Dragon uncaged...

[Timothy Eaton]

Nice work Hrvoje! With all due respect to Gavin, whose posts I often find informative, comparing "looks", capture technology, DR, subtlety, feel, or really any variable between two different films is hopeless if your point of comparison is a couple of jpegs lifted from the internet. And if the point of comparison is one's impression at the time of viewing the two films, fairly useless to anyone else because it is based heavily on the viewer's memory, projection system, choices in post, etc. If the point of comparison is DVDs or Blu-rays, this is so far downstream from the original capture of the images, with so much manipulation in-between, it's ridiculous to talk about what the capture technologies rendered at the time of capture.

Back in the day, every time Eastman came out with a new film stock, we (fellow cinematographers and I) made extensive side by side tests of the stocks, shot under identical circumstances, projected under identical conditions. It's the only way to really examine the capture technology. Putting up internet jpegs is, by comparison, meaningless.

I think this is much the point Mark is was making. Different tools for different folks. A zillion downstream choices determine much of what we see in the final film.

My experience with Epic has been that it generates images that are remarkably malleable in a way that no other digital format could deliver before. Now there are more choices. Great. But tagging one system or another with a "look" based on hearsay, or what we see in the movies, as opposed to testing and in the field experience with the camera systems doesn't seem very productive. It leads to nonsense statements like Joe G's:

Originally Posted by JoeG
I guess I'm a Red "nay sayer" as the MX sensor is substandard in my view (Alexa looks better, as does BMC). I thought it nearly ruined Gatsby, and a few others. Skin tones were crap. When they try pushing it around, it shows its limitations, especially when exposed brightly. The higher end doesn't hold up from my theater seat, and since I paid good money to see these things, my opinion is as valid as anyone else's, be they labeled "nay sayer" or whatever."

Subliminal "feel" has been chosen and latter image dismantled in post.

No "colorist replacement" LUTs or plugins. Thin online jpegs.
Close enough for the purpose.

 

[Gavin Greenwalt]

FFS.... I've said it 1,000 times and I'll say it one last time. Of course you can get from A -> B. I even posted a link to FilmConvert which *DOES* go from A->B. It comes down to 'defaults' though. As I said "there is nothing magical about film, it's just a chemical LUT". I'm wanting better defaults. I am lazy and want a better out-of-the-box experience.

When RED came out everyone was thrilled but many people complained about color. We got REDColor. Wow what an improvement! (said pretty much everyone). And many still complained that it wasn't great and out came REDColor2 and everyone again said "Wow what an improvement!" But people still complained and others bashed them for doubting the greatness of RC2. And out came RC3 and everyone again exclaimed "Wow what an improvement!" but now people are convinced that RC3 is perfectly fine. I'm sure when RC4 comes out everyone will once again exclaim "WOW what an improvement!" You can't keep saying that REDColor is fine and then rejoice and be merry every time a new one comes out and improves on it. I don't want to hear anyone saying RC3 is perfect say they can see a difference in RC4 when it's inevitably released. Because apparently RC3 has no room for improvement. If you go back to CameraRGB and compare it to REDColor3 there is a HUGE improvement. The two can barely be compared. And yet people managed to shoot great images before REDColor. It's not impossible. Before FLUT lots of us complained that it was too easy in the RAW settings to clip channels. People said we were hacks for saying it could be improved. RED implemented FLUT. HUGE improvement. There is a trend where lots of us point out something that could be greatly improved, we get attacked almost universally by fanboys who say RED is perfect and we should just be happy like ___insert__director___ "who doesn't need anything better". And then a year or so later RED drops an update and those same people are now the biggest cheer leaders of RED's amazing prowess in revolutionizing imagery (again). You can't keep revolutionizing something if there isn't a lot of room for improvement. If there was no room for improvement we wouldn't see any difference and nobody would get excited every time it gets better. There is some "Epic" cognitive dissonance going on to resolve those two conflicting opinions.

I don't want to start from a mediocre point and work towards something good, I want to start from something that's good. When I shoot film, the straight off the negative look is far closer to where I want to be than with RC3. Nearly anything is possible. The question is what's easiest. I'm not alone in this perspective. In fact I would say I stand with the majority. I haven't been on one phone call in my life where a DP said "I really want to shoot EPIC because I love its color." Never. Not once. I've never once had a director ask to shoot RED. I always have to fight for RED. And it's *ALWAYS* down to color. Every. Single. Time. RED is supposedly about listening to their customers. Ignoring the opinions of hundreds of professionals who have devoted their life to nitpicking images isn't helping RED, it's exactly the opposite, it hurts RED.

