Current Server Time: 09:33 (Central Europe)

#apertus IRC Channel Logs

2014/03/13

Timezone: UTC


00:39
Bertl
off to bed now ... have a good one everyone!
06:14
sb0
left the channel
06:30
sb0
joined the channel
06:30
niemand
joined the channel
06:39
niemand
left the channel
08:36
Bertl
morning folks!
08:38
sasha_c
joined the channel
08:46
Bertl
evening sasha_c!
08:49
sasha_c
evening Bertl.
08:50
sasha_c
Does Gabe come on here often? I'm interested in asking him more about what he plans to de re: his plans for building medium format digital back for photography
08:51
Bertl
he shows up every now and then and we have a chat
08:52
sasha_c
Thanks, I'll try contacting him via email
08:52
Bertl
yes, that is definitely faster
08:53
sasha_c
Hey, it's very exciting to think about the proposed plans for the beta
08:54
sasha_c
Most of the work you've already done could be reimplemented into this, no?
08:54
Bertl
yes, indeed
08:57
sasha_c
Amazing
09:32
se6astian
joined the channel
09:38
se6astian
good morning
10:00
sasha_c
left the channel
10:12
sasha_c
joined the channel
10:41
Bertl
hey se6astian! how's going?
10:44
se6astian
good, testing rcn profiles already
10:44
se6astian
looking good
10:48
Bertl
testing with my paper-roll-contraption, I've seen that at roughly 30% gray you seem to get the best results
10:48
Bertl
the problem is to get all channels to one even illumination peak
10:48
se6astian
lowering black sun protection seems to slightly improve noise, but it requires a recalibration of rcn
10:48
se6astian
I see
10:49
se6astian
maybe a LUT like the sawtooth can aid visually?
10:49
Bertl
yes, definitely, but I wasn't able to get even illumination here without the contraption
10:50
se6astian
can you take some photographs of the "thing" :)
10:50
Bertl
I will try, but it doesn't look very intersting
10:51
Bertl
it is basically a carton board roll like those in the middle of a toilet paper roll
10:51
Bertl
with a diffusor made out of a special slide foil
10:51
Bertl
at roughly 2-3cm from the sensor
10:52
Bertl
and a small board with an RGB (I tried individual red green blue leds as well) led, which is controlled via the pmod port
10:52
se6astian
I see
10:53
Bertl
the led has a wider angle (120°) and illuminates the diffusor, which refracts the light into the micro lenses of the sensor
10:53
se6astian
I just checked my histogram and was surprised that my green1 and green2 "bumps" are not on top of each other
10:53
se6astian
is that a potential sideeffect of the RCN?
10:53
Bertl
yes, of course
10:53
Bertl
if you adjust it with uneven lighting in the red/blue
10:54
Bertl
the green will be 'corrected' apart
10:54
se6astian
I see
10:54
se6astian
that could be used as a good indicator of the profile results
11:03
se6astian
would a good profile result in fully alligned bumps?
11:05
Bertl
if the illumination is even, yes
11:05
Bertl
i.e. even illumination -> gauss curves at the same location
11:05
Bertl
the RCN then narrows the peaks
11:05
se6astian
I see
11:05
Bertl
i.e. they get higher and less wide in the slope
11:06
se6astian
can I clear the current rcn profile with loading /dev/null with pmem -B 0x60000000 -m W0x60300000+0x8000/W ?
11:06
se6astian
aka pmem -B 0x60000000 -m W0x60300000+0x8000/W < /dev/null
11:08
Bertl
pmem -B 0x60000000 -m Z0x60300000+0x8000
11:10
se6astian
danke
11:10
Bertl
you're welcome!
11:15
se6astian
added to wiki: https://wiki.apertus.org/index.php?title=Axiom_Alpha_Software#Row.2FColoumn_Noise_Correction
11:17
se6astian
hmm, seems running "pmem -B 0x60000000 -m Z0x60300000+0x8000" just crashed my camera
11:17
sasha_c
left the channel
11:17
se6astian
I did turn off image acquisition beforehand but left the histogram open (which runs snaps on its own afaik)
11:18
Bertl
shouldn't have any effect
11:19
Bertl
hmm, but obviously it has
11:19
Bertl
i.e. I can reproduce it here
11:20
Bertl
will investigate
11:21
Bertl
ah, yes, bug in pmem, sec
11:22
se6astian
after reboot I noticed that my green bumps also down line up without any rcn profile
11:22
se6astian
I just checked the offsets but they are equal
11:22
se6astian
also the matrix is fine
11:22
se6astian
maybe the channels (numbers changed)?
11:22
Bertl
how much do they differ?
