Jeffrey D Mathias wrote:
As to Q0, Q1, Q3, Q5:
In some pixel peep tests I found something interesting. All four Q settings looked to give identical (as close as I could tell at 300x) results for a lace curtain in focus. However (and this is the interesting part) the out of focus parts looked to get smother as Q changed from 0 to 1 to 3 to 5. I wonder if the variable compression of Braw actually starts more in these smooth out of focus areas and maybe later in the in-focus areas. This could help explain why all compressed Braw looks so good, and maybe more attention should be directed to the 'soft' areas when really evaluating for quality comparisons. In any event the Q1 seems fantastic and could save about half the space on disk and give about twice the recording time and looks to me to be a real contender to use instead of Q0 or 3:1. Q3 I would give a maybe as there is quite a lot of changes is the 'soft' areas. Q5 I personally would not likely use for my regular photography (did not like the old Q5 either, nor 5:1)(and I have a lot of storage drives.) However, I could see myself using Q5 for interviews... just not for 'art' shots.
I would be interested to hear (read) any other findings on how the Braw compression might work.
You’re one of the few I’ve seen that has noticed exactly how Q compression is working.
It’s based on what’s in focus.
So the amount of compression floats based on how much of the image is in focus. A longer lens where only say 10% of the image is in focus will get a very low data rate. More data is thrown at the sharp area and less is used on the out of focus area.
The Q setting simply specifies the data range bracket.
So if you wanted to really “break” the compression you would create a test image with a LOT of very fine detail that was all in focus, so a wider focal length lens and a deeper stop, say something like 5.6. (More than that and diffraction may affect the result)
The cool thing here is that it’s very easy to change your Q setting as you shoot to vary the compression. If you know you have a wide shot with a lot of detail that’s important for you to hold then use a lower Q number.
If it’s a talking head and the background is soft you can probably get a visually identical result from a higher Q setting.
Jb