How Important is 10-Bit Really?

[webrunner5]

15 hours ago, Mark Romero 2 said:

Hmmm...

Wolfcrow seems to imply that there isn't much of a difference 150Mbs 10-bit and 150Mbs 8 bit. He said he expected a bigger difference between 8 and 10-bit footage after upgrading to the 400Mbs codec (although it doesn't look like he actually made a video comparing the differences between 8-bit and 10-bit at 400Mbs).

 

 

On the end where he is using that duster on the green screen he says there IS a big difference between 8 bit and 10 bit internal,  compared to external, and states he doesn't know why??? WTF LoL. How many of us shoot with a external recorder, not many of us. So to me he is Way off to advertise there is no reason to go 10 bit, or even 4.2.2 video!

If that was the case nobody in the world would ever buy a External Recorder. Atomos would be out of business. Now I know he states IF you don't use a LuT or basically edit the footage, well who doesn't do that other than grandma. I used to follow that guy years ago, but he seems to not be as reliable as he should be to state a bunch of stuff on there that he is not a wiz at. And lets face it hardly anyone is a wiz at all aspects of video.

[webrunner5]

Zak ought to be able to chime in on this. I know he did reply in this thread.  Doesn't he do this stuff as sort of a living? He has to be more qualified to answer this stuff more than me LoL.

I just bought the Original Sony A7s, haven't received it yet,  noone you are not alone anymore, so yeah I am stuck with 8 bit unless I win the Lotto to buy a external recorder that cost damn near as much as I paid for the camera.

[Mark Romero 2]

2 hours ago, webrunner5 said:

On the end where he is using that duster on the green screen he says there IS a big difference between 8 bit and 10 bit internal,  compared to external, and states he doesn't know why??? WTF LoL. How many of us shoot with a external recorder, not many of us. So to me he is Way off to advertise there is no reason to go 10 bit, or even 4.2.2 video!

If that was the case nobody in the world would ever buy a External Recorder. Atomos would be out of business. Now I know he states IF you don't use a LuT or basically edit the footage, well who doesn't do that other than grandma. I used to follow that guy years ago, but he seems to not be as reliable as he should be to state a bunch of stuff on there that he is not a wiz at. And lets face it hardly anyone is a wiz at all aspects of video.

Yeah, i guess he kind of equivocated there. He DID mention (and show) a BIG difference in the noise levels, but then went on to imply that noise "is not that big of a big deal.

BTW: Hope you have some lenses for your a7S. Those Sony lenses ain't cheap. Heck, even the aps-c lenses for my crop-sensor a6500 are pricey Be sure to buy an extra lotto ticket for me

 

[webrunner5]

Yeah I can tell you I will probably only buy 2 at the most Native Sony lenses. My son is thinking of buying a Sony A7r mk II so we might be able to swing a few more of them. But I have a fully electronic EF to Nec adapter with ND filters in it, so I may go Canon EF lenses on it. Yeah Sony lenses are big bucks no doubt. I have maybe 6 different something to Nec adapters left over form my A7, A7r days. So I will go even maybe a few Cine lenses for it?

I have given up on the m4/3 thingy. And yes I really do like the GH5s, but I am going to do the Sony A7s route, the mk II, maybe even the mk III if I live long enough LoL.

And no I am not going to buy a damn Canon FF mirrorless. I am not going to reward them with my money for just dragging their feet for years now. Sony has tried hard to push the envelope, and I will reward them, not Canon.

[kidzrevil]

The problem with 8 bit vs 10 bit argument is how the footage is being shot. People shoot underexposed with the totally incorrect white balance and try to wrestle with it in post and complain that 8 bit isn’t good enough. Well the same thing happens with 10 bit footage shot in correctly and underexposed you just get a billion shades of crap color over the 16.7 million in 8 bit. If you shoot as close to the final look as possible you literally cannot go wrong in post. If you are trying to convert from one color space to another using a LUT thats a different story. 

I usually shoot a good rec709 image and adjust it properly in post and im fine. You probably wont be able to tell the difference between my 14bit raw footage. It all comes from a 14bit sensor anyway 

[Mark Romero 2]

6 minutes ago, kidzrevil said:

The problem with 8 bit vs 10 bit argument is how the footage is being shot. People shoot underexposed with the totally incorrect white balance and try to wrestle with it in post and complain that 8 bit isn’t good enough. Well the same thing happens with 10 bit footage shot in correctly and underexposed you just get a billion shades of crap color over the 16.7 million in 8 bit. If you shoot as close to the final look as possible you literally cannot go wrong in post. If you are trying to convert from one color space to another using a LUT thats a different story. 

I usually shoot a good rec709 image and adjust it properly in post and im fine. You probably wont be able to tell the difference between my 14bit raw footage. It all comes from a 14bit sensor anyway 

When you say that you usually shoot a good rec709 image, do you literally mean you are using the 709 gamma? Or do you mean more like you try to get a shot where all the dynamic range is going to fit into rec709 easily? Or something else that I am missing entirely?

