Current Server Time: 18:47 (Central Europe)

#apertus IRC Channel Logs

2019/06/02

Timezone: UTC


02:17
Bertl
off to bed now ... have a good one everyone!
02:17
Bertl
changed nick to: Bertl_zZ
03:00
futarisIRCcloud
joined the channel
05:20
futarisIRCcloud
left the channel
07:31
futarisIRCcloud
joined the channel
08:07
niemand
joined the channel
08:07
niemand
left the channel
08:07
niemand
joined the channel
08:54
BAndiT1983|away
changed nick to: BAndiT1983
10:15
Nira|away
changed nick to: Nira
10:31
futarisIRCcloud
left the channel
10:47
Nira
changed nick to: Nira|away
10:50
Nira|away
changed nick to: Nira
11:02
se6astian|away
changed nick to: se6astian
11:22
illwieckz
left the channel
11:35
illwieckz
joined the channel
11:36
Bertl_zZ
changed nick to: Bertl
11:36
Bertl
morning folks!
11:52
apurvanandan
joined the channel
11:53
futarisIRCcloud
joined the channel
12:24
se6astian
changed nick to: se6astian|away
13:05
Dev_
joined the channel
13:06
Dev_
left the channel
13:12
BAndiT1983
Dev_, you should consider getting a proper IRC client, like hexchat or similar
13:31
Dev
joined the channel
13:31
Dev
changed nick to: Guest40935
13:33
Guest40935
left the channel
13:34
dev_
joined the channel
13:35
dev_
Hi BAndiT1983 , Switched to Hexchat
13:35
BAndiT1983
hi dev_, very good
13:35
BAndiT1983
how's the progress?
13:35
BAndiT1983
haven't compiled your code yet
13:36
dev_
I was trying to change MLVLoader so it can load more than one frame
13:38
dev_
I am using a simple vector to store the frames initially
13:38
BAndiT1983
seen it, later it should be automated of course
13:38
BAndiT1983
is the approach clear at the moment or should i create a timeline diagram?
13:39
BAndiT1983
timeline of an object i mean, e.g. how OCImage object is passed through several processors in the pipeline
13:42
dev_
The is what i have understood for the timeline : OCImage object takes parameters like aspect ratio and imageformat etc, it is passed to bayerdownscler which do the processing which is quit unclear to me , but at the end we have data in three channel (red, green and blue) in OCImage object. which will be used by debayering algorithim to debayer the data
13:42
dev_
please correct me , if i am wrong
13:42
apurvanandan
left the channel
13:42
BAndiT1983
downscaler is the simplest method, as it only takes known pixels of bayer data
13:43
BAndiT1983
https://en.wikipedia.org/wiki/Bayer_filter
13:43
BAndiT1983
look at the second image at wikipedia, there you see, that each channel has some know pixels, unknown ones are covered by other colors
13:44
BAndiT1983
so the downscaler grabs the known ones, usually a block of 2x2, e.g. RGGB, and calculates 1 pixel out of it, G1 and G2 are added together and divided by 2 to get average value
13:46
dev_
is etract method present in downscaler doing this ?
13:47
BAndiT1983
you describe it correctly, but be always aware that the object is to be seen independent of the pipeline, you can imagine it like 2 channels of data flow, first you have an empty OCImage and you have the pipeline, then at first image loader will load image data, forward it to downscaler or debayer, afterwards the data will be set in ocimage, this is very basic explanation and later i will explain more advanced approach, but focus on brute force
13:47
BAndiT1983
method first
13:47
BAndiT1983
of course
13:47
BAndiT1983
downscaler was implemented and, if i remember correctly, also covered by unit tests
13:47
dev_
https://github.com/apertus-open-source-cinema/opencine/blob/dev/Source/OCcore/Image/BayerFrameDownscaler.cpp
13:48
dev_
okay
13:48
dev_
hmm , could able to understand the brute force approach
13:48
BAndiT1983
https://github.com/apertus-open-source-cinema/opencine/blob/dev/Source/UnitTests/BayerFrameDownscalerTests.cpp
13:49
BAndiT1983
brute force is just what you do at the moment with your prototype, later we need to rework it to proper dataflow
13:49
BAndiT1983
so the files or frames are estimated automatically and pushed through the pipeline, set in OCImage objects and gathered in some big chunk of memory
13:50
BAndiT1983
