Current Server Time: 02:25 (Central Europe)

#apertus IRC Channel Logs

2018/06/05

Timezone: UTC


01:35
rton
left the channel
01:51
BAndiT1983
changed nick to: BAndiT1983|away
02:05
aombk
joined the channel
02:55
TofuLynx
left the channel
03:05
fredy
left the channel
03:06
fredy__
joined the channel
03:15
jucar
joined the channel
03:23
Bertl
off to bed now ... have a good one everyone!
03:23
Bertl
changed nick to: Bertl_zZ
03:31
jucar
left the channel
05:39
ymc98_1
joined the channel
06:20
niemand
joined the channel
06:20
niemand
left the channel
06:20
niemand
joined the channel
07:03
g3gg0
joined the channel
07:43
se6astian|away
changed nick to: se6astian
07:44
niemand
left the channel
08:09
ymc98_1
left the channel
08:10
ymc98_1
joined the channel
08:15
ymc98_1
left the channel
08:27
ymc98
joined the channel
08:44
sebix
joined the channel
08:44
sebix
left the channel
08:44
sebix
joined the channel
08:47
fredy__
changed nick to: fredy
08:48
BAndiT1983|away
changed nick to: BAndiT1983
09:12
ymc98
left the channel
09:22
ymc98
joined the channel
09:30
rton
joined the channel
10:11
aombk2
joined the channel
10:14
aombk
left the channel
10:18
Bertl_zZ
changed nick to: Bertl
10:18
Bertl
morning folks!
12:30
danieeel
joined the channel
12:32
danieeeel
joined the channel
12:32
danieel
left the channel
12:36
danieeel
left the channel
13:40
RexOrCine|away
changed nick to: RexOrCine
14:17
g3gg0-afk
joined the channel
14:17
g3gg0-afk
left the channel
14:18
g3gg0-mobile
joined the channel
14:18
g3gg0-mobile
Hi
14:23
Bertl
hey
14:25
supragya
joined the channel
14:25
supragya
hi g3gg0 g3gg0-mobile BAndiT1983
14:26
g3gg0-mobile
Hi supragya
14:26
supragya
you asked for mlv file?
14:26
g3gg0-mobile
Yeah
14:26
g3gg0-mobile
Are you on limited bandwidth?
14:26
supragya
well, you could generate one by running the emulator at your end :)
14:27
supragya
yes, limited bandwidth
14:27
g3gg0-mobile
Okay, will y
14:27
supragya
however, I could do a small mlv with a couple of frames
14:27
g3gg0-mobile
Will usr your generator then
14:27
supragya
generator splits the files in two right now: meta and frames
14:27
supragya
will need to join those
14:28
supragya
currently fuse based system is taken out, had some issues with it
14:28
supragya
best is to clone locally and run ./setup.sh
14:28
supragya
that will do everything for you
14:28
supragya
axiom.mlv
14:29
supragya
btw, how are we proceeding now? I understood little on the direction we need to take
14:29
g3gg0-mobile
Okay will do this tonight
14:30
g3gg0-mobile
Whats planned for Phase 1 evaluation?
14:30
supragya
according to proposal?
14:32
g3gg0-mobile
Yep. Cannot acccess it right now as i am in a Workshop
14:32
supragya
Evaluation period​: Analysis phase ends: report findings and suitability of each
14:32
supragya
container format. Report if custom containers are feasible.
14:32
supragya
This is evaluation 2:
14:32
supragya
Evaluation period​:​Provide a functional and working system that can
14:32
supragya
encode/decode the decided video container format.
14:32
supragya
Preparation: ​Provide evidence with respect to where and how any performance
14:32
supragya
improvements can be made.
14:33
g3gg0-mobile
We have to be totally clear what the workflow should look like
14:34
supragya
okay... where do you say we should work on then?
14:34
supragya
design?
14:35
g3gg0-mobile
Workflow yeah
14:36
g3gg0-mobile
I thought we initially committed on writing cdng with a cache file inbetween
14:36
supragya
well, buffers are a problem for sure yes
14:37
supragya
also, mlv2cdng need to be implemented next?
14:37
g3gg0-mobile
Mlvfs already does this step
14:37
danieeeel
changed nick to: danieel
14:37
g3gg0-mobile
Hi
14:38
supragya
Am i missing the hiccup?
14:39
supragya
if current recorder is able to encode mlv, and mlv2cdng can be done, it only needs modification for example when PLR or dualexpo is added to the stream
14:39
g3gg0-mobile
Can you sketch the exact scenario of shooting a video, and whwre cdng comes into play?
