F5: F5/F55 1080p

Thanks Nate!

At that level of media/storage intensity, raw becomes a lot less challenging to work with and process. And to get 16-bit files for roughly the same storage cost as 10-bit 4:2:2 ProRes... well it makes the idea of shooting raw a lot more appealing to my mind.

As someone who's not likely to need 4K deliverables any time in the next couple of years. 2K raw might just see a bit of use from me.
 
Grug said:
Thanks Nate!

At that level of media/storage intensity, raw becomes a lot less challenging to work with and process. And to get 16-bit files for roughly the same storage cost as 10-bit 4:2:2 ProRes... well it makes the idea of shooting raw a lot more appealing to my mind.

As someone who's not likely to need 4K deliverables any time in the next couple of years. 2K raw might just see a bit of use from me.
Click to expand...

2K raw also tends to be almost real time or real time on most recent machines, especially if the manus debayering takes advantage of the GPU. Sony's does, unlike Red.
 
I never got around to comparing actual data rates, but wow I find it cool and interesting that at 24p Sony 2k raw 16-bit is 240mbps whereas Alexa 12-bit RGB 444 ProRes is 264mbps?

I know raw sensor data requires less data bits... but again WOW knowing this it sure makes me feel better about just shooting 2k raw 16-bit on the Sony.
 
I think there is a major open question.
What is 2k RAW?
Assuming 4k RAW is slightly compressed 8M pixel image where R:G:B = 2M:4M:2M,
2k RAW might be 2M pixel one where R:G:B = 0.5M:1M:0.5M and data rate will be reduced to a quarter.
But I wonder such a rough down sampling is not balanced with most of 16bit RAW application in terms of the resolution.
If 2k RAW means 3M pixel image where R:G:B = 1M:1M:1M, the data rate could be reduced only by a quarter and seams not to fit to HFR.
On the other hand, 2k XAVC 422 is 4M pixel image where R:G:B = 1M:2M:1M.
I think XAVC 422 with 10bit S-log2 will be better balance between the resolution and color depth for 2k.
 
Sorry, figure was mistaken.
wrong: If 2k RAW means 3M pixel image where R:G:B = 1M:1M:1M,
correct: If 2k RAW means 6M pixel image where R:G:B = 2M:2M:2M,
 
Thank you for your comment.
I know F5/F55 is not cropping and basic theory about chroma sampling and color filter array like bayer pattern.

My biggest question is what is 2k RAW, where the sensor is 4k and RAW should be 4k in principle.
That's why I'm sure some kind of demosaic and/or down sampling process is necessary.
Regarding the amount of the information, bayer RAW data is equal to 420 chroma sampled data for a same frame size.

So I think if the 2k RAW means bayer patterned sampling, 422 XAVC could be better in terms of resolution.
If the 2k RAW means 444 RGB, the image quality will be superb though data rate will stay at 75% of 4k RAW.
 
acky said:
I think there is a major open question.
What is 2k RAW?
Click to expand...
That's a very good question. Surely, by definition, the raw output from a 4k sensor is 4k.

2K raw is about a quarter the size of 4k, so to get 2k "raw" from a 4k sensor, you must either discard 75% of the sensor data, use some heavy compression, or discard and compress.

Sony refer to the 2K raw as "derived 2K" or "super sampled 2k", but what does that mean, exactly?
 
At best most native 2k sensors will derive 1.6k, the F's however I suspect will derive a full 2k from its 11MP (8 effective) sensor and frankly that is good enough for me. I like 2k over HD as the slight increase is enough to give me the padding needed for post.
 
Yes, with current state of hard disk technology, 2K raw does sound like an attractive way to go. Hopefully it won't be too long before some downloadable samples make their way into the wild, and we'll be able to see how it performs for ourselves.
 
You must log in or register to reply here.
Back
Top