11:23
Bertl
channel numbers only change when you flip or do not flip
11:23
se6astian
screenshot, one sec
11:24
se6astian
https://cloud.gerade.org/public.php?service=gallery&t=88644302ffde4f6e5659627a3c8bbf32&path=//Axiom#88644302ffde4f6e5659627a3c8bbf32/Axiom/alpha/histogram02.jpg
11:25
se6astian
image flipping is set to 2 = flip in Y (recommended)
11:25
Bertl
okay, that is way off, i.e. your channels are wrong
11:25
Bertl
the blue and red look like the greens
11:25
Bertl
most likely the first green is the blue and the second the red
11:26
se6astian
if I set flipping to off = 0 it seems to be correct again
11:26
se6astian
in the histogram
11:26
Bertl
so you do not account for the flipping then :)
11:26
se6astian
not yet
11:27
se6astian
I assume flipping was set to 0 by default in the beginning? (when I created the histogram code)
11:28
Bertl
probably, back in the early days :)
11:28
se6astian
ok :)
11:36
Bertl
btw, the cmv_hist3 now supports the -C option
11:36
Bertl
which allows you to specify a percentage of the width/height which will get sampled
11:37
Bertl
e.g. -C50 will only sample the center 1/4th of the entire image
11:39
se6astian
nice, parameter is percent of height/width in center?
11:39
Bertl
yes, default is 100
11:43
se6astian
great
11:43
se6astian
new pmem on 13thfloor?
11:45
Bertl
still working on it
11:45
Bertl
it seems to be a problem with the glibc not with my code
11:45
se6astian
ah ok
11:46
Bertl
or maybe not ...
11:51
Bertl
okay, I guess I fixed it so far
11:52
Bertl
checking now
11:52
se6astian
my camera seems to crash a lot today, I first thought it might be a loose connection, but I am not convinced of that anymore
11:55
Bertl
strange things happen here too
11:56
philippej
joined the channel
11:56
Bertl
maybe the xilinx tools built a bad image ...
12:04
se6astian
I also seem to have solid color areas in over/under expsoure areas after rcn calibration
12:05
se6astian
I did enable clipping though
12:10
Bertl
let me rebuild the image, just to make sure
12:12
Bertl
pmem has been updated, to clear the rcn you probably want to use ./cmv_rcn3 -z
12:13
Bertl
you can do it with pmem though, but the command is:
12:13
Bertl
pmem -B 0x60300000 -m F0x60300000+0x8000=0/W
12:13
Bertl
i.e. zero does not work, as it uses undefined access sizes
12:16
se6astian
ok
12:23
se6astian
updated wiki
12:28
philippej
left the channel
12:39
Bertl
okay, bitstream has been updated, also wmem_conf.sh changed
12:41
Bertl
btw, how do you enable the rcn clip?
12:43
se6astian
fil_reg 11 0xFC31F000
12:43
Bertl
okay, that's fine :)
12:44
Bertl
ah I see, the upper limit clip doesn't work as expected
12:52
sasha_c
joined the channel
13:00
se6astian
PLR GUI ready: https://twitter.com/ApertusOSCinema/status/444109187711107072
13:00
Bertl
PLR stands for?
13:02
se6astian
piecewise linear response
13:03
Bertl
ah, nice
13:24
se6astian1
joined the channel
13:25
se6astian
left the channel
14:06
troy_s
left the channel
14:08
troy_s
joined the channel
14:09
sasha_c
left the channel
14:22
troy_s
greets all
14:23
troy_s
se6astian1: You made progress on the >70% RGB values!?!
14:25
danieel
now we know what caused the delay of BM cameras :)
14:25
Bertl
I think we are making good progress
14:25
troy_s
Bertl: Fantastic.
14:25
Bertl
I build a contraption which allows to measure the sensor without the illumination problem
14:25
troy_s
Bertl: I can pry a little time as soon as you think you have it on the handle.
14:26
troy_s
Bertl: The >70% woes?
14:26
Bertl
I'm currently fixing a few minor bugs in the image/rcn pipeline
14:26
troy_s
Bertl: Were there some there that were revealed in our tests?
14:26
Bertl
well, yes, we have a linearization LUT now, and the contraption allows me to evenly illuminate the sensor with selected colors
14:27
Bertl
so testing the linear range is rather simple
14:27
troy_s
Bertl: If se6astian1 desperately wants a matrix, we can use the existing 4.5 shaper + matrix we have.
14:28
troy_s
Bertl: You built a bellows?
14:29
Bertl
well, kind of .. it is a simple tube (paper) with a diffusor in the middle between the sensor and a very precisely controlled led illumination
14:30
troy_s
Bertl: Any tests that further your hypothesis?
14:30
troy_s
Bertl: (that >70 RGB the values are junk?)
14:30
Bertl
the leds are synchronized with the frame request trigger
14:31
Bertl
so I can control how much light hits the sensor per channel
14:31
troy_s
Bertl: Ah.
