Shooting SuperFlat Help?

[Dave Pierpont]

Hi guys here at work reading up on shooting SuperFlat with the 550d and I have a few questions. I have done color correction/grading throughout my 4 year college career however, I never really went to far in depth into Apple Color. My question is, will I be able to get the most out of the SuperFlat picture style just by bringing a file into FCP and using the color corrector there?

I know that using Color will give me better results and more options, but just by using Color Corrector in FCP will this allow me to achieve the dynamic range after messing with the colors on a SuperFlat image?

[Ian Holb]

The SuperFlat craze needs to die.

The author of that picture style admitted he got better results just using Neutral with contrast and sharpness turned down. Superflat and its ill-brethren ExtraFlat have done more damage than good.

[Jerry Porter]

What he said ^^^^^^^ you will never get the skin tones right in super flat. Neutral and turn it down a notch or two.

[Dave Pierpont]

thanks for the quick responses guys. After looking at some of the transformations people have got out of the SuperFlat look it seems like you guys are right. Neutral with those adjustments you mentioned sound good. Anyway locked here at work will run some tests when I get back home.

Thanks again

[Perrone Ford]

I've always thought that the Superflat goes too far, but neutral didn't go quite far enough for some of my needs. I have become rather enamored of the Marvel Cine pre-set. Colors are excellent, it's not TOO flat, but boy did it make a difference over standard when I needed to get some sky and horizon shots. I think I'll be sticking with it.

[Jerry Porter]

I have Marvel loaded into mine as well but I find for most work the Neutral setting gives me the best out of the camera shots.

[Perrone Ford]

Originally Posted by Jerry Porter

I have Marvel loaded into mine as well but I find for most work the Neutral setting gives me the best out of the camera shots.

I will agree with you there. I tend to be setting up for grading in post, and in that regard I prefer Marvel. If I am shooting for something straight out of the camera, neutral is probably best. But I have kodachrome and Velvia loaded just in case I need to push things HARD the other way.

[Jerry Porter]

I have never used the Koda setting for video, but I love it for stills. I think we have the same things loaded on our cameras....

[Khoi Pham]

Originally Posted by Perrone Ford

I've always thought that the Superflat goes too far, but neutral didn't go quite far enough for some of my needs. I have become rather enamored of the Marvel Cine pre-set. Colors are excellent, it's not TOO flat, but boy did it make a difference over standard when I needed to get some sky and horizon shots. I think I'll be sticking with it.

Same here, I think superflat is too flat and neutral with contrast all the way down is still crushed the black a little so I made my own, kind of in between those 2.

[David W. Jones]

Many styles or looks have been created over the years for different cameras by people that haven't a clue how to even read a waveform monitor or vector scope. And then blindly downloaded and used by people because of a post on the internet claiming perfection.

[Evan Donn]

David's absolutely right - I just just looked at some samples from the Marvel Cine preset and it has the same problem as superflat. I keep seeing people saying they use these so they can shoot for the grade but I don't think they understand what's really going on behind the scenes - you are losing information that cannot be regained in grading.

Any preset that's created in Canon's picture style editor and uses the curve control to modify the contrast is likely to have these problems unless the modification is extremely minimal. There is no control over the bias through the points you add to the curve - and the default bias is strong and horizontal. Instead of getting a smooth curve that transitions linearly from point to point you tend to get flat spots at each point - if you add more than one or two points it starts getting worse and worse. The effect of this is that values on either side of each point get squashed down to the same value - and you get patches in your image where a range of shades of gray all end up as nearly the same values. If we were storing the resulting data as 14 bit (I believe this is the internal accuracy) you might be able to recover some of the data later when grading, but when it's converted to 8 bit you're collapsing 64 close shades of gray down to one - and that data is lost forever.

I don't know if this is a limitation of the camera hardware/firmware or simply the picture style editor itself. If it's just the picture style editor there's the possibility someone could build a better tool that would let us build 'flatter' profiles that still retained a linear range of midtones, or even do things like profiles optimized to retain the greatest data in the range of properly-exposed skintones. As it stands now though the tools just aren't there and it's much easier to produce a worse image than a better one with custom picture profiles.

[Evan Donn]

And to address the original post a bit...

Originally Posted by Dave Pierpont

I know that using Color will give me better results and more options, but just by using Color Corrector in FCP will this allow me to achieve the dynamic range after messing with the colors on a SuperFlat image?

Regardless of which picture profile you shoot with you should take the time to learn Color - it's goes so far beyond what you can achieve in FCP it's amazing, and once you get past the initial learning curve it's also much faster to work in than FCP - except that you'll probably spend more time on the color correction just because of the options you now have.

[Perrone Ford]

Originally Posted by Evan Donn

David's absolutely right - I just just looked at some samples from the Marvel Cine preset and it has the same problem as superflat. I keep seeing people saying they use these so they can shoot for the grade but I don't think they understand what's really going on behind the scenes - you are losing information that cannot be regained in grading.

While I can't argue this point, I think we need to look at it in the proper context. The decision to shoot on an HDSLR in and of itself represents a significant compromise from the beginning of production. This is not RAW. It's not 10bit, 12bit, or 14bit capture. We are not creating DPX files on the back end.

Similarly, in production we make the choice of what to lose and what to keep each time we place a light, or determine framing. The underlying question during production is always the same. What must I preserve, and what can I afford to let go. The decision to shoot with ANY picture profile, even the standard or neutral ones in the camera, mean that certain information will be kept, and other information will be lost. It's inevitable. And we must choose what we keep and what we throw away.

To be fair, what we are calling "grading" here might be better characterized as setting a "look" like making a LUT. We do what we can in post to achieve the best image we can. Superflat, and Marvel allow us to capture information that standard and neutral do not. And we can manipulate that data in post to give a "look" that we consider pleasing. Yes, information is lost. But if the we achieve our goal of a pleasing image, then I don't think it matters all that much that we've thrown some out. Scanning film at 2K instead of 8K throws away a tremendous amount of information. Far more than what is lost in the decision to shoot with one of these pre-sets. And yet, it's done every week in Hollywood.

So let's not lose sight of the big picture. Make the in-camera choice that suits your needs, and shoot something that looks good. Really, it's that simple.

[Evan Donn]

I understand what you're saying Perrone, and I agree - naturally any of these settings require us to make decisions about what to keep and what to discard. I'm assuming that's our starting point though, and my point is that attempting to change the curve significantly in the picture profile editor discards midrange information in a non-linear manner that cannot be controlled sufficiently by the user - so they may not realize what exactly is being discarded.

Even with only two points of control - say one at each end to stretch the highlights and shadows - you end up with your midrange compressed in a non-linear manner. You'll have a flatter spot in the curve midway between the two points - which means a range of slight tonal variations around that point all get shifted to the same value. Once they are the same value there's no way to differentiate them later in grading. Adding more points only makes things worse, turning the curve into a weird curved stair step.

[Perrone Ford]

Understood and hopefully others will understand the ramifications of making these choices. This is why I felt superflat went too far.

I've been testing the Marvel Cine now for a while on on manner of subjects, from faces indoor and outdoor, to nature scenes, and architecture. I am comfortable with the tonal range it gives me. I will be setting up some indoor interview tests with it in a week or two, as well as some outdoor model photography and I will see how it responds. If I am ok with it there, I'll continue to use it.

I still have the internal camera presets to use should I feel the need. And this is one of the wonderful things about these user presets. They are merely tools in the toolbox. You can choose to move away from them at any moment if desired.