An easier way to grade log footage - Tutorial

[John Brawley]

You didn't call me out John. You insulted me.

I said put up or shut up.

And I said it WELL AFTER you refused to elaborate on these quotes or use images to explain what you're seeing...

What I see is a whole bunch of zones in the color gamut of BM where we have to jump through a lot of hoops in order to get something faintly normal compared to other parts of the gamut and even other cameras. Its as if certain zones were 12 or 14 bit and others were 8 bit within the same image ! This is how I know we are on the edge of the color gamut. Unfortunately skies and the blush in faces are often right there. The image falls apart from a color point of view and the transitions are not smooth, they are washed out and become "colored" rather than enriched. I most commonly see this in skies being cyan rather than blue and reds being orange...until they are suddenly magenta. Very little room to find a range of reds. This has a direct impact on faces, which often lack nuance or transitions.

And

Well every colorist I know feels the same. There are issues all over the gamut. The color is thin and falls apart even from Raw. Most of what I see here are slightly desaturated images, which users here now feel is normal or "real". But the images are actually quite dull. This is not a style thing. It is a problem that issues from color depth, or rather the lack of it in certain parts of the spectrum and the way the sensors are responding and mapped. The truth is there is crossover when balancing out skies that remain persistently cyan rather than blue, causing other parts of the spectrum to distort when corrected. I can spot a video shot outside on a BM camera from a mile away because of it for instance. Skin tones also often lack nuance which shows up as a lack of saturation in the cheeks, the lips or if someone blushes...so people become sallow with a single wash of color, rather than having true seperation of skin tones.

I'm all for learning and being shown otherwise. Please enlighten us with IMAGES instead of opinion.

They don't have to be your images, again, you ignored that idea too.

When the guy pretty much leading the team responsible for the latest generation of Blackmagic's look out of camera asked you the same question you again ignored him.

If you had a genuine concern then you had, and still have (!) a great opportunity to have THE VERY GUY who could address your view and make a change in this product listening and waiting for you to demonstrate and back up your view.

You ignored him totally.

So yeah, I said put up or shut up. It hurt your feelings apparently.

You seem disingenuous when you posit a view that isn't widely shared, ignore the lead developer who could address the concern you raise, then sulk about being insulted after not having to explain yourself.

I don't think you get to get away with that on on a forum dedicated to the Blackmagic camera range. I'm going to call you out for that because what you're posting is an opinion and yet you're not actively seeking to help, move forward or have any way of dealing with it for the better. You just want it known...

Please prove me wrong, I'd love to be shown up. Here's your chance for revenge if you're insulted.

For me you definitely crossed a line.
Not sure you will understand what I am getting at, but I hope you do.

I'm actually challenging your assertion in this "court" of public opinion. If you're going to refuse to use images, then I'm forced to use words to robustly call upon you to prove your claim or have it considered mis-informed.

I'm really happy to be proven wrong and you have very simple means to do so. Post some images by way of example.

By the way, you're very welcome to report any post, it will go to moderators (not me) and you can have your opinion heard.

I've never censored any discussion in my role as a moderator. I get to trade here on my reputation. You don't in my view because we don't know who you are and you offer nothing other than an opinion.

I'm all for genuine discussion. I learn when I'm wrong and I love being shown there's a better way. So show me.

JB

[DPStewart]

Your opinions and experience mean nothing if you can't back it up with examples for the rest of us. For all we know you're a hack, or you're one of the best in the world. We don't know so your opinion has no weight if you don't post footage or examples of your work and the issues you describe. That's just how it is. All opinions are not equal. Esepcially when you go against what others have experienced here. You're right, this isn't a court of law and you don't get the benefit of the doubt and are not innocent until proven guilty. Statistically you're more likely a hack, and until proven otherwise your opinions are judged as such.

Yes. Unfortunately this IS pretty much how a public forum is going to end up operating. Don't worry about it though - it's not personal.

Now, I know you're probably thinking "But Dane.. we always believe YOU without question!", but let's be honest here... that's just because my mother is sending you checks.