14:31
troy_s
Bertl: And this allows you to calibrate the response?
14:31
Bertl
yes, as I know what I expect and what I get from the sensor
14:32
danieel
so you get a linearization lut - but is it stable? exposure time / temperature wise
14:33
Bertl
time will tell :)
14:33
troy_s
danieel: The sensor is _mostly_ (68%?) linear as expected.
14:33
troy_s
danieel: It is the >70% RGB values that are causing the worst fits that cannot be shaped out.
14:33
troy_s
(the floor is uniform)
15:22
Bertl
se6astian1: okay, there is definitely another bug in the rcn code, have to simulate this and see where it goes wrong
15:38
se6astian1
ok, thanks!
15:39
rainer_vie
joined the channel
15:43
rainer_vie
\o/
15:43
Bertl
hey, how's going?
15:45
rainer_vie
thanks fine... yourself?
15:45
Bertl
fine so far, on the bug hunt atm :)
15:46
se6astian1
Hi rainer_vie, I just invited you to the axiom-dev mailing list
15:46
rainer_vie
oh great...
15:47
rainer_vie
debugging is sometimes a nice work and sometimes it can be pain in the a...
15:47
rainer_vie
tomorrow i have the first chat with departure funding
15:48
rainer_vie
hope to get a meeting with them in a week or so
15:48
rainer_vie
till then i need to make a small presentation
15:50
se6astian1
for the media copying/viewing vault?
15:50
rainer_vie
btw I finished a first version of my Clipfinder Program http://dcs.co.at/pics/Clipfinder.jpg
15:50
rainer_vie
yes... I want to couple it with openCine
15:51
rainer_vie
Malte had a good point....
15:52
rainer_vie
one tiny board with an lcd which copy/hashes only
15:52
rainer_vie
and a bigger one which is scaleable
15:53
rainer_vie
the bigger one should then have also the openCIne on it
15:54
rainer_vie
on the HP of openCine it is very good described what it should can... so then there should be a Hardware
15:54
rainer_vie
which is portable and rugged solid
15:56
se6astian1
sounds good
15:57
rainer_vie
I hope that the guy at departure thinks that also... :)
15:58
se6astian1
well they are just responsible for the administrative part of the grant and while they have experience with previous calls they are not involved when it comes to making a decision isnt it?
15:59
rainer_vie
yeah thats some gremium which makes the decision... the good thing with departure is, that they found 50%... so I will calculate with the double to get whats needed :)
16:01
rainer_vie
bertl what do you think will it take on manpower to develop openCine in a basic version???
16:02
rainer_vie
or sebastian...
16:04
Bertl
rainer_vie: no idea
16:15
se6astian1
hard to say
16:16
se6astian1
actually most of the software already exists
16:16
se6astian1
and just needs to be "glued" together :)
16:16
se6astian1
so for getting DNG realtime playback its probably very little work/time if you find someone who knows what he is doing
16:17
se6astian1
the more you dig into color grading then the more complex it will get
16:18
rainer_vie
the first version should have only settings for debayer
16:18
rainer_vie
a next version could have then a primary grade option I think
16:18
rainer_vie
displaying in realtime waveform etc. will also take time I think
16:19
rainer_vie
do you know shotcut ?
16:23
rainer_vie
it has a primary color correction already inside... and there are various open vfx projects I found which are dealing with that...
16:32
se6astian1
I did not try it yet
16:32
se6astian1
but since these are all open source tools, thats the glueing together I mean
16:34
rainer_vie
sure...
16:55
troy_s
left the channel
16:56
rainer_vie
left the channel
16:57
troy_s
joined the channel
16:57
rainer_vie
joined the channel
17:04
se6astian1
after some testing it appears that the histogram seems to be crashing the camera mostly
17:05
se6astian1
histogram plus automatic image acquisation at the same time maybe
17:07
Bertl
do you trigger a snap in the histogramm acquisition?
17:07
Bertl
if so, try leaving the -s out
17:08
Bertl
IIRC, the hist code probably needs an update
17:08
se6astian1
I removed the -s option as soon as the automatic acquisition was implemented
17:09
se6astian1
yes crashed again when I had the histogram page openend and started automatic acuqisition
17:09
se6astian1
ok time for cooking dinner then :)
17:10
Bertl
and crash means?
17:20
se6astian1
ssh stops responding, webserver stps responding, FPGA seems to continue operation
17:20
se6astian1
so a CPU crash I presume
17:20
se6astian1
*stops
17:22
Bertl
okay, that sounds more like a stall in the AXI bus then
17:23
Bertl
could you try to send a magic-sysrq when this happens again via the serial console?
17:26
se6astian1
sure, how do I dod that?
