F5: 444 10-bit vs 422 10-bit internal F camera

G

Gene Sung

Veteran
The F cameras obviously still have some big advantages over the FS7 - Much better compressed RAW workflow (I've heard the FS7 has the same RAW as the FS700 which is uncompressed on the 7Q, so it's crazy big). If I ever have to shoot RAW for a project, I'd probably just rent an F55 or something.

One other thing I noticed is that the F's have 444 10-bit, which I don't believe the FS7 will have. Is there a big difference between 444 10-bit vs 422 10-bit?

I shot some RAW footage and exported it from After Effects as 422 Pro-res HQ and also 4444 Pro-Res which I believe is 12 bit. I did a very extreme grade on it and I could hardly tell the difference. The 4444 seem to be a tiny hair better, but very, very minor.

So I'm wondering how different the Sony 444 vs 422 is out of the camera? Is there a noticeable improvement in IQ and grading ability?
 
gene_can_sing said:
The F cameras obviously still have some big advantages over the FS7 - Much better compressed RAW workflow (I've heard the FS7 has the same RAW as the FS700 which is uncompressed on the 7Q, so it's crazy big). If I ever have to shoot RAW for a project, I'd probably just rent an F55 or something.

One other thing I noticed is that the F's have 444 10-bit, which I don't believe the FS7 will have. Is there a big difference between 444 10-bit vs 422 10-bit?

I shot some RAW footage and exported it from After Effects as 422 Pro-res HQ and also 4444 Pro-Res which I believe is 12 bit. I did a very extreme grade on it and I could hardly tell the difference. The 4444 seem to be a tiny hair better, but very, very minor.

So I'm wondering how different the Sony 444 vs 422 is out of the camera? Is there a noticeable improvement in IQ and grading ability?
Click to expand...

Answering this question gets a bit tricky. If you're not sure what the difference between 4:2:2 and 4:4:4 is, here's a great article to read:
http://www.dvxuser.com/articles/colorspace/

If you didn't read the linked article, the shortest way to put it is this: with 4:2:2 every pair of pixels will have their chroma (color) values averaged, while maintaining separate luma (brightness values). With 4:4:4 every pixel retains it's own chroma and luma values. Obviously then, 4:4:4 is a more precise storage format that retains a greater amount of data. Now, the reason why these compressions work in the first place is because the human eye doesn't see what's being done on such a tiny scale, so unless you're doing things like keying from a greenscreen or massively altering the image colors, you're not likely to see a difference, even after grading.

Here's where it gets even trickier though: even though 4:4:4 is a more precise storage format, it really is only useful if the data incoming from the sensor has enough color depth. The FS7 and F5's 4k bayer sensor should easily provide enough color data for good 4:4:4 at 1080p... but not at 4k. I'm sure others can explain better than I can, but it's the same reason why people were dubious about whether the 4:4:4 output of the F3 was really derived from enough photosites.
 
You can also record compressed raw MXF on the FS-7 using the IFR7+AXS5 just as you can do on the FS700. It is a far less expensive solution to record to the 7Q or Shogun (when it ships and whenever the upgrade to FS7 raw is released).

The SR 444 codec is indeed only HD.

Ned
 
Viddovation said:
Here's where it gets even trickier though: even though 4:4:4 is a more precise storage format, it really is only useful if the data incoming from the sensor has enough color depth.
Click to expand...
Yes, and this can't be emphasised too much (though I'd refer here to chroma resolution rather than colour depth). Colour sampling only refers to the recording format - the container, if you like - and a big container is only worthwhile if you have enough content to make use of it! Putting one litre of water in a 2 litre bottle won't give you any more water. Same principle here.

It follows that for acquisition from a Bayer sensor, and if the basic recording format is the same as the sensor dimensions, there's little point in recording anything higher than 4:2:0. It won't capture any extra chroma resolution, but will need a much higher bitrate and give much bigger file sizes.
 
You must log in or register to reply here.
Back
Top