as well as unencumbered in all ways.
20:40
Bertl
yeah, not talking bayer pattern images, more talking debayered here
20:41
Bertl
that's why I consider tiff a good choice
20:41
gcolburn
if I don't have much luck with that open source package for the calibration I was thinking of implementing this algorithm here: http://www.cs.sfu.ca/~funt/AIC2012_Funt+Bastani_IntensityIndependentColourCalibrationAsPublished.pdf
20:41
troy_s
gcolburn: Unadvisable
20:41
troy_s
gcolburn: Wiser is to simply get to Argyll working, which is assured and vastly more reliable.
20:42
troy_s
Bertl: TIFF is sub optimal.
20:42
Bertl
maybe troy_s can help with the error on argyll?
20:42
troy_s
Bertl: DPX for debayers as it includes a TRC.
20:42
Bertl
yeah, no problem with dpx there, as it is supported
20:43
gcolburn
troy_s: have you used Argyll before?
20:43
troy_s
gcolburn: Oh yes.
20:43
gcolburn
troy_s: Okay, you might be able to get it working faster then
20:43
troy_s
gcolburn: I use it to generate 3D LUTs for display profile transforms
20:44
troy_s
gcolburn: As I said, I would do this but I am busy as crap at the moment sadly, and worse, about a continent away from my computer.
20:45
troy_s
Bertl: It is in the best interest to limit the file choice. Make it as focused and streamlined as possible. As much as I am pro DPX, it is a layer of abstraction upon a lower level format like DNG.
20:46
troy_s
Bertl: So given reflection, I am leaning towards rexbron's suggestion. Only thing that would make it an idealized choice is if we can find a dev to get OIIO support proper. That would be a huge asset.
20:46
troy_s
(and the BMC folks would also likely find a supporter or two)
20:47
troy_s
Bertl: The (tremendous) advantage to this support would be that OIIO then allows everyone to "one stop shop" transformations from DNG to EXR and DPX via OIIO.
20:47
troy_s
And that pretty much wraps up every postproduction pipeline out there.
20:48
troy_s
Editorially it is much more complex.
20:48
troy_s
As both DNxHD and ProRes are encumbered IIRC. (The latter certainly, the former almost certainly)
20:49
troy_s
And given that they are the two defacto formats, with the latter the larger target, editorial becomes much more complex. I suspect a stubbed "glue" app to go to either is the most optimal solution.
20:50
troy_s
IE: A standalone GUI / CLI application that allows for a DNG to ProRes / DNxHD transfer with a one-light baked in optionally.
20:51
troy_s
(via FFMPEG or the more robust FFMBC)
20:51
troy_s
gcolburn: What issue are you having exactly sir?
20:56
troy_s
Bertl: Is the IT8 a Wolf?
20:56
troy_s
Bertl: gcolburn will need the data files.
20:57
Bertl
troy_s: yup, http://vserver.13thfloor.at/Stuff/AXIOM/RAW/IT8_incand.png
20:58
Bertl
change .png to .raw16 for the bayer images
20:58
Bertl
-s
20:58
Bertl
actually .raw16.xz (i.e. compressed)
20:59
troy_s
Bertl: There should be a spectro text file as well
20:59
troy_s
Rxxxxxx.txt
21:00
Bertl
se6astian: any Rxxxxxx.txt files?
21:01
Bertl
https://github.com/hughsie/shared-color-targets/tree/master/targets/wolf_faust/reflective
21:01
Bertl
seems to have them or at least something related
21:01
Bertl
https://github.com/hughsie/shared-color-targets/blob/master/targets/wolf_faust/reflective/R100604.it8
21:02
Bertl
https://github.com/hughsie/shared-color-targets/raw/master/targets/wolf_faust/reflective/R100604.it8 (as raw file)
21:03
troy_s
Bertl: Shipped per IT8
21:03
troy_s
they are the spectro readings of the specific print you received
21:04
troy_s
(rather like the profile for a specific monitor)
21:04
troy_s
Generics are junk
21:04
Bertl
as the number matches the one on the IT8 chart?
21:04
Bertl
(the Rxxxxxx)
21:04
troy_s
and are worthless as the knowledge people that suggest they are useful.
21:04
se6astian
I will check at the university
21:05
troy_s
I believe the Rxxxxxx.txt file is specific to the IT8.
21:06
se6astian
I found the chart in some pile of documents, I doubt anyone still knows if there ever was a file that came with it let alone know then where that file could be by now....