17:27
Bertl
really depends on your serial console terminal app
17:27
Bertl
http://en.wikipedia.org/wiki/Magic_SysRq_key
17:28
Bertl
with minicom, it is CTRL-A Z F
17:29
se6astian1
I see, I will look into it
17:30
Bertl
i.e. with a typical terminal emulator, it's just sendig a break character
17:31
Bertl
(followed by e.g. 'h')
17:31
se6astian1
I am using putty
17:33
Bertl
for the serial console?
17:49
se6astian1
yes
17:49
Bertl
well, no idea then, maybe check the help?
17:49
se6astian1
will do
17:49
Bertl
(look for sending a break character)
17:54
se6astian1
found it
17:54
se6astian1
[ 94.241855] SysRq : HELP : loglevel(0-9) reboot(b) crash(c) terminate-all-tasks(e) memory-full-oom-kill(f) kill-all-tasks(i) thaw-filesystems(j) sak(k) show-backtrace-all-active-cpus(l) show-memory-usage(m) nice-all-RT-tasks(n) poweroff(o) show-registers(p) show-all-timers(q) unraw(r) sync(s) show-task-states(t) unmount(u) show-blocked-tasks(w)
17:55
Bertl
yep, looks good
17:55
Bertl
try this when it 'crashes'
17:55
se6astian1
ok
17:56
Bertl
btw, new bitstream and tools shortly
17:56
se6astian1
great
17:56
se6astian1
crashed, does not accept any input anymore
17:57
Bertl
yeah, then it is probably the AXI splitter
17:57
se6astian1
ok
17:57
Bertl
try to avoid writing/reading from more than one register at a time
17:58
Bertl
mikkael is already looking into it, AFAIC
18:02
se6astian1
thanks
18:07
Bertl
okay, bitstream updated, rcn3 as well
18:07
Bertl
there should be no clipping issues anymore, but the rcn data has changed
18:08
Bertl
i.e. you need to recreate a profile, but the rcn3 tool should do a good job now
18:08
se6astian1
thanks
18:08
se6astian1
will try
18:14
se6astian1
but I do need to enable clipping still before right?
18:19
Bertl
before what?
18:20
Bertl
well, you need to enable clipping if you want the pipeline to clip :)
18:22
rainer_vie
bertl your bayer pattern is 4 pixel ?
18:22
Bertl
yes, RG/GB
18:22
rainer_vie
dcraw can't decode with my command line
18:23
Bertl
se6astian1: new rcn3 tool, which can now take several samples
18:23
Bertl
i.e. just specify -N 3 or -N 5 (it will average the measurements)
18:23
Bertl
rainer_vie: I didn't find a way to actually feed the data into dcraw back then
18:24
Bertl
it seemed to me, that it doesn't handle 'raw' data :)
18:24
rainer_vie
with what do you debayer the raw stills ?
18:24
Bertl
imagemagick
18:25
rainer_vie
ohhh
18:25
Bertl
https://wiki.apertus.org/index.php?title=Axiom_Alpha_Software#Simple_debayer_with_ImageMagick:
18:25
se6astian1
ah nice, I just ran a test with clipping enabled and a single sample
18:25
niemand
joined the channel
18:25
Bertl
wb niemand!
18:26
niemand
hi Bertl
18:27
se6astian1
single snap works good, a bit dark now without daylight so might not be optimal from exposure level pov
18:27
se6astian1
will test with some more light and multiple snaps now
18:29
rainer_vie
Bertl, Imagemagick says it doesn't know the -clone
18:29
Bertl
what version?
18:30
rainer_vie
1.3.16
18:31
Bertl
hmm, maybe you want to update, 6.7.5 here :)
18:31
rainer_vie
I installed it with apt-get so its a very old version
18:32
Bertl
no, it's an ancient version :)
18:33
Bertl
but the 1.3.16 version looks like a replacement
18:33
Bertl
i.e. maybe a supposed to be compatible package?
18:34
Bertl
yeah, that's graphicsmagic
18:34
Bertl
a branch/fork from imagemagick 5.5.2 (2002)
18:36
Bertl
but you might ask the graphicsmagic maintainers how to achieve similar (if possible)
18:36
rainer_vie
right I'm compiling the actual version
18:37
Bertl
there should be a package in debian/ubuntu I guess
18:37
niemand
rainer_vie: s/actual/current/
18:37
Bertl
anything between 6.8 and 7.6 should do
18:38
niemand
or 'latest'. but actual is 'tatsächlich' :P
18:40
se6astian1
just ran rcn with N 5
18:40
se6astian1
very interesting results
18:40
Bertl
did you update the tool?
18:40
Bertl
if everything is black, then no :)
18:40
se6astian1
I did update
18:41
Bertl
2f3792f25ce3ace3706a4473ace372e6 cmv_rcn3
18:41
se6astian1
FPN is so heavily reduced that its not even ah you updated it again ;)
18:41
se6astian1
I have an 20 minute old version then
18:41
Bertl
ancient!
