Does an external recorder eliminate SLog3 Banding?

[Jadesroom]

Will recording prores 422 to an external recorder when shooting Slog3 with an A7Sii eliminate banding?

[Don Kotlos]

Eliminate not, but improve yes. 

Banding depends on bit space, color space, compression, and of course how close are the tonalities that you want to represent in the log space. 

The source of the banding is when you are trying to represent fine tonalities that don't exist in the file because of the limited resolution in bits especially when in recording in a log space. But color sampling such as 4:2:0 and heavy compression can exaggerate the banding (watch this recent discussion for example). 

Other than external recording you can do the following:

1. Use a less aggressive log space such as slog2. 

2. Increase the saturation in the camera settings.

3. Bake in a LUT with the external recorder. 

4. Use a smaller color space such as sgmamut3.cine or Pro or Cinema. 

[Shirozina]

If you want to guarantee a banding free image then shoot slog-2 as it uses most of the 8bit space. Correctly exposed and WB’d slog-2 to an ext recorder is a very robust image that will only fall apart with major abuse in the grade unlike the internal codec that is so compressed it already broken before you grade it......

No doubt some will jump in and say the ext hdmi out is only 8bit so it can’t be any good......

 

[maxotics]

Self appointed LOG police here.  A camera's normal gamma profile is designed to maximize color saturation across an 8/24-bit gamma curve.  As soon as you shoot LOG, you accept a trade off between gray-scale DR and chroma.  The improvement you'll get shooting external 422 (if the camera shoots internal 420) has nothing to do with the banding from chroma spread out too thinly on a gamma curve, it is about compression.  To reduce data 420 substitutes more color than 422.   Two different types of banding (or color loss).  I'm just extending what @Don Kotlos said above.  My 2-cents is don't shoot a LOG profile that will give you noticeable banding (which depends on what you shoot) and expect a more lossless CODEC to save you

[Shirozina]

26 minutes ago, maxotics said:

Self appointed LOG police here.  A camera's normal gamma profile is designed to maximize color saturation across an 8/24-bit gamma curve.  As soon as you shoot LOG, you accept a trade off between gray-scale DR and chroma.  The improvement you'll get shooting external 422 (if the camera shoots internal 420) has nothing to do with the banding from chroma spread out too thinly on a gamma curve, it is about compression.  To reduce data 420 substitutes more color than 422.   Two different types of banding (or color loss).  I'm just extending what @Don Kotlos said above.  My 2-cents is don't shoot a LOG profile that will give you noticeable banding (which depends on what you shoot) and expect a more lossless CODEC to save you

Tell me - what actual experience do you have shooting a Sony A7s or r MK2 to an external recorder in Log and grading it?

[maxotics]

13 minutes ago, Shirozina said:

Tell me - what actual experience do you have shooting a Sony A7s or r MK2 to an external recorder in Log and grading it?

I'm glad you asked.  Very little experience.  What I did do is create 32 images representing 16 million colors and created video using various standard gamma profiles and LOG profiles and compared the evidence.  I also have a Ninja Blade where I could look at the difference between 422 HDMI out and internal 420.   Unlike some people here, I try to test my knowledge and not claim to have turned lead into gold through some special camera settings for knees, black level, saturation and other, sorry, horseshit which most people don't fully understand.  If you can grade a beautiful image, bully for you!  But if you're going to challenge me on the technical differences between say standard profiles and LOG, then I'm all ears!  That you believe I need to be good at subjective grading doesn't do it for me      

If there is something wrong in what I said, say what it is, and why.

[Shirozina]

17 minutes ago, maxotics said:

I'm glad you asked.  Very little experience.  What I did do is create 32 images representing 16 million colors and created video using various standard gamma profiles and LOG profiles and compared the evidence.  I also have a Ninja Blade where I could look at the difference between 422 HDMI out and internal 420.   Unlike some people here, I try to test my knowledge and not claim to have turned lead into gold through some special camera settings for knees, black level, saturation and other, sorry, horseshit which most people don't fully understand.  If you can grade a beautiful image, bully for you!  But if you're going to challenge me on the technical differences between say standard profiles and LOG, then I'm all ears!  That you believe I need to be good at subjective grading doesn't do it for me      

If there is something wrong in what I said, say what it is, and why.

Please explain in detail your methodology - Sounds like I could sell my external recorder if your theory is better than my eyes....

[maxotics]

34 minutes ago, Shirozina said:

Please explain in detail your methodology - Sounds like I could sell my external recorder if your theory is better than my eyes....

Why sell your external recorder?  Was that what I suggested?  I only said that banding can be a result of an aggressive LOG profile, as Don said, or from too little color information due to a high compression CODEC that uses 420.  We both have different roles, right?  You shoot and grade real stuff I assume.  I'd hire you over me to go shoot something.  I test stuff, a more scientific approach.  We're here to educate each other, I hope   Just so the OP can make sense of this I will explain some of this stuff in more detail, as I understand it!