21:07
Bertl
anyway, I think the R100604.it8 is at least a close approximation, as it seems to originate from Wolf Faust and has the same numer as on the IT8 chart I have here :)
21:07
Bertl
*number
21:09
Bertl
or if you mistrust the person on github, maybe this here is the better source: http://www.targets.coloraid.de/
21:09
gcolburn
that's where I downloaded from
21:10
gcolburn
for my testing thus far
21:11
Bertl
I think that those are the original files anyway, i.e. the zip contains the .txt troy_s was referring to
21:13
troy_s
Bertl: They ship per chart sadly
21:13
troy_s
but anyways...
21:13
Bertl
I'm not sure they exist per chart though
21:14
Bertl
i.e. I'm not convinced that they differ in the same charge
21:14
gcolburn
"The individually measured target reference files and OEM target files are only available on request by email "
21:15
gcolburn
I think the ones on the site are batch averages, and you can email to get another copy of your specific chart's measurements.
21:15
Bertl
the question is, how to identify the specific chart?
21:16
gcolburn
good question :)
21:16
gcolburn
how old is the chart?
21:16
Bertl
I'd say, from the image, it must be june 2010, no?
21:16
gcolburn
the read me in the download stated after about two years its best to get a new chart
21:17
gcolburn
ah. didn't notice that. I don't have the image up
21:17
Bertl
but maybe 2010:06 means something completely different
21:18
Bertl
but to me it definitely looks like the year and month
21:20
troy_s
Bertl: “IT 8.7 compatible reference file containing the color measurements for the target.
21:20
troy_s
Bertl: ;)
21:24
Bertl
yeah, I just wonder how you get those files when you only have the chart, because the chart has no unique identifier besides the charge number and the date
21:27
troy_s
Bertl: I seem to recall each batch checked and stuffed into the data
21:27
troy_s
(per print batch)
21:29
gcolburn
Bertl: was there any glare on the IT8 chart? it kind of looks like there might be but I can't tell for sure
21:30
Bertl
yes, I had to fight to keep it low, the chart is very reflective
21:30
Bertl
I would have expected some kind of diffuse reflection, but it is actually all specular
21:31
Bertl
just look at the bootom, the gray there is uniform on the chart
21:32
Bertl
but you can see the table and camera being reflected from the chart :)
21:32
gcolburn
okay. how much trouble would it be to get another shot with the defect columns off of the patches? the relative error in the profile that was generated is too high. if we crop the image I think it will have a hard time picking out the patches since they won't be the right size
21:32
Bertl
give me a few minutes and I fix the sensor errors for you :)
21:33
troy_s
Here is the link for posterity
21:33
troy_s
Have a bot tag it please
21:33
troy_s
http://trumpetpower.com/photos/Exposure#Normalizing_exposure
21:34
troy_s
Bertl: Ideally a broad diffuse source. Exposure needs to also heavily saturate the data - as in VERY well exposed but still within the linear range of the data.
21:46
rexbron
troy_s: I'm still confused as to why you have a hate on for tiff
21:47
rexbron
Also, as for OIIO support of DNG, the plumbing should already be there, as DNG is an extention of TIFF
21:48
rexbron
What is beyond my knowledge for the implimentation is there needs to be a feedback mechanism to pass settings to the underlying plug in (white balance, exposure comp etc).
21:48
troy_s
rexbron: Because it has no true gain beyond DPX / EXR.
21:49
troy_s
rexbron: EXR is the single format really required (TRC / color as an aside / parallel need of course)
21:49
rexbron
troy_s: gain? What do you mean, advantage? Unlike DPX, it's not ambigous as to whether the byte level storeage is gamma compensated or not! Remember the issues with the ASC-CDL compliance suite?
21:49
troy_s
EXR is more robust in all facets.
21:49
rexbron
troy_s: sure, at the expense of storeage space :)
21:49
troy_s
DPX has a TRC metadata tag
21:50
rexbron
as does Tiff, it supports arbitrary metadata
21:50
troy_s
rexbron: Storage is free. It isn't storage either... it is precision
21:50
troy_s
rexbron: The question is TIFF vs EXR. EXR wins.
21:50
troy_s
(both in full, half, and stereo)
21:50
rexbron
troy_s: then I'll expect a cheque for the 30Tb of hard drives needed for my low budget feature
21:51
troy_s
TIFF gains nothing over half!