18:41
Bertl
:)
18:42
se6astian1
the results are still interesting, with N5 the FPN is reduced so strongly that its not even really visible anymore when I boost the contrast to the max
18:42
se6astian1
but I noticed that it also does auto white balance at the same time ;)
18:42
Bertl
wait for the actual results :)
18:42
se6astian1
before the wall had a yellowish/red tint from the tungsten light
18:42
se6astian1
after the profiling it was pure white :)
18:43
Bertl
yes, as I said, you want to get even illumination over all channels
18:43
Bertl
(before you run rcn3)
18:43
se6astian1
so I need to adjust the matrix ?
18:43
Bertl
matrix is not involved
18:43
Bertl
i.e. it is in the output pipeline
18:43
se6astian1
I have no daylight anymore now
18:44
se6astian1
how can I get evenly exposed channels....
18:44
Bertl
no idea, but for now you probably have to live with unwanted white balance :)
18:44
se6astian1
ok :)
18:45
se6astian1
the command to save the FPN profile changed as well didnt it?
18:45
se6astian1
pmem complains about wrong parameters with the command on the wiki
18:46
Bertl
what's the command on the wiki?
18:46
se6astian1
ah wait
18:46
se6astian1
my bad
18:46
se6astian1
putty edited the command unintentionally
18:46
Bertl
tsts
18:47
se6astian1
updating again now :)
18:47
se6astian1
you compile and upload faster than I can download and update my camera
18:47
se6astian1
:D
18:48
se6astian1
is the source of those cmv_* tools on github already btw?
18:48
rainer_vie
sebastian you need a 6k HMI in your livingroom
18:49
rainer_vie
then its good illuminated
18:49
se6astian1
and well heated :)
18:49
Bertl
nope, haven't cleaned up the tools yet
18:49
se6astian1
ok
18:56
se6astian1
the very latest version of the rcn tool does not do whitebalance changes anymore
18:56
Bertl
it will do it implicitely but very subtle
18:57
Bertl
i.e. differences in brightness across the paired row/col channels will cause the tool to try to compensate
18:57
se6astian1
and the pattern is a bit more visible again ... :)
18:57
se6astian1
maybe because I changed the lighting source now
18:57
se6astian1
maybe best if I retry this with sunlight
18:58
Bertl
also be careful, the rcn cannot help with over/under exposure
18:58
Bertl
i.e. if you have a perfectly corrected 50% gray image
18:58
Bertl
and then you switch to white (overexposed) the colum/row noise will be visible again
18:59
Bertl
you can compensate for this with the LUTs before the RCN correction
19:00
Bertl
I think I will shift the range by 2 bits there, so that you can go over and under the expected values by a factor of 4
19:03
rainer_vie
all libraries in /usr/local/lib but when calling convert it does'nt find libMagickWand-6.Q16.so.2
19:03
Bertl
probably your library path is not configured to include /usr/local/lib
19:03
Bertl
what distro are you running?
19:04
rainer_vie
ubuntu 13.10
19:04
rainer_vie
I could copy all them from usr/local/lib to usr/lib
19:04
Bertl
https://launchpad.net/ubuntu/saucy/+package/imagemagick
19:04
Bertl
why not just install the imagemagick package?
19:05
rainer_vie
it is 6.7 not 6.8
19:06
Bertl
okay, so give it a try then :)
19:06
Bertl
btw, here is a walk through how to build a newer package
19:06
Bertl
http://namhuy.net/1730/install-imagemagick-6-8-linux-mint-ubuntu.html
19:07
rainer_vie
kk copied all from /usr/local/lib to /usr/lib now it works
19:07
Bertl
ouch, okay :)
19:08
rainer_vie
great... but the debayer takes time....
19:09
Bertl
it's a lot of data and probably depends on the available memory
19:10
rainer_vie
sure
19:12
rainer_vie
is the sensor 4:3 ? aspect is 1,333
19:12
Bertl
yes
19:13
rainer_vie
oh nice, so we can use anamorphic lenses as well
19:13
rainer_vie
for 16:9 you scale the image or read a window?
19:14
rainer_vie
ahhh
19:14
rainer_vie
forget it
19:16
rainer_vie
I meant do you read always the hole sensor array or a window if you need 16:9.
19:16
Bertl
there is a simple way to define vertical regions
19:16
Bertl
i.e. stripes, which get transferred
19:17
Bertl
se6astian1: which reminds me, that we should test this feature
19:18
Bertl
(as it will increase the framerate or possible exposure values)
19:19
rainer_vie
^ ^ in most cases we shoot 16:9... as conform to the DCI spec a 1:1,9 window will be also interresting
19:19
Bertl
1:1.9?