*steps off rant-soap-box*

 

[Timothy Eaton]

FFS.... I've said it 1,000 times and I'll say it one last time. Of course you can get from A -> B. I even posted a link to FilmConvert which *DOES* go from A->B. It comes down to 'defaults' though. As I said "there is nothing magical about film, it's just a chemical LUT". I'm wanting better defaults. I am lazy and want a better out-of-the-box experience.

When RED came out everyone was thrilled but many people complained about color. We got REDColor. Wow what an improvement! (said pretty much everyone). And many still complained that it wasn't great and out came REDColor2 and everyone again said "Wow what an improvement!" But people still complained and others bashed them for doubting the greatness of RC2. And out came RC3 and everyone again exclaimed "Wow what an improvement!" but now people are convinced that RC3 is perfectly fine. I'm sure when RC4 comes out everyone will once again exclaim "WOW what an improvement!" You can't keep saying that REDColor is fine and then rejoice and be merry every time a new one comes out and improves on it. I don't want to hear anyone saying RC3 is perfect say they can see a difference in RC4 when it's inevitably released. Because apparently RC3 has no room for improvement. If you go back to CameraRGB and compare it to REDColor3 there is a HUGE improvement. The two can barely be compared. And yet people managed to shoot great images before REDColor. It's not impossible. Before FLUT lots of us complained that it was too easy in the RAW settings to clip channels. People said we were hacks for saying it could be improved. RED implemented FLUT. HUGE improvement. There is a trend where lots of us point out something that could be greatly improved, we get attacked almost universally by fanboys who say RED is perfect and we should just be happy like ___insert__director___ "who doesn't need anything better". And then a year or so later RED drops an update and those same people are now the biggest cheer leaders of RED's amazing prowess in revolutionizing imagery (again). You can't keep revolutionizing something if there isn't a lot of room for improvement. If there was no room for improvement we wouldn't see any difference and nobody would get excited every time it gets better. There is some "Epic" cognitive dissonance going on to resolve those two conflicting opinions.

I don't want to start from a mediocre point and work towards something good, I want to start from something that's good. When I shoot film, the straight off the negative look is far closer to where I want to be than with RC3. Nearly anything is possible. The question is what's easiest. I'm not alone in this perspective. In fact I would say I stand with the majority. I haven't been on one phone call in my life where a DP said "I really want to shoot EPIC because I love its color." Never. Not once. I've never once had a director ask to shoot RED. I always have to fight for RED. And it's *ALWAYS* down to color. Every. Single. Time. RED is supposedly about listening to their customers. Ignoring the opinions of hundreds of professionals who have devoted their life to nitpicking images isn't helping RED, it's exactly the opposite, it hurts RED.

*steps off rant-soap-box*

For the record Gavin, as far as my post goes, I never said Epic or RC3 are perfect. Never said professional opinion isn't worth listening to. Never said Red color couldn't improve. All I said was it's not very helpful to put up a couple of internet jpegs as a way of making comparisons. I think Hrvoje pretty well demonstrated the fallacy of that approach.

All the best, Gavin, but no need to rant on my account. No need for FFS.

Tim Eaton

 

[Rudi Herbert]

Now, now Gavin! and of course, you are totally right.

 

[Hans von Sonntag]

You can, it's just really up to you in post. A good colorist will give you any look you want. Film-like, digital crispy lettuce, and everything in between.

I kindly disagree.

Any colourist is dependant on the colour he gets first hand. It is a difference to grade off film-scans, Red Raw (here the colour interpretation is vital, eg. RedColor, RedColour3) or Alexa files. Each camera has a different way to capture colour or at least to interpret the sensor's data as colour in RGB. This fact leads the colourist into a path she/he follows artistically. Best example for this is the often seen washed-out look in TV-commercials. This look has been inspired by Alexa LogC files that made directors curious.