21:51
troy_s
Egads
21:51
rexbron
I don't disagree
21:51
troy_s
Or even 10bit DPX
21:51
rexbron
uhh, it's moot when compared to DPX really
21:51
rexbron
except for herritage
21:51
troy_s
My point isn't a TIFF hate on, just that it is only acceptable when the others are unavailable for whatever reason
21:52
rexbron
DPX. Is. For. Printing Density.
21:52
troy_s
rexbron: Transfer. Period.
21:52
troy_s
rexbron: All house transfers are DPX.
21:52
troy_s
Largely due to the TRC issue
21:53
rexbron
troy_s: for convention and historical reasons alone
21:53
troy_s
(1.6 Cineon seems the bog standard)
21:53
troy_s
rexbron: The point is... it has never been TIFF. And apples to apples, no sense.
21:53
troy_s
Total sense for mattes
21:53
troy_s
Or other assets that it makes sense for.
21:54
troy_s
But still images less so.
21:54
troy_s
Convention is a design context.
21:54
rexbron
my qualms with DPX have to do with implimentation issues and color assumtions, that we have both come across, in the lower level libraries.
21:54
troy_s
rexbron: Let's face it
21:54
rexbron
troy_s: and the context has changed!
21:55
troy_s
rexbron: EVERYONE has screwed color up, and any format / app that can insulate that from the artist is a toxic format.
21:55
troy_s
rexbron: Agree on the shooting front, _NOT_ on the post house front.
21:56
troy_s
rexbron: If you wanted to sit in on a Baselight say, or get a film print, etc. Those are still real contextuals.
21:56
troy_s
rexbron: And to be fair to DPX, it too is robust as hell. It isn't just a legacy “gosh we all hate xxx format” - it fills its role well. (Albeit a little too powerful and doesall)
21:57
rexbron
troy_s: sure, for those contexts. For my purposes and on all the films I've posted, I go Tiff because I trust it
21:57
troy_s
(as in it almost does too much)
21:57
troy_s
TIFF can transfer curve flags and meta that 99% of apps likely ignore
21:57
troy_s
has
21:57
rexbron
opposed to DPX? Which can equally be ignored no?
21:58
rexbron
Format crusades...
21:58
troy_s
So it too is riddled with issues. DPX is well suited for a log starting point, and given convention, it is rare that it doesn't get treated as such.
21:58
troy_s
Again, I am not against tiff.
21:58
rexbron
In any case, go to OIIO and you get both
21:58
rexbron
well tested implimentations
21:58
troy_s
But if I send you a DPX, I can assure you you will at least assume it is a 1.6 log image
21:58
troy_s
Fair.
21:58
troy_s
?
21:59
rexbron
Am I a user or a library in this context?
21:59
troy_s
Where TIFF is bog standard sRGB curve which is rather goofball in imaging of this sort (photographic)
21:59
troy_s
Any app.
21:59
troy_s
The convention is my point.
22:00
troy_s
So perhaps you misread my opinion of TIFF. I would rec no other format to be honest. Given still images of a photographic nature aimed t post production however, that context leans me to DPX for the still component. Can TIFF work?
22:02
troy_s
Yes. If used carefully with someone very aware.
22:03
rexbron
I don't know where in the TIFF spec is states a TRC
22:05
troy_s
rexbron: Don't quote me but I recall the spec has a slot for the TRC
22:05
rexbron
I'd be curious to know
22:05
troy_s
rexbron: And I bet 99.9% of all apps utterly ignore it.
22:05
troy_s
rexbron: Grr... wait.
22:05
rexbron
DNGs do have linearization tables
22:08
troy_s
rexbron: Page 84
22:08
troy_s
TiffSpec
22:08
troy_s
TransferFunction
22:08
troy_s
:)
22:09
rexbron
Pixel components can be gamma-compensated, companded, non-uniformly quantized, or coded in some other way. The TransferFunction maps the pixel components from a non-linear BitsPerSample (e.g. 8-bit) form into a 16-bit linear form without a perceptible loss of accuracy.
22:09
rexbron
it's for linerizing the data
22:10
troy_s
http://pastebin.com/LfTAPM9b
22:31
gcolburn
left the channel
22:41
se6astian
good night!
22:41
se6astian
left the channel
22:53
Bertl
FWIW, here a 'fixed' version of the chart, no 'debayering' per se, just an overlay of the colors
22:53
Bertl
http://vserver.13thfloor.at/Stuff/AXIOM/RAW/IT8_incand.fix.png
22:53
Bertl
(the sensor artefacts have been removed)