Dragon compared to 35mm Vision3 200T

[Ivan McCullough]

.... Feed a CTO filter 5600ish white light and it'll output 3200ish k light. As far as I'm concerned that's just an optical LUT. ......

Ahem, CTO, but I agree with everything else you say. Gotta love that best part of a century of 'colour science' ;-)

edit, weird correction matrix glitch batman

 

[Hrvoje Simic]

Try this one.

(re-tweak on highly compressed grabs, yadda yadda...)

Yeah I am eager to see a debayer without the green cast someday. It's been dogging RED for half a decade now. Was really hoping we could put that behind us completely with Dragon. I highly doubt it's the OLPF though. I can't see a reason why colorscience couldn't correct it though. So hopefully DragonColor/DragonColor?2? nails it once and for all.

Please take into consideration that viewing graded material is not just viewing an imaging device/medium characteristics but colorist's & DP's influence as well.

The way this film stock is graded is just one and this is just one film stock. Attempting to match it completely from the get-go by a digital camera manufacturer would be shooting a moving target relative to various expectations and points of reference.

 

[Mark K.]

And I guess that's why I still love film. By almost every objective measurement, Dragon should match 5213 - but there's just something magical about that lovely dancing celluloid strip that current digital sensors still can't match.

I wonder if vibrating sensors (ala Penelope Delta) and random pattern sensors (ala Fuji's mirrorless cameras) might be the answer?

 

[Martin Stevens]

And I guess that's why I still love film. By almost every objective measurement, Dragon should match 5213 - but there's just something magical about that lovely dancing celluloid strip that current digital sensors still can't match.

I wonder if vibrating sensors (ala Penelope Delta) and random pattern sensors (ala Fuji's mirrorless cameras) might be the answer?

Film Convert is perhaps an answer for you.

 

[Mark K.]

Film Convert is perhaps an answer for you.

It's already an intrinsic part of my workflow (galling though it is to have to fake it!).

 

[Vadim Bobkovsky]

but there's just something magical about that lovely dancing celluloid strip

aka random grain?

 

[Mark K.]

aka random grain?

I'm sure that's part of it. It's a subtle component, but it's certainly there.

 

[Paris Remillard]

The film skin tones are just hard to beat. The green in the temple, under the eye under the cheekbone, around the mouth, along the collarbone on the Dragon image... It's just not pleasant to me. And like others have mentioned, not terribly easy to grade out all of the time. I'm not a camera or software engineer, so I speak with no deep understanding of what exactly is involved, but I'd like to understand why a film emulsion can be made to avoid those color casts, yet digital sensors/software combinations, and in my experience, particularly Reds, have so much trouble in this area.

 

[Hans von Sonntag]

The film skin tones are just hard to beat. The green in the temple, under the eye under the cheekbone, around the mouth, along the collarbone on the Dragon image... It's just not pleasant to me. And like others have mentioned, not terribly easy to grade out all of the time. I'm not a camera or software engineer, so I speak with no deep understanding of what exactly is involved, but I'd like to understand why a film emulsion can be made to avoid those color casts, yet digital sensors/software combinations, and in my experience, particularly Reds, have so much trouble in this area.

Before I jumped on the Red wagon 35mm and S16mm was my acquisition format. After the shoot the stock got to the lab and the neg. was delivered to the post house where the Ursa or Spirit resided. Negative has an orange masks and needs a WB in the telecine before one can start to grade. There was never a single shot that did not need a tweak her or there. Cast in any directions be it orange to blue or magenta to green were the norm. I always had an eye on the batch number making sure that we had maximum consistency in colour rendition. It is a misconception to believe that digital alone has a problem with casts. There are colour sciences, different sensors or film emulsions, gamma curves, LUTs and many combinations in all possible directions. There is no right way to skin the cat.

In this particular case the colourist gave the film-originated clip a tendency to magenta, and the Dragon shot clip shows a litte green cast. Could easily be the other way round. Last Wednesday I directed a shoot for a TVC where we had a whitish overexposed BG. One of the HMI showed a visible green cast. I could swear that this cast got worse over time. Or not. Both Epic cameras reacted slightly different in look and feel. It's the angle the camera is taken, it's the sensor's temperature, it's the day the camera was build, there are tons of reasons.

The only way to evaluate meaningfully sensors and emulsions is to play with the material in a well equipped grading suite by your self, or better with other experts in a discussion while everyone is allowed to touch the levers.

Hans

 

[Gavin Greenwalt]

It's more complex than "Magenta cast" or "green cast" though. I can't look at the re-graded-graded shots on my home screen which is garbage and I can't see any color variation at all like on a well calibrated screen. But speaking generally, the 'problem' is more subtle where you can have both more magenta and more green simultaneously while say blues match perfectly. Honestly I believe the "problem" is that RED subjectively prefers a "neutral" color palette that's more photo-realistic. Kodak has a more aggressively subjective spin on reality. Different wavelengths and combinations of colors just excite different mediums differently. IR contamination is a great example of color shifting without a way to simply white balance out the problem.

Nobody disagrees that there is variability in film. But the variability generally is consistently different from the variability of RED etc. Similarly there is variability in Cooke S4s but they're still generally warmer than a Master Prime.

 

[Paris Remillard]

Exactly. It's the non uniformity that's the problem.

 

[Marc Wielage]

It's more complex than "Magenta cast" or "green cast" though. I can't look at the re-graded-graded shots on my home screen which is garbage and I can't see any color variation at all like on a well calibrated screen. But speaking generally, the 'problem' is more subtle where you can have both more magenta and more green simultaneously while say blues match perfectly. Honestly I believe the "problem" is that RED subjectively prefers a "neutral" color palette that's more photo-realistic. Kodak has a more aggressively subjective spin on reality.