14:39
supragya
Okay
14:39
supragya
From my perpective, this was what i thought from day one:
14:40
supragya
Person A goes, fires up the axiom and the recorder, and records the mlv
14:40
supragya
comes home, saves the mlv somewhere and then does post production
14:40
supragya
1st step in post production is mlv->CDNG and then everything else
14:40
BAndiT1983
there is something missing -> backup to multiple disks onb-set
14:40
BAndiT1983
*on-set
14:41
supragya
backup to multiple disks?
14:41
supragya
like a hot backup?
14:41
BAndiT1983
yep, so the data is not lost, just a thing which people always pointing at when speaking about using axiom on set
14:41
se6astian
backup is optional but good practice indeed
14:42
supragya
biggest thing is to solve the buffer issue
14:42
supragya
and how to make sure that we write to disk as fast as the stream
14:42
supragya
so we don't lose data
14:43
BAndiT1983
you cannot write as fast as the stream as the disk also needs some time
14:43
supragya
also a minor thing: current system does not split mlv into m00, m01 etc
14:43
BAndiT1983
and there are some components in betwtween whcih also require their time
14:43
BAndiT1983
buffering in ram and then writing is only usable thing at the moment
14:43
supragya
BAndiT1983, then over time ram will fill up and how do we act on bufferoverflows etc?
14:44
g3gg0-mobile
left the channel
14:44
supragya
stream is pushed to us, more like UDP that's the problem. Take it or leave it thing
14:45
BAndiT1983
?
14:45
BAndiT1983
you get data in ram, when some amount is there, then write to disk, plain and simple
14:47
supragya
seems okay, and the only option
14:47
g3gg0-mobile
joined the channel
14:47
g3gg0-mobile
Annoying -.-
14:47
supragya
btw, is there anything we are not clear about how the system should work?
14:48
supragya
g3gg0-mobile, need your thoughts on this please.
14:48
ymc98_1
joined the channel
14:49
ymc98
left the channel
14:50
g3gg0-mobile
We talked about that the system shall convert to cdng while recording and later
14:51
g3gg0-mobile
One thread is writing frames into a cache file(mlv)
14:51
BAndiT1983
nope, not while recording, it can leave to lost data on-set
14:51
BAndiT1983
*it can lead to
14:51
BAndiT1983
record mlv and nothing else, otherwise there are too many risks of lost data and wasted money
14:53
BAndiT1983
when the shoot is over and sd-card or whatever is backuped, then the user can decide to convert the data, but only then
14:53
g3gg0-mobile2
joined the channel
14:53
g3gg0-mobile2
Okay
14:53
supragya
I am sure, I have been living under this assumption (that BAndiT1983 put forth) since the day we had chat on how system should look like
14:53
supragya
hence the emulator
14:53
g3gg0-mobile2
Then its fine
14:54
BAndiT1983
supragya, you can implement conversion and we can add it to OC later, or so, to show the workflow to the people
14:54
BAndiT1983
also had trouble with raw2dng lately and it has to be reviewe after long time
14:54
BAndiT1983
*reviewed
14:55
g3gg0-mobile2
Then look into converting the mlv to cdng using e.g. mlvfs
14:55
BAndiT1983
what is the advantage to go over FUSE while converting?
14:55
g3gg0-mobile
left the channel
14:56
supragya
hmm, yes
14:56
g3gg0-mobile2
It already produces proper cdng and allows either on-demand conversion (quite effortless)
14:56
BAndiT1983
MLV reader in OC gets the data directly from the file and could be used for conversion?
14:56
supragya
maybe because it is already giving correct cdng
14:56
BAndiT1983
then you can use it
14:57
BAndiT1983
just wondering about performance of it
14:57
supragya
shouldn't be a big overhead
14:57
supragya
however, since on demand can be done
14:57
supragya
why not take it down and make it a program that converts it without fuse
14:58
g3gg0-mobile2
If we can pass through raw12 plr wothiut conversion it should be very inexpensive
14:58
supragya
that way, we have a strong starting point to add plr specific changes
14:58
g3gg0-mobile2
Sure, could be ported to OC
14:58
g3gg0-mobile2
Would support this idea
14:59
supragya
looked into dng specific methods on bitbucket (not sure if they are mlv2cdng) but, much of it is canon specific details, am i wrong?