Many filmmakers believe their cameras record color just like our eyes/brains do.  But they don't.  Cameras can only see in gray-scale actually.  The manufacturers put color filters over alternating pixels and then composite a "color".  But it's not really a color.  It's just a brightness value of light through a filter with the noise floor of the silicon as a reference/floor.  Too much brightness, and the value doesn't change.  The #1 confusion I see is the belief that these values have an implied relative brightness; that is, that the brightness difference between 100 and 200 would mean double the brightness as we see it.  That SHOULD be the case, but isn't.  The camera records linear, we see exponentially.  But even that is IN THE WEEDs.  The fact is, as YOU WOULD BE THE FIRST TO POINT OUT, the real world of grading is getting an image that looks right to the viewer, or conveys the feel the filmmaker wants.  That will  mean adjusting the relationship of brightness across the image... 

But but but..., you must have enough color data to fit your display gamma or you will notice where you don't have enough color, which is banding in smooth color gradients.  You can't widen your gamma (LOG) in a fixed data color space (8-bit) and assume that you will always have enough color information in your image to fully saturate a gradient.  Don was basically saying, the less aggressive LOG you use, the less chance you have to run into banding.  AGREE!!!!  And he'd know more than I about which LOG profiles do what in the real world.

Let me say this again in other words.  Extending a gamma beyond the amount of color data you have can increase banding.  Similarly, forcing more color into a narrow gamma will result in an image that has no discernable visual difference, which is why LOGs can have benefits in certain circumstances (they exploit what may be unnecessary color data). But you have to really know your shit to know when that is!  What I learned in my experiments is I did NOT know my sh_t

As for the external recorder, I was just pointing out that its benefit can affect banding, but is a separate issue from how it happens when using a LOG gamma for shooting in an 8bit space?  The OP should really be clear about the differences, right, before assuming one will fix the other?  If he was, I don't believe he would have asked the question because it's very subjective, or too variable to give an objective answer, right?

Are we friends again?

[Emanuel]

I bet bit rate will end in a solid key to avoid it as much as when "only" 8bit 422:

sample (not mine) extracted from (minute 28:30):
https://vimeo.com/114978513

[Gregormannschaft]

21 hours ago, maxotics said:

Why sell your external recorder?  Was that what I suggested?  I only said that banding can be a result of an aggressive LOG profile, as Don said, or from too little color information due to a high compression CODEC that uses 420.  We both have different roles, right?  You shoot and grade real stuff I assume.  I'd hire you over me to go shoot something.  I test stuff, a more scientific approach.  We're here to educate each other, I hope   Just so the OP can make sense of this I will explain some of this stuff in more detail, as I understand it!

Many filmmakers believe their cameras record color just like our eyes/brains do.  But they don't.  Cameras can only see in gray-scale actually.  The manufacturers put color filters over alternating pixels and then composite a "color".  But it's not really a color.  It's just a brightness value of light through a filter with the noise floor of the silicon as a reference/floor.  Too much brightness, and the value doesn't change.  The #1 confusion I see is the belief that these values have an implied relative brightness; that is, that the brightness difference between 100 and 200 would mean double the brightness as we see it.  That SHOULD be the case, but isn't.  The camera records linear, we see exponentially.  But even that is IN THE WEEDs.  The fact is, as YOU WOULD BE THE FIRST TO POINT OUT, the real world of grading is getting an image that looks right to the viewer, or conveys the feel the filmmaker wants.  That will  mean adjusting the relationship of brightness across the image... 

But but but..., you must have enough color data to fit your display gamma or you will notice where you don't have enough color, which is banding in smooth color gradients.  You can't widen your gamma (LOG) in a fixed data color space (8-bit) and assume that you will always have enough color information in your image to fully saturate a gradient.  Don was basically saying, the less aggressive LOG you use, the less chance you have to run into banding.  AGREE!!!!  And he'd know more than I about which LOG profiles do what in the real world.

Let me say this again in other words.  Extending a gamma beyond the amount of color data you have can increase banding.  Similarly, forcing more color into a narrow gamma will result in an image that has no discernable visual difference, which is why LOGs can have benefits in certain circumstances (they exploit what may be unnecessary color data). But you have to really know your shit to know when that is!  What I learned in my experiments is I did NOT know my sh_t

As for the external recorder, I was just pointing out that its benefit can affect banding, but is a separate issue from how it happens when using a LOG gamma for shooting in an 8bit space?  The OP should really be clear about the differences, right, before assuming one will fix the other?  If he was, I don't believe he would have asked the question because it's very subjective, or too variable to give an objective answer, right?

Are we friends again?

Really appreciate this level of detail and passion, I feel like I'm learning a lot from your posts even though I don't really agree with not using the LOG profiles. But, your thoughts mirror that of someone I met at a film workshop in Brighton who hated the Sony LOG profiles in 8 bit and suggested only to shoot in the Cine profiles. I get the argument, and in layman's terms, the improvement I see is that you get less macroblocking, less colour noise and in general, just a cleaner, bolder image. Would that be correct?