19:20
rainer_vie
in digital cinemas the projectors have three formats:
19:20
se6astian1
yes, will do soon
19:20
rainer_vie
full, flat and scope
19:20
Bertl
rainer_vie: sounds strange, that would be 1600x3072?
19:21
niemand
left the channel
19:22
rainer_vie
2048x1080 or 4096x2160 is the full format
19:23
niemand
joined the channel
19:23
Bertl
ah, that's 1.9 then (or 1.9:1)
19:23
rainer_vie
1998x1080 or 3996x2160 is flat (Academy 1:1,85)
19:24
Bertl
why are you switching H and V (i.e. inverting the fraction?)
19:25
rainer_vie
right 1,9:1 or 1,85:1 or 2,39:1
19:26
rainer_vie
so thats the worldwide aspect ratios in cinema
19:26
Bertl
yeah, no problem with any of those I guess
19:26
rainer_vie
1,85:1 and widescreen 2,39:1 is common
19:27
rainer_vie
also very common is to use the hole sensor with 1,33 : 1 and scale it then to widescreen
19:27
rainer_vie
with anamorphic lenses
19:28
Bertl
yes, I hear about this practice
19:29
rainer_vie
but it is not very often used in europe it is more a hollywood style ;)
19:31
Bertl
I've heard it gives a specific 'look'
19:31
Bertl
not that I've ever actively noticed to be honest :)
19:32
Bertl
se6astian1: new bitstream and rcn3 tool, also new linear_conf.sh and setup.sh defaults
19:32
rainer_vie
yes, but the newer anamorphotic lenses are better corrected, so they made filters for the anamorphic lens flare... this horizontal flares... weired thingsd
19:33
Bertl
se6astian1: they allow now to stretch the 'linear space' about a factor of 4 over the resulting range, thus allowing for proper clipping on over and under exposure (i.e. to white and black)
19:34
Bertl
you can easily play with the feature by using linear_conf.sh 0.30 (for example)
19:34
Bertl
0.25 is default, second argument is an offset
19:34
Bertl
I will make the 0.25 = 1.0 by scaling the input value when we see that 2 bit is a good choice
19:35
Bertl
rainer_vie: yeah, I can imagine that flare effects look slightly different
19:36
Bertl
so, let's find the linear range now :)
19:36
rainer_vie
ok... so I will play with my listobjects in my python project
19:37
rainer_vie
that will get funny after three glasses of red wine
19:37
sb0
left the channel
19:37
niemand
left the channel
19:38
Bertl
:)
19:39
sb0
joined the channel
19:40
sasha_c
joined the channel
19:42
Bertl
morning sasha_c!
19:43
sasha_c
morning Bertl (even though it's 9:43pm in Vienna, hehe)
19:44
Bertl
https://github.com/apertus-open-source-cinema/alpha-software/blob/master/cmv_hdmi2/registers.txt
19:45
Bertl
it isn't complete yet, but it describes the first two register blocks :)
19:45
sasha_c
You read my mind. I have the day off from work today, so I can get started on adding those details to the wiki that you requested.
19:45
Bertl
I'll try to complete it in the next few hours
19:46
Bertl
it would be nice to find a somewhat appealing table format for the wiki
19:47
Bertl
similar to what I did here: https://wiki.apertus.org/index.php?title=ZedBoard
19:48
Bertl
you might want to look at the CMV12k datasheet, the format used there is not uncommon and I think somewhat easy to read
19:48
sasha_c
I see. And I'll add this under the CMV12000 page: https://wiki.apertus.org/index.php?title=CMV12000
19:48
Bertl
no, make a new one
19:48
sasha_c
ok
19:48
Bertl
the CMV12k registers are only a small part
19:58
sasha_c
I've been able to dig up Reference:CMV12000-datasheet-v6 from my inbox, but this is dated from 2012. Is there a more recent data sheet?
20:02
Bertl
yes, there is on github, but it hasn't changed much layout wise
20:03
Bertl
so for having a reference how registers might be documented in a table it should suffice
20:03
sasha_c
cool, thanks
20:04
Bertl
for the actual register values (we will transcribe them sooner or later to the wiki as well) I'd say we wait till we know more about the sensor
20:04
sasha_c
ok
20:04
Bertl
so primary focus is on the registers we created/provide
20:31
sasha_c
Is this suitable: https://wiki.apertus.org/index.php?title=CMV12000_to_HDMI_registers
20:31
sasha_c
(for a start)
20:33
Bertl
hmm, well, the table might work for the actual registers
20:34
Bertl
but you are currently listing the register blocks
20:34
Bertl
(so the heading is misleading)
20:34
sasha_c
I'll change the heading to "CMV12000 Register Blocks" then?