Red's raw approach has the advantage to give their user-base a variety of colour interpretation possibilities such as RC1 - RC3 or CameraRGB which unfortunately is gone in RCX but still lives in many SDKs in 3rd party applications such as Resolve or Smoke. I mostly choose CameraRGB because it's closer to that what I get from Arri's Alexa and in IMHO more pleasant to my eye than RedColour3 or much worse RedColour2 and it's bad red channel. RedSpace was also an interesting Colour Science that leads to nice skin tones.

Personally I would love to have a colour science that is closer to film. Everyone who has tried Filmconvert as suggsted by Gavin will be strived by the uniqueness of the looks that come from LUTs which mimic different filmstocks quite well. And yes, any good colourist can copy that. But not from scratch and not with an example to follow. Modifying the matrix in a way that the resulting LUT looks in all shots like eg. 5298 is not a thing of minutes but a work of hours if not days.

The beauty of film was not the perfect copy of the colours you see in reality but their pleasing interpretation. I would love to see RedFilm isn the next colour science release.

Hans

 

[Blair S. Paulsen]

Sensor technology is the foundation of digital imaging performance. Dragon was designed for digital cinema using the best technology Jim and the RedTeam could find. While I don't know the provenance of the MX sensor, IMO its inherent response profile has always required some compensation in processing that reduces color linearity. The fact it looks as good as it does is a tribute to Graeme and a validation of the value of malleability when "developing" from RAW data.

I expect the Dragon sensor to be a significant step forward in color precision that quiets this whole issue.

Cheers - #19

 

[Gavin Greenwalt]

All I said was it's not very helpful to put up a couple of internet jpegs as a way of making comparisons.

Tim Eaton

Yes... if only I had also posted an apples to apples comparison between film and digital that was of the exact same shot.... that would have made things far more clear. :| If only... and if only I had even stated "for an apples to apples comparison two good examples would be" I guess hindsight is 20/20 in this case.

I think Hrvoje pretty well demonstrated the fallacy of that approach.

Hrvoje only proved my point that you *can* match RED to film but that by default people are releasing imo inferior footage because they don't have that reference point to start from and they get lost in the woods. I personally agree with those who say that Gatsby looks bad. I love Baz's films. But I just can't get past the look of Gatsby. A lot of that is the CG. But a lot of that starts with the color science.

It's like starting from a blank canvas. If I tell you paint using the tonality of film almost nobody will be able to just pull that out of thin air. If I give you a photo though it's relatively easy to match the contrast and color. It's why it's so important to shoot reference photos for CG. It's really hard to imagine photorealism. It's a lot easier to take a reference photo and match it.

Similarly there is a lot of Arrested Development shots that looks horrifically bad. Weird orange splotchiness. I'm sure they *could* use secondaries to get it in line... but apparently they didn't. And as a result sticking to the defaults produced a poor looking product.

Also you very clear said that those who think skin tones are off on RED are simply imagining it and can't actually tell the difference:

Some of this skin tone conversation has the same feel as wine snobbery, the idea that oenophiles are possessed of a higher sensibility that the rest of us simply don't have access to, especially as it pertains to the cost of the wine. Blind taste testing, however, has put to rest most of the wine nonsense, "experts" having been fooled in test after test after test.

If you can't tell the difference why bother improving it? If you can't tell the difference it's perfect.

 

[Timothy Eaton]

Yes... if only I had also posted an apples to apples comparison between film and digital that was of the exact same shot.... that would have made things far more clear. :| If only... and if only I had even stated "for an apples to apples comparison two good examples would be" I guess hindsight is 20/20 in this case.

Hrvoje only proved my point that you *can* match RED to film but that by default people are releasing imo inferior footage because they don't have that reference point to start from and they get lost in the woods. I personally agree with those who say that Gatsby looks bad. I love Baz's films. But I just can't get past the look of Gatsby. A lot of that is the CG. But a lot of that starts with the color science.

It's like starting from a blank canvas. If I tell you paint using the tonality of film almost nobody will be able to just pull that out of thin air. If I give you a photo though it's relatively easy to match the contrast and color. It's why it's so important to shoot reference photos for CG. It's really hard to imagine photorealism. It's a lot easier to take a reference photo and match it.

Similarly there is a lot of Arrested Development shots that looks horrifically bad. Weird orange splotchiness. I'm sure they *could* use secondaries to get it in line... but apparently they didn't. And as a result sticking to the defaults produced a poor looking product.

