Current Server Time: 05:42 (Central Europe)

#apertus IRC Channel Logs

2013/12/28

Timezone: UTC


03:41
jucar
joined the channel
05:05
jucar
left the channel
10:29
jucar
joined the channel
11:00
philippej
joined the channel
12:53
philippej
left the channel
12:59
jucar
left the channel
15:22
philippej
joined the channel
16:10
Bertl
morning everyone!
16:19
philippej
left the channel
18:00
jucar
joined the channel
21:10
jucar
left the channel
21:42
rexbron
left the channel
21:42
troy_s
left the channel
21:43
troy_s
joined the channel
21:43
rexbron
joined the channel
21:43
rexbron
left the channel
21:43
rexbron
joined the channel
21:49
gcolburn
joined the channel
21:49
gcolburn
hello
21:50
Bertl
hello
21:51
gcolburn
how are you doing?
21:52
Bertl
well, struggling with an FPGA bug and having some back pain
21:52
gcolburn
sorry to hear that. been sitting too long?
21:53
gcolburn
I'm trying to get the bayer2rgb code working that you linked to previously. The image is coming out all garbled. I thought I'd see what command you have been using. Here is what I have:
21:53
gcolburn
./bayer2rgb --input IT8_2900_1.raw16 --output image.tif --width 4096 --height 3072 --bpp 16 --first GBRG --method VNG --tiff --swap
21:54
Bertl
ah, swap is not working on that code IIRC
21:54
Bertl
so you have to manually swap it before processing or use a different architecture
21:55
gcolburn
okay. so perhaps I'll hack the code
21:55
gcolburn
are you planning on writing your own demosaiq code for the HDMI streaming?
21:55
Bertl
yeah that was my first idea, but it isn't trivial to do with the code
21:55
gcolburn
ok
21:55
gcolburn
bummer
21:56
Bertl
I do not plan to do sophisticated debayering at all
21:56
Bertl
we basically get the RGB values and we will use them as is
21:56
Bertl
as we are down scaling anyway, I don't see a big problem there
21:56
gcolburn
so you'd use the average of nearest neighbor or something?
21:57
Bertl
simple overlay in the first step, average of neighbors later
21:58
gcolburn
okay
21:59
gcolburn
by the way I've made good progress on porting to C++
21:59
gcolburn
I can read in all the raw tags (just not their values)
21:59
gcolburn
the code is very fast
21:59
Bertl
sounds good ...
21:59
gcolburn
I'll probably read the image next before I read all the tag values
21:59
gcolburn
because I could do simple manipulations at that point
22:00
gcolburn
not sure if you guys have use for it, but I also got libJpeg compiled and linked to statically so I could produce jpegs
22:00
Bertl
ah, btw, what is the status with lossless .raw16 -> .dng and .dng -> .raw16
22:00
gcolburn
ah yes
22:00
gcolburn
let me work on that this weekend
22:01
gcolburn
so after the raw image is read in, I was planning on taking the rest of the bytes in the .raw16 file and dumping it into the private meta data
22:02
gcolburn
I don't think I need to have knowledge of the structure of the register data (not sure what format its in)
22:02
Bertl
no, just dump the 256 bytes somewhere
22:03
Bertl
all that matters is that the information is not lost and the .raw16 can be regenerated
22:03
gcolburn
yeah
22:03
gcolburn
i'll try that right now
22:04
gcolburn
will it always be 256?
22:04
gcolburn
or should I just calculate the size so it can expand?
22:06
troy_s
Greets Gabe.
22:06
Bertl
as long as we stick to the cmv12k, it will always be 256
22:06
Bertl
we might at some point add additional information like temperature or voltage readings from the frontend, but that's a different story
22:07
gcolburn
okay
22:08
gcolburn
Hi troy
22:08
troy_s
gcolburn: Is this generic image loading code?
22:08
troy_s
gcolburn: As in "raw in -> whatever out"?
22:08
troy_s
gcolburn: Because if it is, there are two libraries you should hook in or at least have optics for hooking in.
22:35
Wescotte
joined the channel