20:35
Bertl
i.e. the xxxxx means that there are addresses where the 'x' are
20:35
Bertl
I'd suggest to reduce this specific table to only two columns
20:35
Bertl
the address and the description
20:35
Bertl
default values do not make much sense, and we have no real name for the blocks
20:36
Bertl
https://wiki.apertus.org/index.php?title=Axiom_Alpha_Software#Register_Memory_Space
20:36
Bertl
maybe this helps to get an idea what I'm talking about
20:37
sasha_c
yes, it does. Just made the changes to the wiki. How does it look now then?
20:38
Bertl
reduce the heading to 'Address' and 'Description' and we're fine
20:39
Bertl
note that the 4 column table you had is fine for the actual registers
20:39
sasha_c
ok, nice
20:42
Bertl
http://www.analog.com/static/imported-files/eval_boards/ADV7511_Programming_Guide.pdf
20:42
Bertl
check out the register tables there, they have some nice/usable ideas as well
20:42
Bertl
I'm not sure the RW/RO column makes much sense in retrospect, because we have clearly divided ranges
20:43
Bertl
i.e. either a block is read/write or read only or split somewhere
20:44
Bertl
0x60000000 - 0x60100018 all read only
20:44
Bertl
0x60100000 - 0x60100018 all read only
20:44
Bertl
I mean
20:44
sasha_c
ok, I think I understand this
20:44
Bertl
0x60100100 - 0x6010013C all read/write
20:45
Bertl
the bitmasks presented in the adv7511 datasheet might be useful
20:45
Bertl
i.e. the ***110**** style
20:45
Bertl
if it can be visualized properly
20:46
Bertl
but it is in no way necessary
20:56
troy_s
rainer_vie: Shotcut is a joke for grading.
20:56
troy_s
rainer_vie: The larger issue with grading is intra application transfer: CDL, LUT formats, etc. etc.
20:59
troy_s
Bertl se6astian1 Does it mean we get new daylight charts?
20:59
troy_s
(or any chart)
20:59
troy_s
:)
20:59
rainer_vie
sure... I pointed that out, because it has a grading function... we definitely have an other aproach
20:59
Bertl
troy_s: yes, I'd say so, just needs some daylight I guess :)
21:00
troy_s
Bertl: Woop!
21:00
troy_s
Bertl: Good progress?
21:00
Bertl
well, I have one strange issue left with the linearization
21:01
Bertl
but I think it actually is a problem with the dsp block I'm using
21:01
Bertl
so should be sorted out soon
21:01
rainer_vie
final raw format will be DNG right ?
21:02
Bertl
maybe, maybe not
21:02
rainer_vie
what could it be else?
21:03
Bertl
raw frames, raw streams for example
21:03
Bertl
our very personal switch byte 7 with byt 23 format?
21:04
Bertl
*byte
21:04
Bertl
I don't know, I haven't looked into possible container formats yet
21:04
Bertl
troy_s: btw, we have a new .raw12 format :)
21:07
rainer_vie
DNG is currently a common format which is supported by a lot of applications out there
21:11
rainer_vie
ARRI for example made an online LUT generator for their LOG-C format: http://www.arri.com/camera/digital_cameras/tools/lut_generator/lut_generator/
21:13
Bertl
okay?
21:14
rainer_vie
regarding troy... it would be cool to make luts for color space conversions for REC709, DCI XYZ etc....
21:15
Bertl
I think the matrix is better suited for this purpose
21:16
rainer_vie
you mean on the camera or later on ?
21:17
Bertl
doesn't matter, it should be a linear transform, no?
21:19
Bertl
what I mean is, no LUT required for linear color space transformations
21:24
rainer_vie
so you get the 12 bit from the sensor and convert it to 16 bit and debayer it... the final picture is a linear 16 bit image right?
21:24
Bertl
no, we currently do not debayer it at all
21:24
Bertl
the sensor pipeline does FPN correction and scaling of the linear range
21:25
rainer_vie
yes but for the video outputs (HDMI and SDI) and later on on the raw conversion it's debayered
21:26
Bertl
the hdmi pipeline uses the RG/GB values from each bayer block to apply a color transformation, gamma LUT, possible overlay and sensds the data as RGB 242 to the encoder
21:31
rainer_vie
ok... so the picture which is displayed on the hdmi interface will be the same as debayered currently for example with imagemagick
21:31
Bertl
very similar, yes
21:31
Bertl
the hdmi encoder does some interpolation as well
21:31
rainer_vie
sure
21:32
Bertl
and we still have to configure YCrCb to test if it improves the image
21:32
rainer_vie
component signal will be needed for SDI anyway
21:33
rainer_vie
10 bit 4:2:2 YCbCr would be great
21:34
rainer_vie
will it be possible to give out progressive scan and psf ?