[jimagine]

I learn when I'm wrong and I love being shown there's a better way. So show me.
JB

As a creative director I just want to say that this level of evolved thought is beyond a lot of people, but thank you for it.

If we could get people in government to learn this, the world wouldn't be what it is right now.

[yoclay]

I'm all for learning and being shown otherwise. Please enlighten us with IMAGES instead of opinion.

They don't have to be your images, again, you ignored that idea too.


JB

Ok, so let me get this straight. You want me to post examples with images that I haven't worked on and don't know what the original footage looks like ?

Right.

That sounds like a losing battle to begin with.
But something I considered and had previously rejected.

For those of us who are trying to understand your position better, could you demonstrate with publicly available footage? I'm sure plenty of people here would happily send you dngs.

Keys,
I had thought about using some of the footage in the subforum with people's work here to illustrate what I have been seeing, however, I am not sure that most people would appreciate something they have put their blood, sweat and tears into, being used to illustrate "issues". This can be delicate, because in my experience no one wants to have the fruits of their creativity examined in this way. And rightly so,they have worked hard on it. Typically then too they become defensive and say they "wanted it to be that way". However, I often have my doubts because people tend to remain unaware of the limitations before them unless they have a broad range of experiences.

It’s challenging to speak about work that I have not personally consulted on.

However, the best I can do is to show you what I am seeing.

Typically for me, most BM images are pretty easily identifiable. They are PASTEL.
Very few primaries, mostly complimentary colors.

Juan, has done a good job of illustrating by his 4 step transform process, that there is definitely some hoops to jump through. His process clearly illustrates why I think BM needs to revisit it's "color science". That process he has shown, shouldn't be necessary.

So, let's start with skies. i see a tremendous amount of magenta/green(or cyan) crossovers. Those two complimentary colors opposing each other in the same sky. What I don't see a lot of are BLUE skies. There is a problem for me in the color science/palette of BM cameras on this level. TYpically people are trying to find a very delicate balance of magenta/cyan to find blue, but it is never really blue and it falls apart quickly if saturation is added to the image. So what is the solution for most members ?

Well, they make skies which are so strongly dominated by cyan that there is no chance of the dreaded magenta showing up. I have included a magenta sky image to show how much more disturbing it is.

The issue is a shift in color over certain density levels. This makes it extremely difficult to achieve an even sky or primary colors such as a pure-ish blue or red. Some will say it is the palette of BM cameras. However, I view it as a limitation. I think a real palette should be the result of a choice by the artist, not the manufacturer. And no, I don't see this degree of an issue with an Alexa or even many Canon cameras. Sony has other issues, but not this one.

I am not saying that a blue cannot be achieved in the sky of a BM image (I am sure some of you will go out and prove me wrong - hopefully) , but what I am saying is that I have to jump through far too many hoops to get there. This is extremely time consuming and this is a common experience that is noted by many colorists I know.

So here are some images to illustrate what I am speaking of. It took me less than 5 minutes of searching here. Cyan rather than blue skies dominate here. Again, these are not my images or even ones I have worked on, however, they do illustrate what I see regularly from BM cameras. My apologies to the hard working people here who have produced them, but I have been pushed to show examples of what I am seeing on a regular basis, so I am responding with what is most easily at hand.

I have written in white on the images and below them. Again, these are illustrations of what I am seeing regularly and speaking about, not things I have worked on.

Screen Shot 2017-12-14 at 08.14.12.jpgScreen Shot 2017-12-17 at 13.04.03.jpgScreen Shot 2017-12-14 at 08.11.11.jpgScreen Shot 2017-12-14 at 08.11.30.jpgScreen Shot 2017-12-14 at 08.11.38.jpgScreen Shot 2017-12-14 at 08.15.20.jpg

[Steven Abrams]

I've seen the same magenta or cyan skies too.

Look at this, magenta at the edges of the sky, cyan in the middle. Magenta shadows.



Oh wait, this is Alexa Mini.



Well what about this magenta sky?



Oh, Amira.