15:00
g3gg0-mobile2
Matrices probably, yes
15:01
g3gg0-mobile2
Most likely there is even more
15:01
g3gg0-mobile2
iirc mlv2cdng was used as basis for mlvfs
15:01
supragya
can you point me to a starting point?
15:02
g3gg0-mobile2
https://www.magiclantern.fm/forum/index.php?topic=13152.0
15:03
supragya
just to be clear, I will try to take FUSE out for our specific system.
15:03
supragya
is that okay?
15:03
supragya
FUSE complicates a little (for starting point)
15:05
BAndiT1983
now you are at the point where i was wondering about fuse for it at first place
15:05
g3gg0-mobile2
You mean for camera emulation? Yeah
15:05
BAndiT1983
if you have useful interface, then any system can be used with it
15:06
supragya
BAndiT1983, did not get you... are you thinking about FUSE for mlv2dng?
15:06
supragya
*cdng
15:06
BAndiT1983
it was the idea of g3gg0, my point was, that you should access the MLV data directly first and try to convert it
15:07
g3gg0-mobile2
My plan was to reuse mlvfs
15:08
g3gg0-mobile2
Provides already proper mlv parsing, preview code and cdng generation
15:08
supragya
can you think of what changes would be needed in mlvfs (for our mlv till now, without PLR etc)?
15:09
supragya
*grammar
15:09
supragya
(mine)
15:10
g3gg0-mobile2
According to danieel, cdng does 1:1 accept our payload and adding a tag for the linearization table should be enough
15:10
g3gg0-mobile2
Comparable for what mlvfs does for jp92 (compressed mlv)
15:10
g3gg0-mobile2
Id say thats a week of effort
15:14
g3gg0-mobile2
BAndiT1983, is it okay for you if supragya spends a week on this?
15:15
g3gg0-mobile2
Then we have one possible workflow which allows CDNG being produced with reasonable effort
15:16
BAndiT1983
he can try, although i'm not sure about the performance
15:17
supragya
I am sorry, but what does linearisation table do exactly?
15:17
BAndiT1983
some google search brings varying results using fuse, with major slowdowns
15:17
supragya
are we not going from 12bit to 12bit raw from mlv to cdng
15:17
danieel
g3gg0: i was not sure if it is 1:1, but if you pack the bits corretly, then yes
15:17
BAndiT1983
i'Ve posted a paper on that before
15:18
BAndiT1983
http://www.rcsumner.net/raw_guide/
15:18
supragya
BAndiT1983, on linearisation?
15:19
BAndiT1983
yes
15:19
BAndiT1983
same paper is linked on the page
15:19
BAndiT1983
usually it was conversion from 12bit to 16 bit linear data, but this varies
15:20
BAndiT1983
page 7 in the pdf is about linearization
15:21
ymc98_1
left the channel
15:22
g3gg0-mobile2
We had two options: convert raw12-plr to raw16-lin during conversion
15:22
g3gg0-mobile2
This was the first choice
15:23
g3gg0-mobile2
The other one, discussed yesterday, was to simply pass the raw12-plr
15:23
g3gg0-mobile2
And provide a lut for linearization
15:24
danieel
i have the 12bit bit order snippet, can I paste?
15:24
supragya
so the problem statement is to make a LUT for conversion from raw12-plr to raw16-lin?
15:25
g3gg0-mobile2
A) convert by hand to raw16-lin using probably a per-pixel-lut as alex proposed
15:26
g3gg0-mobile2
B) just pass raw12-plr and provide the cdng tag for the lut
15:27
g3gg0-mobile2
Who votes for which option? I prefer B) if it is possible
15:27
supragya
B seems better
15:27
Bertl
off for now ... bbl
15:27
Bertl
changed nick to: Bertl_oO
15:28
danieel
C) convert to 16bit in per pixel and apply a global lut/curve to reduce data to 12 bit
15:29
supragya
wont that mean 12bitPLR is linearisable in 12bits itself?
15:29
g3gg0-mobile2
Hehe
15:29
danieel
you can compare all - and see if the psnr is above the rounding erros :)
15:30
danieel
hard to tell if the sensor is that bad in doing plr, that it needs per pixel correction
15:30
supragya
PSNR -> Peak signal to noise ratio?
15:30
g3gg0-mobile2
C) is like A) ;)
15:31
danieel
just 3/4 of the storage/bandwidth
15:31
g3gg0-mobile2
Yep. Preferrable
15:31
supragya
aren't we losing details in C ?
15:31
supragya
or some other things?
15:31
g3gg0-mobile2
Probably noise
15:32
BAndiT1983
what about dark frame?