Where's Alexa on that color palette?

BTW, call me crazy, but I have my home screen calibrated every so often. My eyes bug out and my head explodes if I don't. Very messy when that happens, plus we have to clean all the blood and brain matter off the ceiling.

Nobody disagrees that there is variability in film. But the variability generally is consistently different from the variability of RED etc. Similarly there is variability in Cooke S4s but they're still generally warmer than a Master Prime.

I've often noted that when you change absolutely nothing in lighting or exposure, if you zoom in on (say) a 10-1 zoom lens, you get one color, and then when you zoom all the way out, you get a slightly different color, particularly in terms of white balance. What I generally see is a yellow/blue offset -- maybe yellowish in telephoto, colder on the wide. So the lenses absolutely, positively add a few drops of color. It's measurable on the scopes and very visible with test charts. Even worse when you add multiple cameras at different angles.

I think the color correction issues between digital and film can be balanced out by about 90%, but there are some indefinable differences that are hard to account for.

 

[Hrvoje Simic]

The film skin tones are just hard to beat. The green in the temple, under the eye under the cheekbone, around the mouth, along the collarbone on the Dragon image... It's just not pleasant to me. And like others have mentioned, not terribly easy to grade out all of the time. I'm not a camera or software engineer, so I speak with no deep understanding of what exactly is involved, but I'd like to understand why a film emulsion can be made to avoid those color casts, yet digital sensors/software combinations, and in my experience, particularly Reds, have so much trouble in this area.

Unless you are observing raw or uncompressed material on a calibrated monitor, all bets are off. What you are seeing here is compression on top of compression on top of compression. This is where splotches often appear because compression throws a lot of data away, leaving limited color palette to be displayed.

The right example I posted is additionally pushed on top of that web compression, and this process further stresses the limitations of compression, revealing something which should not be there in raw or uncompressed material. If you zoom in you might see those weird tonalities in macro blocks which are a consequence of high compression.

If you observe these images on a couple of different non-calibrated screens you might notice that those questionable greener tones are differently shown between them, maybe even on different places. Computer or mobile device screen is additionally "pumping" or "stretching" colors, showing what may not be there in original files. Any gamma, color balance or gamut offset from your screen further enhances the inaccuracies of compression, by further "drifting away" from the original, and doing so in digital, bit depth dependent stepping manner.

For more objective impression raw or uncompressed grabs would be required. Highly compressed examples like these are useless for observing details and focusing on them based on this is a waste of time. These examples may serve only to provide a rough impression.

 

[Gavin Greenwalt]

Where's Alexa on that color palette?

My impression has always been that Alexa goes more neutral than film (which is to say a little more green) but it's a more consistent color cast than say RED which has divergent magenta and green cast in the same frame.

 

[Hrvoje Simic]

Honestly I believe the "problem" is that RED subjectively prefers a "neutral" color palette that's more photo-realistic. Kodak has a more aggressively subjective spin on reality.

Why is that a "problem" ?

Digital UHD RAW camera. Relatively neutral starting point.
Upping the pixel count > more realistic details
More neutral color palette > more realistic tones

 

[Ivan McCullough]

Why is that a "problem" ?

You may as well direct this question at the majority of shooters who prefer how the Alexa looks or 5219....

 

[Hrvoje Simic]

You may as well direct this question at the majority of shooters who prefer how the Alexa looks or 5219....

Although those looks are greatly affected by DoPs and colorists, the main difference is in philosophy.

Why mimicing other cameras and stocks from the get-go with a RAW imaging device if there are so many new looks which can be made and these can be crafted as well ? To avoid post work ?

 

[Ivan McCullough]

Although those looks are greatly affected by DoPs and colorists, the main difference is in philosophy.

Why mimicing other cameras and stocks from the get-go with a RAW imaging device if there are so many new looks which can be made and these can be crafted as well ? To avoid post work ?

Because nothing is neutral.

One companies log is not another companies log.

one companies sensor is strong in some areas and weak in others and vice versa. Someone in the company makes a choice and decides what is 'neutral' or 'log' or 'RAW' and then still has the audacity to talk of colour science and olpf's as if they matter in one conversation but when it is suggested that the camera has a particular 'look' out of the box gets all defensive.

The consensus seems to be that Alexa looks nice, that 5219 looks nice, that RED has issues. Maybe RED should pay attention.

 

[Nick Morrison]

You may as well direct this question at the majority of shooters who prefer how the Alexa looks or 5219....

The Alexa is great, but I know for a fact DIT's in NY are really starting to enjoy working with Dragon (even some that had previously not been sold on MX). In particular, the strong color and clean look straight out of the camera is really winning people over. I firmly believe Dragon is slowly starting to re-calibrate how a portion of the industry views RED.

Remember, RED came out the gate before Alexa, with an unusual philosophy and workflow, and not everyone has jumped on board. I find it interesting that over the years the industry has started to navigate slowly over to RED's original position in 2007: 4K, RAW, etc.

Dragon however, addresses the more human issues like "color" and "look" that the technical achievements before never could. Which is exciting, and reveals a lot about how RED has grown as a company.

Looking forward to later this year and especially next...when lots of DRAGON shot content really starts to come out.

 

[Hans von Sonntag]

Remember, RED came out the gate before Alexa, with an unusual philosophy and workflow, and not everyone has jumped on board. I find it interesting that over the years the industry has started to navigate slowly over to RED's original position in 2007: 4K, RAW, etc.

Just for the record, Arri had the D20 with a S-35 sized sensor on their rental shelves before Red's RedOne entered the market. The SI2K was in 2007 available with compressed RAW and Dalsa had long before Red with the Origin a 4K camera available to rent. This all does not belittle Red's achievements.

Hans