[kidzrevil]

@markr041 well for starters most if not all 8 bit cameras only shoot in the rec709 spec so I should really rephrase that. I personally expose for what I want to keep in the grade I don’t worry about dynamic range. I make sure my white balance accurate, my iso as low as possible, and my shutter speed at a minimum of 1/60. Exposed correctly the camera is compressing the 14 bit signal into an 8 bit codec. We literally do the same thing working with RAW photos so I figure I’d save a step and get the look I want IN camera. Now when I color correct my footage all the data I want is in the file and I have the flexibility neccessary to get the look I want in post. The trick is to get 90% of the look in camera instead if shooting flat or LOG especially since LUT’s are far from accurate

[mercer]

@kidzrevil interesting. Do you do a quick Rec709 transfer in Resolve for 5D3 footage, or do you use BMD Film and turn to Rec709 after?

[kidzrevil]

@mercer  I convert to BMDfilm (which is phenomenal) and I save the master so I can use it with filmconvert,impulz,koji etc. when grading in ACES I use the Canon 5D IDT or BMDfilm

[Emanuel]

@webrunner5 Go on Sigma glass, Don : ) Funny you had dismissed the GH5S suddenly now :-)

[mercer]

29 minutes ago, kidzrevil said:

@mercer  I convert to BMDfilm (which is phenomenal) and I save the master so I can use it with filmconvert,impulz,koji etc. when grading in ACES I use the Canon 5D IDT or BMDfilm

Good to know. Even Resolve’s default Rec709 settings look pretty good with ML Raw footage for a quick turnaround to ProRes and a simple grade in FCPX... or Premiere in your case. 

As I said on IG, right now I’m just messing with footage while I shoot it but when I get closer to the editing stage I really need to get more serious about color and since I am just a tinkerer now, I am contemplating an ACES workflow... we’ll see... I’m pretty impatient. lol. 

[kidzrevil]

@mercer ACES is lovely with the Canon 5D raw. You can get an HDR image out of it and upscale to 4k if you do process the footage properly. Too bad ACES  sucks with compressed footage like my sony ! Im just in love with Davinci right now man I think I am going to buy the studio version for the noise reduction feature but im conflicted since I already own NEAT video for premiere ! Thank God Davinci is great for roundtrip workflows

[kidzrevil]

this is what 8bit footage looks like when you get it right in camera and process it correctly in post.

[jonpais]

Struggling to come up with a counter-argument to ‘you’ve got to expose and WB correctly in the first place’: I suppose you could also say there’s no advantage to shooting with a $4,500 Zeiss Otus if you don’t learn to focus properly. But when you do, wow! ? 

A. All dogs can fly.

B. Fido is a dog.

C. Fido can fly.

The premise is fallacious, as is the conclusion. If I were to say that I’ve been working with 8-bit all my life and have never encountered banding; that 99% of what we’ve seen up until now has been 8-bit; that none of my clients have requested 8-bit; that Canon 8-bit is jaw-droppingly good; or that I see no banding on YouTube; or any number of other arguments - those are all anecdotal evidence. For every video someone can produce showing no banding, I can find one that does. There may well  be situations where a client demands 10-bit; I know nothing about chroma keying, but it seems 10-bit or higher is necessary; and HDR requires 10-bit as well. Never mind Sony’s transitional 8-bit HLG: there are a number of reasons they stuck with 8-bit that’ve got nothing whatsoever to do with image quality. Some say bit rate’s more important than bit depth, but the two obviously go hand-in-hand - you don’t offer a higher bit depth without higher bit rate - I don’t think 4K 10-bit 100Mbps would make much sense. Outside of the echo chamber, there is hardly a reviewer who hasn’t regretted the absence of 10-bit on some newer releases, and I can’t recall watching a single tutorial by any colorist worth his salt who would not prefer working with higher bit depth footage. Someone has said that 10-bit on these consumer cameras  isn’t really 10-bit at all - but nobody’s arguing that there’s no difference between 10-bit or 12-bit Alexa footage and that of a consumer camera. Each camera is different, every NLE is different, people’s needs and perceptions are different, and I think we should respect that. 

[Yurolov]

The question is vexed because not all 8 bit and 10 bit images are equal, which may have more to do with the manner in which the image is compressed than the bit depth. 

For instance if I shoot 10bit and raw on the same camera, the colors are the same. The only difference I notice is that the 4:2:2 10 bit compression degrades the image. The raw is also more malleable. But then again canons 8bit cinema line is pushable too.  

If anyone watched max yurevs review of the c200 you will know that his opinion is that there isn't much difference between the 8bit and raw files in terms of image or dr. 

This may be so. I haven't done the tests. But I think it is near to the truth in which case we should really be asking how effective is the compression rather than what's the bit depth. Obviously that's not something that can be marketed. That's why the term 10 bit is so ambiguous when it comes to image quality. 

