12bit vs 10 bit test

A is 12
B is 10

I tried to cheat, but you saved them in a 16bit TIFF.
 
rbirnbaum said:
Oh yeah? How can you tell?
Click to expand...

I pushed them pretty hard in Davinci, and there looked to be slightly less artifacts in A. But the size of your head is bigger in "frame B", which gives some advantage in perception of gradation. The difference is small enough to my eye, on an 8bit monitor, that you could tell me either one is 10 or 12, and I would believe you.

Is "A" slightly expose darker than B? or is that a side effect of different bit depth?
 
Last edited:
I am looking at it in a calibrated Wide Gamut display and the difference is almost negligible. Slightly sharper in the 'in focus' area and more detail in the background. Lighting is the same. Window backlighting. Nothing changed. Only difference is the head position that could lead to a tonality difference.

But how about skin tones?
Highlights?
Color?
 
A has more information = 12 bit. Greater than 10-bit is diminishing returns unless doing extensive grading, especially when the final target is 8-bit. The new HDR standard for TVs support 10- & 12-bit formats; even if the difference isn't major, if bandwidth is there might as well use it when supported.
 
So yes, A is 12 bit and B is 10 bit... but really, I can't tell the difference except in really tiny details. I will be doing more tests in the next few days. Overexposing and with more controlled light and focus puller to ensure accuracy. Sometimes a still may seem a little less sharp because there was microscopic movement that blurred it at 180 degree shutter.

Thanks again for checking it out and giving me your feedback.

I am really trying to find out if 12bit is really worth it, or if it isn't, if there are some instances when switching to 12bit may bring some benefits.

Data saving is significant, so unless there are huge benefits to it, I don't want to be shooting everything in ProRes 444 12bit for the sake of it.
 
rbirnbaum said:
So yes, A is 12 bit and B is 10 bit... but really, I can't tell the difference except in really tiny details. I will be doing more tests in the next few days. Overexposing and with more controlled light and focus puller to ensure accuracy. Sometimes a still may seem a little less sharp because there was microscopic movement that blurred it at 180 degree shutter.

Thanks again for checking it out and giving me your feedback.

I am really trying to find out if 12bit is really worth it, or if it isn't, if there are some instances when switching to 12bit may bring some benefits.

Data saving is significant, so unless there are huge benefits to it, I don't want to be shooting everything in ProRes 444 12bit for the sake of it.
Click to expand...

Definitely smart to be doing your own tests.

If possible, I'd also suggest giving the footage to a professional Colorist, so they can push and pull the footage and give you their professional assessment. I only say this because a Colorist's job is very complex and if you're working as a DP, you may be handing footage off to Colorists...at least some of the time. This way you can get some feedback from their perspective as well.

It may be that some cases 10bit is fine, while for others you may want 12bit. And you may be able to find this out yourself, but hey, just a suggestion. After all, I don't know your level of expertise in the post world and would t want to assume.
 
Thanks Rafael,

All I did was just put a MacGregor's Alexa LUT on top of the footage and adjust WB. The same node was copied onto the other footage.

My experience is both professional and personal, but no big features or anything. Your colorist suggestion is great. I should hand it over too. I did push both in Resolve and I got to say that the differences were negligible to me. But I need to do more tests.

Maybe shooting in DNxHD instead of PRO RES 444 could make a bigger difference...? I feel like the internal processing of the camera is so good, that it is hard to tell them apart. But honestly, that is usually the case with most 10bit vs 12bit footage I see.

I'll let you guys know what the colorist says!
 
I currently only have the 10bit option due to the limitations of the AJA mini. Will be interesting to see how big a difference you find. One of the easiest ways to see if it is helping in the highlights, is to over expose the face to the point of partial blow out. We are very critical of what a face should look like, so finding differences should be easier.
 
I am aiming to do some harsher light tests and some lower light tests to see if there are any advantages when recovering and with noise levels this week if the week helps. Rainy here in NYC.
 
rbirnbaum said:
I am aiming to do some harsher light tests and some lower light tests to see if there are any advantages when recovering and with noise levels this week if the week helps. Rainy here in NYC.
Click to expand...

That sounds great!
 
TESTS UPLOADED

So, here they are. 12 bit vs 10 bit over and under exposure.

Settings:
Shot in Sensor Native Tungsten - You have to correct WB in post.
Cine EI
SGamut SLog
Lens Sony PL Prime 85mm at T2.8
Filters: IRND Firecrest 0.6 0.9 1.2 1.5 (Combining 0.6+0.9)

OVEREXPOSURE
Sun blasting on my face making me squint like a cowboy.

UNDEREXPOSURE
Moved to shaded area and dropped in a 2.7 IRND
Then I took ND off and shot at T11

Full Folder is 5gb enjoy.

https://www.dropbox.com/sh/gqe26tmjxbrb4ym/AABAInEidNgC0gQ4nOJLRjbMa?dl=0
Below a graded JPEG. Not the actual footage.

12bit Test Ricardo Birnbaum_1.4.2.jpg
 
Last edited:
It seems the images use a Rec709 color space gamut and gamma.
In the case I do not see much of a point comparing 10 and 12 bit renditions.
 
Cary Knoop said:
It seems the images use a Rec709 color space gamut and gamma.
In the case I do not see much of a point comparing 10 and 12 bit renditions.
Click to expand...

You mean the videos? The videos are shot in Slog and Sgamut - No Rec709

How are you deriving these are Rec709 from the clips?
 
rbirnbaum said:
You mean the videos? The videos are shot in Slog and Sgamut - No Rec709

How are you deriving these are Rec709 from the clips?
Click to expand...
Oh I see you included the original source videos, I thought you only provided the resulting Rec709 pictures for comparison.
My mistake.

Which slog and gamut exactly?

Edited to add: I checked the ND 06.mov both the 10 and 12 bit versions and that footage is utterly overexposed, completely useless.


overexposed.jpg
 
Last edited:
Cary Knoop said:
Oh I see you included the original source videos, I thought you only provided the resulting Rec709 pictures for comparison.
My mistake.

Which slog and gamut exactly?

Edited to add: I checked the ND 06.mov both the 10 and 12 bit versions and that footage is utterly overexposed, completely useless.


View attachment 131823
Click to expand...

"Utterly Overexposed" was the purpose of the whole test. "Completely useless" is an opinion.

The whole point of this was to push the exposure to see where it would fail, how much we could recover, and whether there was any difference between 10 bit and 12 bit in terms of recovery. The same with the underexposure. Of course they are harsh and brutal and not within anyone's exposure preferences. At least not mine.

Would love to hear your opinion with regards to the differences 10 bit vs 12 bit.

In terms of color and log profile, it is in the description of the test. slog sgamut
 
You must log in or register to reply here.
Back
Top