15:34
g3gg0-mobile2
Id say not scope yet
15:34
danieel
either burn-in into raw data, or match with the reel number / camera model
15:34
danieel
(& camera sn)
15:35
g3gg0-mobile2
Tool support is quite bad iirc
15:35
g3gg0-mobile2
So id put it into cdng conversion
15:35
danieel
yes, i can remember the still tools have dark frame / flat field correction which is matched by that
15:36
supragya
well, I may start missing things which are necessary this way :)
15:36
danieel
if you burn this in, then there is a question wheter to mask and interpolate the hot pixels :)
15:36
supragya
g3gg0-mobile2, am I to implement B ?
15:36
BAndiT1983
implement conversion first, as the discussion is drifting off a lot
15:37
BAndiT1983
we should walk step by step
15:37
danieel
because the bad pixel list feature is also a problematic feature of the tools
15:38
g3gg0-mobile2
Supragya look into efforts required for B) or C)
15:39
danieel
these things could have a configuration option, generalized as: ignore/reference/embed/apply , for the various levels of processing
15:39
g3gg0-mobile2
Assume A) is similar to C)
15:40
supragya
se6astian, ping!
15:43
supragya
g3gg0-mobile2, analysing B and C (using tools like matlab)?
15:43
supragya
or scilab?
15:43
g3gg0-mobile2
You decide
15:44
supragya
okay!
15:44
g3gg0-mobile2
What do you want to Analyse exactly?
15:45
danieel
on a real world plr shot or artificially generated one? so that the sensor "shittiness" can be adjusted
15:45
g3gg0-mobile2
Do we have a sensor model?
15:45
supragya
real world plr would be good to test experimentally
15:46
supragya
artificial generation would be mathematical
15:46
BAndiT1983
is it in the scope of gsoc?
15:47
supragya
as far as I wrote in proposal (to make an emulation) (eval 2), no. But I really never knew what i was writing for during proposal submission
15:48
supragya
this may well be a part of gsoc
15:48
supragya
if we consider we are chosing a raw container for AXIOM Beta specifically
15:48
BAndiT1983
but is it really important at the moment?
15:49
supragya
it is really upto you, BAndiT1983 and g3gg0-mobile2 to choose the direction
15:49
BAndiT1983
can'T you just create some useful curve and test with it? as the task was about container for axiom and not all the processing and bits
15:49
danieel
but back to the question - if one (alex) thinks there *IS* a need of per pixel curve, then he must have the proof/measurement/simulation already in place
15:50
supragya
if we think raw12 is without plr, it does not need a LUT since it is linear (am I wrong)?
15:50
supragya
if plr is introduced, there are additional blocks for it in mlv system which would not be a big problem to add in current emulation
15:50
danieel
i would definitely start with A, to dont block the whole thing.. and when the proper 16bit linearizaton is implemented, then packing it with a lut/curve to 12 bit is trivial (C)
15:50
g3gg0-mobile2
Supragya look into efforts required for B) or C)
15:51
g3gg0-mobile2
Wait. First check efforts.
15:51
supragya
after that BAndiT1983, yes when mlv2cdng conversion is to be done, this discussion would crop up eventually
15:51
supragya
how to account for plr in cdng
15:52
danieel
then also check meaningfullness, if the perpixel plr relies on some unusual calibration step which is not developed, it makes no sense to implement B/C
15:53
supragya
g3gg0-mobile2, sure... will do some research, and get back to you within a few days with some thoughts on it
15:53
supragya
things are blurry for me now... but will sort it out with communications in this week
15:56
g3gg0-mobile2
Please ask if things are unclear
15:57
supragya
sure
15:58
g3gg0-mobile2
Gtg. Continue per mail please
15:58
supragya
Going away now. Thanks for time, all of you, BAndiT1983 g3gg0-mobile2, danieel.
15:58
g3gg0-mobile2
Ok cu
15:58
se6astian
changed nick to: se6astian|away
16:01
danieel
np, off for shopping too
16:01
supragya
left the channel
16:05
g3gg0-mobile2
left the channel
16:26
g3gg0-mobile
joined the channel
17:02
supragya
joined the channel
17:26
supragya
left the channel
18:10
se6astian|away
changed nick to: se6astian
18:34
g3gg0-mobile
left the channel
18:44
TofuLynx
joined the channel
18:48
TofuLynx
Good Afternoon!