The only problem is that for me, the LOG profiles give you an aesthetic quality superior to that of the CINE profies, or even the standard contrasty profiles. And it's less about technical colour spaces and more about gradients. I'm happy to trade a little noise for a more pleasing image, with smoother rolloffs and shadow gradients. The initial first impression you get from the LOG profiles is a much better one that the contrasty, cleaner profiles that have this off-putting video quality to them. 

[no_connection]

The only reason you get banding is improper dithering. Nothing more nothing less. At some point you run into excessive noise at which point you need higher bit depth. A 4bit log would not get banding but it would be extremely noisy to do so. A 10bit file can be less noisy than 8bit to avoid banding. At some point you could get rid of noise and accept that the banding is so small that it's not a problem. But you would still have to reintroduce dither to avoid banding when viewing it. Just look at animated movies.

[PannySVHS]

@maxotics, Hey, awesome post! External recorder will help, as that XAVC is highly compressed. It´s not the 8bit part but the compression showing banding.

[webrunner5]

9 hours ago, no_connection said:

The only reason you get banding is improper dithering. Nothing more nothing less. At some point you run into excessive noise at which point you need higher bit depth. A 4bit log would not get banding but it would be extremely noisy to do so. A 10bit file can be less noisy than 8bit to avoid banding. At some point you could get rid of noise and accept that the banding is so small that it's not a problem. But you would still have to reintroduce dither to avoid banding when viewing it. Just look at animated movies.

I am not too sure dithering and banding go together? Dithering and noise yes. And noise in blocks not bands. Banding as I know it is based on the compression codec, ergo not enough bits.

[maxotics]

11 hours ago, Gregormannschaft said:

I'm happy to trade a little noise for a more pleasing image, with smoother rolloffs and shadow gradients.

I'm glad my post was good in a passionate way because I certainly mean them that way!  I try to say it as much as possible, that there is a place for LOG and I would, and do, use it myself.  It has a look you simply can't get any other way.  But it has a trade-off, as you point out.  My feeling is you can't really understand LOG until you understand what a standard profile is meant to do, which is match all our perceptible colors to the display gamma.   That is I DID NOT understand LOGs until I understood those fundamentals.  I was working on measuring the various noise ratios in various LOG profiles last year, but reporting my first findings here was very aggravating and lot of people pushed back.  So though I have no doubt that @Don Kotlos is giving good advice about which LOG profiles to use, it would be nice if we had more objective data.  Or better tests.  When I see test of LOG done on YouTube in a low DR scene I nearly burst a blood vessel   Right now I'm working on materials to show how RAW works, so I'm basically continuing what I did last year, but starting from the beginning!  Again, thanks for the encouragement!    

[Don Kotlos]

@Gregormannschaft Slog2 can give great results especially in high contrast conditions like outside or even during a night walk in the city when there are very bright and saturated colors. But what I have found is that once I grade it, I end up with a file that looks very close to the Cine2 profile. Since I am not a hard grader I almost always shoot Cine2 unless the I need the extended dynamic range.  

My main problems with slog2 (other than the occasional banding) are:

1. The skin tonalities look much better with Cine modes with minimal work in post. 

2. Slog2 is actually clipping the hardest when compared to any of the Cine profiles. So to make the highlights look good I have to make the clipping smoother when grading by clipping even more but more gracefully. Take a look at my post here and observe how fast the highlights drop from the clipping point:

You also see how Cine2 (& Cine3 but not shown) use the full legal range whereas S-log2/Cine1/Cine4 need to be adjusted.

This video is very informative as well and you can compare the 6400 slog2 to the 3200 cine2 (to reflect the overexposing that slog2 needs): 

 

[no_connection]

On 12/20/2017 at 2:23 AM, webrunner5 said:

I am not too sure dithering and banding go together? Dithering and noise yes. And noise in blocks not bands. Banding as I know it is based on the compression codec, ergo not enough bits.

Well this is a huge topic and kinda hard to explain in a small post, I am by no means an expert but Ill try to explain it.

In short you trade noise for bit depth.

https://en.wikipedia.org/wiki/Ordered_dithering
https://en.wikipedia.org/wiki/Dither

So you cram a higher bitdepth from the sensor into a more limited depth you see in the camera file. But that is not the whole story. by doing it right you noise shape the signal for the intended codec so that no matter what, you are not going to get any banding, and depending on the output format that might be more or less noisy. I was impressed when I examined the X-T2 files in Fusion and saw just enough noise to avoid banding but not much more. Do note that it does require some bandwidth, but so does higher bit depth.

If you compress too much you will have nether bit depth nor noise, so yes that will make banding, but emulating 10bit to 8bit dithered is not much of a problem, and the noise needed is going to be less than the sensor noise anyway, by my guess. You are still going to want that 10bit to be dithered as well.

You can blame banding on codec but it's really a bad implementation of all of the above.

So in short, noise is the solution to banding, but it takes proper implementation to get it right.

BTW if you want to experiment with dithering for video, MadVR renderer (using for example MPC-HC) have an option to dither the output in realtime with different intended bitdepths.

*edit* guess the bit depth of the attached images