maybe we don't need separate OCImage objects and can store only image data, as the frames are equal mostly, besides special data, but this is also something for later, when we start optimizations
13:51
BAndiT1983
just focus on getting the data from downscaler into OCImage and store OCImage objects in a big memory block
13:52
dev_
Okay , I am trying this for MLV file now,
13:53
dev_
I will update u soon.
13:53
BAndiT1983
it shouldn't matter, as the approach is always the same
13:53
BAndiT1983
ok, looking forward to your update
13:54
dev_
it shouldn't matter, as the approach is always the same : Yes, I got it
13:54
dev_
Thanks
14:03
futarisIRCcloud
left the channel
14:30
illwieckz
left the channel
14:35
dev_
left the channel
14:43
illwieckz
joined the channel
14:44
BAndiT1983
changed nick to: BAndiT1983|away
14:44
BAndiT1983|away
changed nick to: BAndiT1983
14:50
apurvanandan
joined the channel
15:02
apurvanandan
left the channel
15:04
apurvanandan
joined the channel
15:22
Y_G
joined the channel
15:45
dev_
joined the channel
15:48
BAndiT1983
hi Y_G, before we turn trello into too much chat, i will reply here, if you can cover bases of the zynq_info with current implementation in daemon, then you can continue with other thigns
15:48
BAndiT1983
*things
15:48
BAndiT1983
have you implemented a new version of zynq_info with daemon CLI?
15:49
RexOrCine|away
changed nick to: RexOrCine
15:49
dev_
left the channel
15:52
Y_G
zynq_info has been broken down into various parts , so we can retrieve individual settings with the command DaemonCLI voltages get vccint and similar
16:16
BAndiT1983
Y_G, we should think about proper naming, e.g. vcc_int or vcc_aux, but maybe there are better sounding names, maybe Bertl can help there
16:19
BAndiT1983
off for now, will be back later
16:19
BAndiT1983
changed nick to: BAndiT1983|away
16:27
apurvanandan
left the channel
16:32
Bertl
well, they are labeled VCC_INT and VCC_AUX so naming them differently would probably just confuse
16:32
apurvanandan
joined the channel
17:36
se6astian|away
changed nick to: se6astian
17:58
apurvanandan
left the channel
17:59
apurvanandan
joined the channel
18:10
apurvanandan
left the channel
18:19
Y_G
left the channel
18:53
BAndiT1983|away
changed nick to: BAndiT1983
18:55
BAndiT1983
Bertl, ok, lets keep them as vcc_int etc., would go for lower-case
19:08
RexOrCine
https://wiki.apertus.org/index.php/AXIOM_Remote
19:13
BAndiT1983
oha, new images, nice
19:43
apurvanandan
joined the channel
20:36
apurvanandan
left the channel
21:49
apurvanandan
joined the channel
21:49
BAndiT1983
changed nick to: BAndiT1983|away
21:54
BAndiT1983|away
changed nick to: BAndiT1983
21:54
BAndiT1983
apurvanandan, have you had any success?
21:55
apurvanandan[m]
Trying till now with urjtag, its is much updated. Sorting out the firmware to load and how to load on my jtag
22:05
Nira
changed nick to: Nira|away
22:11
BAndiT1983
fxload is to be used, if i'm not mistaken
22:11
apurvanandan
left the channel
22:11
se6astian
hi apurvanandan[m], you around?
22:12
apurvanandan[m]
Yes hi se6astian
22:13
se6astian
great, we have the package for you almost ready
22:13
se6astian
ship to the address you provided or a different one?
22:14
apurvanandan[m]
BAndiT1983: yes fxload is used, I am seeing how firmware will be generated
22:15
apurvanandan[m]
se6astian: The address i have is fine for me
22:15
BAndiT1983
one of the links i've sent you is explaining it, you have to use sdcc
22:15
se6astian
great, thanks
22:15
se6astian
off for now, good night
22:15
se6astian
changed nick to: se6astian|away
22:15
BAndiT1983
also off, good night
22:16
BAndiT1983
changed nick to: BAndiT1983|away
22:19
apurvanandan[m]
BAndiT1983: yes i have gone through that, but the firmware isn't compiling by that method, so figuring that out
22:20
apurvanandan[m]
Good Night
22:33
apurvanandan
joined the channel
22:33
Nira|away
changed nick to: Nira
22:41
illwieckz
left the channel
22:48
Nira
changed nick to: Nira|away
22:49
illwieckz
joined the channel
23:19
niemand
left the channel
23:53
apurvanandan
left the channel
23:55
apurvanandan
joined the channel
00:19
RexOrCine
changed nick to: RexOrCine|away