Uhhh... this cyan one!



Damn, Alexa. Okay, but this one has magenta in the sky next to the left mountain, and then turns cyan as well.



Alexa??

Cyan?




What about some weird ass color with a magenta stripe through it?



...And these took me less than 5 minutes to find also. None of these skies look "normal" to me either.

[yoclay]

You know it's almost like this guy is I dunno, full of shit.

Ummm. Sorry, but I have tried to respond exactly as JB and KEY asked me too, with other people's images, as I cannot use the footage from my workplace. This comment above is out of line and I would appreciate that a moderator deals with this. This is precisely the kind of attitude which I am talking about. It's defensive, aggressive and uncalled for.

[shijan]

Ok, so let me get this straight. You want me to post examples with images that I haven't worked on and don't know what the original footage looks like ?

Right.

That sounds like a losing battle to begin with.
But something I considered and had previously rejected.



Keys,
I had thought about using some of the footage in the subforum with people's work here to illustrate what I have been seeing, however, I am not sure that most people would appreciate something they have put their blood, sweat and tears into, being used to illustrate "issues". This can be delicate, because in my experience no one wants to have the fruits of their creativity examined in this way. And rightly so,they have worked hard on it. Typically then too they become defensive and say they "wanted it to be that way". However, I often have my doubts because people tend to remain unaware of the limitations before them unless they have a broad range of experiences.

It’s challenging to speak about work that I have not personally consulted on.

However, the best I can do is to show you what I am seeing.

Typically for me, most BM images are pretty easily identifiable. They are PASTEL.
Very few primaries, mostly complimentary colors.

Juan, has done a good job of illustrating by his 4 step transform process, that there is definitely some hoops to jump through. His process clearly illustrates why I think BM needs to revisit it's "color science". That process he has shown, shouldn't be necessary.

So, let's start with skies. i see a tremendous amount of magenta/green(or cyan) crossovers. Those two complimentary colors opposing each other in the same sky. What I don't see a lot of are BLUE skies. There is a problem for me in the color science/palette of BM cameras on this level. TYpically people are trying to find a very delicate balance of magenta/cyan to find blue, but it is never really blue and it falls apart quickly if saturation is added to the image. So what is the solution for most members ?

Well, they make skies which are so strongly dominated by cyan that there is no chance of the dreaded magenta showing up. I have included a magenta sky image to show how much more disturbing it is.

The issue is a shift in color over certain density levels. This makes it extremely difficult to achieve an even sky or primary colors such as a pure-ish blue or red. Some will say it is the palette of BM cameras. However, I view it as a limitation. I think a real palette should be the result of a choice by the artist, not the manufacturer. And no, I don't see this degree of an issue with an Alexa or even many Canon cameras. Sony has other issues, but not this one.

I am not saying that a blue cannot be achieved in the sky of a BM image (I am sure some of you will go out and prove me wrong - hopefully) , but what I am saying is that I have to jump through far too many hoops to get there. This is extremely time consuming and this is a common experience that is noted by many colorists I know.

So here are some images to illustrate what I am speaking of. It took me less than 5 minutes of searching here. Cyan rather than blue skies dominate here. Again, these are not my images or even ones I have worked on, however, they do illustrate what I see regularly from BM cameras. My apologies to the hard working people here who have produced them, but I have been pushed to show examples of what I am seeing on a regular basis, so I am responding with what is most easily at hand.

I have written in white on the images and below them. Again, these are illustrations of what I am seeing regularly and speaking about, not things I have worked on.

Yes, colors in RAW are randomly shifted and undersaturated. But it is not a huge problem. Just use ColorChecker to get perfectly saturated starting point from UM4.6k, BMCC, BMPCC, BMMCC cameras. All other camera manufacturers just use factory calibration correction which is build into the footage or into the input profile and may be not to perfect as well. BM cameras apply similar factory based color correction when you shoot ProRes and it is not perfect.
With RAW you can do your own color science and don't care about anything else.
There is so many color information in RAW files that you can recover colors from 99.9% desaturated image. You can see my tests in this thread http://www.bmcuser.com/showthread.php?21154

Samples from my own BMMCC tests:

[Nomad]

There is no perfect camera. If you shoot a very accurate color chart and check with a vectorscope, not one single camera has all colors fall into the correct target with whatever color matrix the manufacturer passed along with it.