21:34
rainer_vie
switchable
21:34
Bertl
well, SDI doesn't require YCbCr per se, but the conversion is already implemented, so not a big deal
21:35
rainer_vie
yeah... but most of the external SDI equipment will work fine with component signal
21:36
Bertl
yes, I'm sure we will open a whole new can of worms with SDI like we did with HDMI and existing equippment :)
21:36
rainer_vie
:) I'm sure you will get this working
21:37
Bertl
but nothing I consider a real problem there, it mostly will be a lot of work to figure out what this or that equippment wants/needs/expects
21:37
Bertl
of course, open interfaces will be supported sooner than proprietary variations
21:39
rainer_vie
yes thats a big point... only think that on the interfaces like SDI you have standard framerates
21:42
se6astian1
time for bed
21:42
se6astian1
good night!
21:42
rainer_vie
bb sleep well
21:43
se6astian1
left the channel
21:45
rainer_vie
so a maximum of 60fps is possible I think on SDI currently
21:48
rainer_vie
couch time... wish you all the best!
21:48
rainer_vie
bb
21:48
Bertl
cya
21:48
rainer_vie
:)
21:48
rainer_vie
left the channel
22:08
intracube
joined the channel
22:37
ApertusWeb8
joined the channel
22:37
ApertusWeb8
changed nick to: eadmund
22:38
eadmund
yo!
22:38
Bertl
wb eadmund!
22:38
sasha_c_
joined the channel
22:39
eadmund
good news: idea of making the next board for you is under discussion. but we will also probably want to get a replica of previous one running.
22:40
eadmund
I will know in a few days what Cruse decides to use as an initial dev system to figure out this stuff.
22:40
eadmund
eg. Zedboard or Microzed or a Xilinx SDK.
22:41
eadmund
btw, Trenz in Germany have a really nice SOM: http://www.trenz-electronic.de/de/produkte/fpga-boards/trenz-electronic/trenz-electronic-te0720-zynq.html
22:41
troy_s
left the channel
22:42
Bertl
unfortunately proprietary as many others
22:43
eadmund
ah. you mean the design is proprietary, or the chips are proprietary?
22:43
sasha_c_
Bertl, can you let me know if this wiki for the CMV12000 register blocks is suitable: https://wiki.apertus.org/index.php?title=CMV12000_Register_Blocks
22:43
eadmund
or the tools?
22:43
Bertl
well, we won't get around the chips, but the schematic and layout is not known for those modules
22:44
Bertl
so using those would make us 'depend' on this specific manufacturer
22:44
Bertl
(not speaking of the many issues you can have with such designs)
22:44
eadmund
you mean you want a *documented* design
22:45
Bertl
precisely
22:45
Bertl
sasha_c_: looks good, except for the default values
22:45
eadmund
me too :)
22:45
Bertl
sasha_c_: which are currently in the 'Bits' column
22:45
troy_s
joined the channel
22:45
sasha_c
left the channel
22:45
sasha_c_
changed nick to: sasha_c
22:46
sasha_c
Can I simply rename the column from 'bits' to 'default values'?
22:46
Bertl
no, you combined two columns there
22:46
sasha_c
Ah, I see
22:46
Bertl
the Bits e.g. [3:2] and the default e.g. 0x0
22:47
sasha_c
ok, fixing now
22:47
sasha_c
thanks :)
22:47
Bertl
thank you for helping!
22:47
sasha_c
Anytime man
22:47
Bertl
btw, if you manage to get some even/odd light gray/dark gray alternating background it would probably enhance readablility significantly
22:48
sasha_c
ok
22:49
Bertl
eadmund: so the zedboard, microzed and even zybo have at least well documented schematics
22:49
Bertl
but of course, a design with public layouts would be even better
22:50
Bertl
(the zedboard layouts are public as well, btw)
22:51
eadmund
yes.
22:52
eadmund
frankly, I am more distrubed about the connectors/interfaces. The LPC VITA on the zedboard looks horrible. Did you do the design that connects to it?
22:52
Bertl
yes
22:53
eadmund
well, congratulations.
22:53
Bertl
thanks!
22:54
eadmund
so, everything is out of my hand now. I am shifting to looking for cheap tools and a hackable Rigol :)
22:56
eadmund
what are you guys up to? you could rig up an led to help with linearising the sesnor.
22:56
Bertl
already done :)
22:56
eadmund
what did you do?
22:57
Bertl
I built a contraption out of a paper tube, some master film and an RGB LED
22:57
eadmund
and youflash theled?
22:58
Bertl
the foil is used as diffusor and the LED is controlled by the FPGA
22:58
Bertl
i.e. it is locked to the frame request and can be precisely timed
22:58
eadmund
neat!
22:58
Bertl
yeah, works like a charm!
22:59
eadmund
for color calibration leds are not that useful btw
22:59
Bertl
I have no intention to use them for color calibration