Originally Posted by Tim Eaton
Some of this skin tone conversation has the same feel as wine snobbery, the idea that oenophiles are possessed of a higher sensibility that the rest of us simply don't have access to, especially as it pertains to the cost of the wine. Blind taste testing, however, has put to rest most of the wine nonsense, "experts" having been fooled in test after test after test.



If you can't tell the difference why bother improving it? If you can't tell the difference it's perfect.

Well yes, it would have required quite a bit more effort to compare apples to apples, but apples to apples comparisons are infinitely more useful. My experience with Epic and the latest round of color science has been exceptional in terms of skin tones. Not perfect, but very good. But that's my experience. I haven't done apples to apples comparisons between Alexa and Epic, and so I tend not to make direct comparisons.

There are certain broad things I think we can learn from watching distributed films and TV. But I'm not willing, based on what I have seen in the theatre or on BD, or especially on footage from a run and gun internet show like Arrested Development to extrapolate the characteristics of the various camera sensors from that, and then make very certain comparisons of their capabilities. I have no idea, and I suspect you don't either, how Arrested Development was exposed or lit or captured or processed in post. On the other hand, maybe you have examples from your own work that shows what you believe are the differences between sensors. I always find that kind of information quite a bit more useful.

I did not claim that Hrvoje proved that you can match Red to film. Only that with a few simple adjustments he proved how easy it is to change the look of internet jpegs, and therefore how useless they are in telling us the nature of what was originally captured.

Ah, you keep editing your post. My comment about wine snobs and skin tones wasn't meant to say that no one can tell the difference in skin tones between films, only that most people are not comparing identical sources. Comparing wine this way would be like decanting one wine for a couple of days and tasting it, and then tasting a fresh bottle of another wine two days later and comparing the two. Apples and oranges. You may be the exception, Gavin, but so many posters write that after having seen such and such a film - who knows how projected or how processed or how mangled - Red skintones are lousy. I think it's possible to say they can look great - I would use Flight as an example - or not so good, depending on how the faces were lit, exposed and processed. That's all.

All in all, I don't think we're that far off Gavin. We're both looking for Red to keep innovating. That's all good.

Tim

 

[Robert Ruffo New]

Comparing the aesthetic of a movie is like comparing the aesthetic and look of of paintings in a museum. Most people in the art world don't look at a painting and start criticizing what paint brush the artist used. But movies ... WOW ... everyone is an expert.

In a creative, supervised color correction - a director and a sometimes a cinematographer makes choices in the final color correct. Prior to OFFHOLLYWOOD selling our post production operations to Light Iron - we color graded and delivered dozens and dozens of features shot on RED including the Academy Award winning Beginners used as an example in this thread. I have been a part of the production and post production process of dozens of films shot on the same sensor that look ENTIRELY different.

Which ones look "better" .... well ... that's totally subjective. The reality is - they each look the way the director, or cinematographer ... and even sometimes the producer wanted them to look. Whoever is directing the colorist in a DI on a film shot on RAW R3D's has a lot of room to swing - and the swings they choose may or may not be what you think is best. Guess what ... it's not your movie. Certainly, you are entitled to your opinion. It's just as valid as mine. And just as valid as the director of a given film. Guess who has the power to make the choice?

You can debate and compare what movie you think "looks better" all day long if you want. What is a little more relevant to this forum would be to appreciate how DIFFERENT all the movies shot on RED actually look. There's a CRAZY range.

My 2 cents.

100% agree. The camera defines what it can't do, but the artists involved define what it CAN do. Red cameras remove a lot of limits to do any number of things, any number of ways, without switching cameras.

 

[Hrvoje Simic]

Hrvoje only proved my point that you *can* match RED to film but that by default people are releasing imo inferior footage because they don't have that reference point to start from and they get lost in the woods.

They get lost in the woods because there is a desire to replace a colorist or become one without the time and energy invested into understanding that realm. Tools at hand alone can't justify the titles in the signature.

This color drama is blown out of proportion and many times the causes of undesirable outcomes are wrongly attributed to the tool.

I personally agree with those who say that Gatsby looks bad. I love Baz's films. But I just can't get past the look of Gatsby. A lot of that is the CG. But a lot of that starts with the color science.

Gatsby's look is the least affected by color science of the camera.
Feedback for the look should go to the DoP and the colorist.