It all starts with cameras 'seeing' with different sensitivity curves of only three colors than our eyes – otherwise there would be no metamerism. Yes, humans differentiate only three color channels too, but they are NOT RGB and the can't be easily reproduced with todays technology.

Filters on a Bayer pattern for example are less than perfect for differentiating all possible shades of color in nature the way our eyes do. A perfect camera would be prohibitively expensive (and bulky). You might be able to construct something like that in a lab, but you'd not be able to sell it.

So, every manufacturer needs to find a balance between cost, precision and tastes of the targeted market. An Arri is great with skin tones (and the BM cameras come quite close to it). But it has shortcomings in the cyan/blue range too. A Sony has pretty good blue/cyan/green differentiation, but a red rose for it is red, red or red. No blueish red or warm red. Just red (well, the Venice might be different). Canon takes good care of the yellow/orange/red range, but skin tones can be critical, just a bit too 'sunny'. And so on…

There is a reason why DPs tend to choose a camera for a project. And they did choose a film in the analog era, like Fuji vs Kodak (Fassbinder adored Agfa colors for their 'painterly' look of the shadows, though.) Finally, it's the task of the colorist to bring the vision of the director and DP to the screen. It's rarely naturalistic in color.

But good differentiation of shades – which is so hard to achieve for all colors at the same time with todays technology – is very important for a rich image and suspension of disbelief. More important than precision of color, I'd say, since complementary contrast and our eye/brain system's capability of adaptation in a dark cinema can easily fool our memory about the 'right' color. It's skin tone where we are at best, but the rest of the rainbow? I'd challenge you to present you two samples of red short after one another (but not next to each other) and you'd all have difficulties to tell if they are the same or not. Humans can be trained for very high precision when identifying a musical note, but even trained colorists (or professionals in printing or dyeing) have no absolute memory for the shade of a color. These points have been proven in extensive testing by physiological experiments.

Maybe this is all commonplace to you, but I want to bring home one point: there is no perfect camera!
And every manufacturer has to find a balance which is partially subjective…

BTW, you can always take a crop of a scene plus some shots of a good waveform and vectorscope to bring home your criticism without revealing the scene as a whole.

[yoclay]

Yes, colors in RAW are randomly shifted and undersaturated. But it is not a huge problem. Just use ColorChecker to get perfectly saturated starting point from UM4.6k, BMCC, BMPCC, BMMCC cameras. All other camera manufacturers just use factory calibration correction which is build into the footage or into the input profile and may be not to perfect as well. BM cameras apply similar factory based color correction when you shoot ProRes and it is not perfect.
With RAW you can do your own color science and don't care about anything else.
There is so many color information in RAW files that you can recover colors from 99.9% desaturated image. You can see my tests in this thread http://www.bmcuser.com/showthread.php?21154

Samples from my own BMMCC tests:

Hi Shijan,
Those are excellently balanced pictures. Some of the best I have seen.


However, I am not sure that a color checker is the answer (I myself own 5 and prefer the SMPTE charts for skin tones).
The reason is because you are shooting them in an ideal condition with even diffused and controlled lighting in a very limited contrast range.
It is much more simple to match cameras from various manufacturer's under those conditions.
The issue is that when charts are used in uneven lighting conditions, or when the subject is in very different light than the rest of the scene, the performances can be quite different.
Skies react in an entirely different way than the reflective aspect of a chart. They are quite transmissive and the issues I am seeing have to do with color shifts over an extended contrast range.
A color chart essentially functions well in the center of the triangular gamut, however what concerns me is on the edges of the triangular spectrum of the color space for any particular camera and how those are mapped in this case.