For me you have to do your homework on one particular camera and then compare it with others and see for yourself if 10 bit actually makes a difference. It won't be the same across the board. For instance pit the c300 8bit vs the gh5 10 bit. 

C300 still comes out on top imo. End of the day it is which image you prefer. And 10 bit matters little if you get great colors soc which if you know what you are doing is easy enough. It only becomes a problem if you do client work and clients want to change shit. In that case only raw will do. So not sure about people's obsession with 10 bit.

[jonpais]

@Yurolov Not sure if that’s a logical, well-thought-out argument in favor of 8-bit. But heh, we’re all just talking shit here, right? Care to share files of your tests? Care to elaborate on who’s fixated with what? And compression is something that can be and is marketed. Also, I think you’re missing the point entirely: I’m not comparing the GH5 to Canon’s C300 or Alexa. Have you done chroma keying? Are you a professional colorist? Have you delivered HDR for clients? If not, how can you profess to be an authority? You’ve already confessed to having a bias toward Canon, which kind of negates any objectivity on your part. Lastly, regarding getting an image SOOC being all that matters may be true for you, but not so for those of us who shoot Log. Only RAW will do for ‘clients who want to change shit’? That is patently false. Sounds like a third-grader in a remedial English class...

[jonpais]

42 minutes ago, Yurolov said:

The question is vexed because not all 8 bit and 10 bit images are equal, which may have more to do with the manner in which the image is compressed than the bit depth. 

For instance if I shoot 10bit and raw on the same camera, the colors are the same. The only difference I notice is that the 4:2:2 10 bit compression degrades the image. The raw is also more malleable. But then again canons 8bit cinema line is pushable too.  

If anyone watched max yurevs review of the c200 you will know that his opinion is that there isn't much difference between the 8bit and raw files in terms of image or dr. 

This may be so. I haven't done the tests. But I think it is near to the truth in which case we should really be asking how effective is the compression rather than what's the bit depth. Obviously that's not something that can be marketed. That's why the term 10 bit is so ambiguous when it comes to image quality. 

For me you have to do your homework on one particular camera and then compare it with others and see for yourself if 10 bit actually makes a difference. It won't be the same across the board. For instance pit the c300 8bit vs the gh5 10 bit. 

C300 still comes out on top imo. End of the day it is which image you prefer. And 10 bit matters little if you get great colors soc which if you know what you are doing is easy enough. It only becomes a problem if you do client work and clients want to change shit. In that case only raw will do. So not sure about people's obsession with 10 bit.

 

[Yurolov]

29 minutes ago, jonpais said:

@Yurolov Not sure if that’s a logical, well-thought-out argument in favor of 8-bit. But heh, we’re all just talking shit here, right? Care to share files of your tests? Care to elaborate on who’s fixated with what? And compression is something that can be and is marketed. Also, I think you’re missing the point entirely: I’m not comparing the GH5 to Canon’s C300 or Alexa. Have you done chroma keying? Are you a professional colorist? Have you delivered HDR for clients? If not, how can you profess to be an authority? You’ve already confessed to having a bias toward Canon, which kind of negates any objectivity on your part.

To be honest I wasn't replying directly to you - it was just my musings on the topic.

I will surmise my argument and then you can tell me which points you take issue with. Full disclosure - I do have a bias towards the look of canons as against consumer dslrs. 

My argument is a simple one. 8 bit and 10 bit are not equal among cameras, which makes it difficult to determine the actual perceived difference 10 bit has over 8 bit. I believe the camera's compression has more to do with the overall image/gradeability. I believe this because if all things are equal than 10 bit should perform equally among cameras, but it doesn't. 

So for instance 10 bit on a EVA1 is better than 10bit on a GH5.

https://www.youtube.com/watch?v=LAUEuw0PEPI

Just as 8 bit on a Sony camera may be better for grading purposes than 8 bit on another manufacturers, and perhaps it is better than 10 bit on another manufacturers, or maybe it is equivalent. So that the difference between 8 and 10 bit doesn't even matter. 

 https://www.youtube.com/watch?v=AekKwgvS5K0

The difference between raw and 10 bit that I have seen has to do not with color quality but with compression. 

So that Max Yurev believes that the 8 bit image on the canon c200 looks similar to the raw image:

https://www.youtube.com/watch?v=p-fW6vlHnYY 

This is my opinion; I never professed to be an expert, but I am certainly not illogical.  

[Yurolov]

The main video is the dave dugdale matty scott video - unless you are doing high paid client work, in which case you will be using an alexa/red - it really doesn't mean anything. Lighting, talent, sensor, etc will be a lot more important. 10 bit gh4 is no better for it. And the GH5 is not good because it has 10 bit.  

[jonpais]

Couldn’t disagree more. Why the antipathy toward 10-bit?