Getting RAW Video

All things embedded.
*NO FEATURE REQUESTS*

Re: Getting RAW Video

Postby Fred » Mon May 04, 2015 10:22 pm

Do yout think it is possible to write 24 fps LV stream to the SD card?
Fred
 
Posts: 37
Joined: Mon Jul 28, 2014 5:09 am
Location: Budapest
Been thanked: 5 times

Re: Getting RAW Video

Postby squig » Mon May 04, 2015 10:39 pm

nielsdenmark wrote:I follow this thread for progress on weekly basis and it is truely amazing to see the level of thought put into this project.

I wonder what the ultimate perspective is with this ? - What is the highest raw output size/mbps that can be achieved and captured (by any output method) on nikon cameras ?


This site has some figures on Nikon card write speeds http://www.cameramemoryspeed.com/nikon- ... eed-tests/ I can't say with any certainty how these numbers will translate with continuous raw video recording but the Lexar 1066x CF write speed on the D810 is almost identical to the raw video write speed on a 5D MK3. If the SD card Nikons can write 67MBps 1980x804 letterbox 24fps recording would theoretically be possible but I don't have any knowledge of how the Nikons write to the buffer in comparison with the Canons.

Nice work leegong.
squig
 
Posts: 5
Joined: Mon May 04, 2015 10:20 pm
Location: Sydney
Been thanked: 1 time

Re: Getting RAW Video

Postby leegong » Mon May 04, 2015 10:47 pm

12bits 1664x1100 LV RAW is about 2.7MB size , LV RAW patch doesn't modify FPS of Liveview ,
it just gets triggered one time per Liveview frame , then save LV raw frame in RAM memory into SD card ,
so for 30 FPS Liveview , real writing speed of your SD card should be at least 2.7*30 = 81MB/s , for 24 FPS , writing speed
should be at least 2.7*24 = 64.8MB/s .
One option : It's possible for me to apply LV RAW compression with EXPEED hard core before saving it on SD card ,
but i have no idea about if it's fast enough for liveview FPS, it need testing , however this is not so urgent now .
leegong
Core Developer
 
Posts: 2143
Joined: Mon Mar 19, 2012 12:21 am
Location: Hangzhou , China
Been thanked: 553 times

Re: Getting RAW Video

Postby Fred » Mon May 04, 2015 11:13 pm

Thanks, Lee, so a SD extreme pro will suffice, I guess.
Fred
 
Posts: 37
Joined: Mon Jul 28, 2014 5:09 am
Location: Budapest
Been thanked: 5 times

Re: Getting RAW Video

Postby squig » Tue May 05, 2015 12:27 am

Too bad about the 1664x1100 aspect ratio, but that's 2k recording with a 1.5x Anamorphic lens :grin:

A1ex do you have a Nikon body to play with?
squig
 
Posts: 5
Joined: Mon May 04, 2015 10:20 pm
Location: Sydney
Been thanked: 1 time

Re: Getting RAW Video

Postby leegong » Tue May 05, 2015 12:39 am

Now chip4001 becomes the central issue , we have to modify chip4001 setting , force it reading LV RAW from RAM memory
instead of reading LV RAW from chip4000 in normal case, since the current LV RAW patch commands chip4000 write LV RAW into RAM memory
instead of sending it to chip4001 .
leegong
Core Developer
 
Posts: 2143
Joined: Mon Mar 19, 2012 12:21 am
Location: Hangzhou , China
Been thanked: 553 times

Re: Getting RAW Video

Postby nikkor » Tue May 05, 2015 2:29 am

Could writing to both sd cards one the d7000 be an option?
nikkor
 
Posts: 138
Joined: Mon May 20, 2013 11:06 am
Been thanked: 44 times

Re: Getting RAW Video

Postby astronomer » Tue May 05, 2015 12:28 pm

nikkor wrote:Could writing to both sd cards one the d7000 be an option?

Fred wrote:Thanks, Lee, so a SD extreme pro will suffice, I guess.

squig wrote:This site has some figures on Nikon card write speeds http://www.cameramemoryspeed.com/nikon- ... eed-tests/ I can't say with any certainty how these numbers will translate with continuous raw video recording but the Lexar 1066x CF write speed on the D810 is almost identical to the raw video write speed on a 5D MK3. If the SD card Nikons can write 67MBps 1980x804 letterbox 24fps recording would theoretically be possible but I don't have any knowledge of how the Nikons write to the buffer in comparison with the Canons.

Fred wrote:Do yout think it is possible to write 24 fps LV stream to the SD card?


As this is still at a way to early stage, the card speed is at the bottom of what we should worry.

Again, if you do want an answer, I remember much earlier in this project Simeon used a high speed card (something like 45MB/s UHS-1 which D7000 do not support UHS) and the highest speed for the EXPEED 2 SDHC controller was only 30MB/s. Thus don't even think about recording 1664x924 30FPS RAW. It might be possible, to output data to HDMI, but how to format/transfer/record it is another difficult task.
User avatar
astronomer
Developer
 
Posts: 722
Joined: Fri Apr 06, 2012 10:06 am
Location: US
Been thanked: 292 times

Re: Getting RAW Video

Postby leegong » Wed May 06, 2015 6:57 am

I disable all of pre-processings for liveview , shoot many dark frames and normal scenes ,
the historgam shows no scaling for shadow , EDIT : I might be wrong .
but gaping in historgam still exists , highlight overflow still exist .
EDIT : black level on IMX071 side is confirmed to be 0x258 = 600 decimal .
Attachments
LVraw_NormalSecen_PreProcessingOff.png
LVraw_NormalSecen_PreProcessingOff.png (33.57 KiB) Viewed 5295 times
LVraw_PreProcessingOff_NormalSecen.rar
(1.46 MiB) Downloaded 141 times
LVraw_DarkCurrent_ProProcessingOff.rar
(1.65 MiB) Downloaded 118 times
leegong
Core Developer
 
Posts: 2143
Joined: Mon Mar 19, 2012 12:21 am
Location: Hangzhou , China
Been thanked: 553 times

Sensor output is definitively 14Bit!

Postby astronomer » Wed May 06, 2015 9:29 am

I forget to look at the first set of normal images. And they revealed a lot!

14bit-overflow.png


First of all, this high dynamic lighting scene saturated the image sensor.
Then let's look at the transition region next to that light bulb. You can see 3 rings of black defect due to high bit overflow. Thus it means you overflowed 3 times. That's exactly what will happen if you save the lower 12 bit from a 14bit data.

As such, with gaps still at 16ADU, the image sensor is output 10bit in video mode.

Now I'm more convinced that the digital scaling is happening on sensor, because the sensor set black level at 600ADU during video. That means a lot of work ahead to figure out which register is the digital gain. But luckily, we had a whole a lot SPI bus logging a while ago. Comparison with still mode is essential.
User avatar
astronomer
Developer
 
Posts: 722
Joined: Fri Apr 06, 2012 10:06 am
Location: US
Been thanked: 292 times

PreviousNext

Return to Firmware

Who is online

Users browsing this forum: No registered users and 1 guest

cron