When you refer to "the random shifts" which come with desaturation, I also don't think these colors are as easily recoverable as you state. I feel that I am having to "jump through too many hoops" to restore certain aspects of the color spectrum.

I know that that all camera manufacturer's create input profiles for the intrepetation of their data. That is not in question.
What I am asking is precisely that they revisit that input profile, because in my estimation there are too many issues there currently.

[yoclay]

There is no perfect camera. If you shoot a very accurate color chart and check with a vectorscope, not one single camera has all colors fall into the correct target with whatever color matrix the manufacturer passed along with it.

It all starts with cameras 'seeing' with different sensitivity curves of only three colors than our eyes – otherwise there would be no metamerism. Yes, humans differentiate only three color channels too, but they are NOT RGB and the can't be easily reproduced with todays technology.

Filters on a Bayer pattern for example are less than perfect for differentiating all possible shades of color in nature the way our eyes do. A perfect camera would be prohibitively expensive (and bulky). You might be able to construct something like that in a lab, but you'd not be able to sell it.

So, every manufacturer needs to find a balance between cost, precision and tastes of the targeted market. An Arri is great with skin tones (and the BM cameras come quite close to it). But it has shortcomings in the cyan/blue range too. A Sony has pretty good blue/cyan/green differentiation, but a red rose for it is red, red or red. No blueish red or warm red. Just red (well, the Venice might be different). Canon takes good care of the yellow/orange/red range, but skin tones can be critical, just a bit too 'sunny'. And so on…

There is a reason why DPs tend to choose a camera for a project. And they did choose a film in the analog era, like Fuji vs Kodak (Fassbinder adored Agfa colors for their 'painterly' look of the shadows, though.) Finally, it's the task of the colorist to bring the vision of the director and DP to the screen. It's rarely naturalistic in color.

But good differentiation of shades – which is so hard to achieve for all colors at the same time with todays technology – is very important for a rich image and suspension of disbelief. More important than precision of color, I'd say, since complementary contrast and our eye/brain system's capability of adaptation in a dark cinema can easily fool our memory about the 'right' color. It's skin tone where we are at best, but the rest of the rainbow? I'd challenge you to present you two samples of red short after one another (but not next to each other) and you'd all have difficulties to tell if they are the same or not. Humans can be trained for very high precision when identifying a musical note, but even trained colorists (or professionals in printing or dyeing) have no absolute memory for the shade of a color. These points have been proven in extensive testing by physiological experiments.

Maybe this is all commonplace to you, but I want to bring home one point: there is no perfect camera!
And every manufacturer has to find a balance which is partially subjective…

BTW, you can always take a crop of a scene plus some shots of a good waveform and vectorscope to bring home your criticism without revealing the scene as a whole.

Thank you for a very thoughtful post.
I am in agreement with virtually everything you have stated.

Totally agree.
There is no perfect camera.
I deal with imperfect cameras and imperfect human beings every day.
I myself am an imperfect being, just like everyone here.
I know exactly what you are talking about in terms of color memory for instance.
I began my career correcting color in labs in the 80's.
I know for sure that we don't have color memory and function by comparison.
If we tried to come in and correct purely from memory something we worked on the day before, it was laughable how different the correction was.

However, in order for things to improve there still needs to be regular critical evaluation and responses.

One of my issues with BM has to do with the unneveness in color in skies. It is very typical to see a magenta/cyan crossover within the sky.
Which means I have to do all kinds of things to create uniformity in the sky, sometimes even desaturating or throwing away info in order to do this.
While I agree that Arri can go more towards the cyan (though blue is more easily attainable), it is very clear that a much more uniform result is obtainable in a shorter period of time.
I don't experience the same degree of color crossover as the density shifts in that portion of the image.
This is likely because Arri uses a pseudo HDR technology for the highlight rolloff and retains approximately a 1.5 stop gradation/transition in it's highlight values.

I really believe BM needs to look at the responses of their cameras in this part of the spectrum again.
Something is happening where blue is being interpreted as cyan and it would be in everyone's benefit to shift/enlarge/work-on this part of the spectrum.