18:49
BAndiT1983
hi TofuLynx
18:51
g3gg0-mobile
joined the channel
18:56
g3gg0-mobile
left the channel
19:22
TofuLynx
BAndiT1983:
19:23
TofuLynx
I added the manual paths on ProcessingTest Cmakelist
19:23
TofuLynx
and it still says undefined reference to
19:24
BAndiT1983
pastebin your cmakelists.txt please
19:25
TofuLynx
https://pastebin.com/kSV2dRiA
19:25
TofuLynx
Here is it
19:26
BAndiT1983
have you executed cmake ..
19:26
BAndiT1983
?
19:26
TofuLynx
yup
19:27
TofuLynx
I mean
19:27
TofuLynx
Qt Creator auto compiles the cmake
19:27
TofuLynx
Running "/usr/bin/cmake /home/claubit/opencine '-GCodeBlocks - Unix Makefiles'" in /home/claubit/Source/build-opencine-Desktop_Qt_5_10_1_GCC_64bit-Debug.
19:27
BAndiT1983
then commit to your fork and i will try to adjust it
19:27
TofuLynx
ok!
19:28
TofuLynx
Do I commit the 3rd party too_
19:28
TofuLynx
?
19:28
BAndiT1983
yes, please
19:28
TofuLynx
I have a problem
19:29
TofuLynx
the lodepng.cpp doesnt appear on Qt Creator, so it doesnt appear on the modified files list to commit
19:31
BAndiT1983
do you use qtcreator for git?
19:31
TofuLynx
Yep
19:31
BAndiT1983
never worked for me well
19:32
BAndiT1983
i suggest gitkraken or smartgit, more comfortable
19:32
TofuLynx
i think the problem is Qt not detecting the .cpp file
19:33
BAndiT1983
is cmake finishing without errors?
19:33
TofuLynx
yeah
19:33
TofuLynx
also, I will commit now using the terminal
19:33
TofuLynx
so you can adjust
19:36
TofuLynx
Done
19:36
TofuLynx
Can you check it?
19:36
BAndiT1983
give me a moment, have to prepare some things for it first
19:42
TofuLynx
Is it neccesary to add a CMakeLists.txt to lodepng folder?
19:43
aombk2
left the channel
19:43
BAndiT1983
nope
19:44
BAndiT1983
cmakelists is for modules which should build something, but this files will be added statically
19:44
BAndiT1983
alternatively you could copy them to the folder where you need them
19:45
TofuLynx
Ok! :/ This issue is really making me frustrated
19:45
TofuLynx
I will wait for your try :)
19:57
TofuLynx
brb, Dinner time
19:58
BAndiT1983
enjoy your meal
20:21
BAndiT1983
TofuLynx: another thing which is occuring to me, your cmakelists are not up to date with main repo, as my cmake complains about QT5_USE_MODULES which is obsolete
20:33
BAndiT1983
replace your paths to lodepng with this ->
20:33
BAndiT1983
"${CMAKE_SOURCE_DIR}/3rdParty/lodepng/lodepng.cpp"
20:33
BAndiT1983
"${CMAKE_SOURCE_DIR}/3rdParty/lodepng/lodepng.h"
21:15
TofuLynx
Back
21:16
TofuLynx
Let me try it
21:19
TofuLynx
It worked! :D
21:19
TofuLynx
Thanks!
21:25
BAndiT1983
no problem
21:25
TofuLynx
Ok
21:25
TofuLynx
I think SHOODAK2 is now good
21:26
TofuLynx
But I can't really confirm
21:26
TofuLynx
I will send a example to lab chat
21:32
TofuLynx
Sent :)
21:54
BAndiT1983
shoodak2 has more fuzz on the edges or brighter areas
21:55
BAndiT1983
i like gedi more
21:56
BAndiT1983
will you send Stephan also some samples?
21:56
TofuLynx
Yeh I will be
21:56
TofuLynx
I will send*
22:25
TofuLynx
Well, finally I can compare the different algorithms with GIMP
22:25
TofuLynx
It's a great addition :)
22:25
TofuLynx
Now, as I said yesterday, I will review the code and document it.
22:26
TofuLynx
I have to go now, Sleep time :)
22:26
TofuLynx
Good Night Folks!
22:27
TofuLynx
left the channel
22:30
BAndiT1983
changed nick to: BAndiT1983|away
22:34
se6astian
changed nick to: se6astian|away
23:19
RexOrCine
changed nick to: RexOrCine|away
23:29
Bertl_oO
off to bed now ... have a good one everyone!
23:29
Bertl_oO
changed nick to: Bertl_zZ