It's like starting from a blank canvas.

Yes that should be the whole point. IMO.
But, color wise, some people some seem to prefer a canvas already half painted. For those priorities there are plugins.

One of the benefits of 16 bit RAW digital image capture is that it can be very precise and more realistic.
If a potent digital camera mimics the technically (not aesthetically) inferior medium, it reduces it's range of possibilities in the start.

Of course Epic's color performance has room for improvement, but I wouldn't suggest that improvement going into mimicing another medium. For a professional multifunctional digital camera the improvements related to picture quality should be in its accuracy and consistency in most conditions possible. Everything else should be left as a personal preference.

That way the tool is least biased and most creative options are left to the Craftsman.

 

[Tom Gray]

Hopefully I won't sound like a jerk but.... how can any one "professional" make an absurd statement about a photo, taken from a photo, that is posted on the net for God sake and ogle of it. Please people, get it together; you are all sounding like kids looking through the pictures in the Sears catalog at Christmas time in the 1950's!
The photo of the photo looks like what it is, trash! If any of you think that photo is great you're not a DP of any sorts.
cheers

 

[Gunleik Groven]

Hopefully I won't sound like a jerk but.... how can any one "professional" make an absurd statement about a photo, taken from a photo, that is posted on the net for God sake and ogle of it. Please people, get it together; you are all sounding like kids looking through the pictures in the Sears catalog at Christmas time in the 1950's!
The photo of the photo looks like what it is, trash! If any of you think that photo is great you're not a DP of any sorts.
cheers

The people actually posting those images, give the images some more credit, than if there were others posting those....

 

[Bob Gundu]

Tom, you need to understand the exposure tools. Never trust the LCD!

 

[Luis Otero]

...Any colourist is dependant on the colour he gets first hand. It is a difference to grade off film-scans, Red Raw (here the colour interpretation is vital, eg. RedColor, RedColour3) or Alexa files. Each camera has a different way to capture colour or at least to interpret the sensor's data as colour in RGB. This fact leads the colourist into a path she/he follows artistically...

Hopefully your do not start your color grading using any flavor of RedColors, which is for pre-visualization only. If you start your grading with RedLogFilm, it is 100% equivalent to Cineon, which is what the any colorists working with film captured images will use; not with a backed-in color. Then, they will use emulation LUT to achieve the desired stock look.


NOW, let's go back to the thread purpose: any footage from Dragon?

 

[Andy Roberts]

RedColors, which is for pre-visualization only.

Really?

 

[Gavin Greenwalt]

But, color wise, some people some seem to prefer a canvas already half painted. For those priorities there are plugins.One of the benefits of 16 bit RAW digital image capture is that it can be very precise and more realistic. If a potent digital camera mimics the technically (not aesthetically) inferior medium, it reduces it's range of possibilities in the start.

You're suggesting that MX is more "accurate" than film. I would strongly disagree. Every sensor imparts a "bias" on the image. To suggest that RED is more "accurate" than film is to suggest that its bias is better. No sensor will ever perfectly respond the way the eye does. There is far too much dynamic adaptive changes made from second to second as well as a huge amount of image processing in the neural cortex which is greatly affecting our perception. Film is no more or less "accurate" to reality. I would grant that RED is far more accurate when it comes to producing a nice and consistent linear luminance and that is a great help when trying to technically measure reality for things like HDR or compositing. However technical precision in producing a smooth gradient doesn't appear to translate into color (at least with RED).

For 100 years people have chosen film stocks based on preference. So clearly 100 years worth of artists were by your definition doing photography wrong? When a photographer chose Velvia because they liked the look of it they were choosing an inferior medium since it had a unique look? It would be quite the revelation to the photography community to discover that choosing a heavily biased film stock that imparted a unique look was bad. They should have done all of that in the dark room?

(BTW this extends to digital too. I absolutely love my cell phone sensor/processing. The way it sees the world is very unique and I really like its look. I have struggled long and hard to reproduce some of the looks my cell phone creates with my RED. Its skin tones are atrocious but its landscapes are really something marvelous.)

If you start your grading with RedLogFilm, it is 100% equivalent to Cineon, which is what the any colorists working with film captured images will use; not with a backed-in color. Then, they will use emulation LUT to achieve the desired stock look.

Color is not Gamma. RLF only sets the 1D luminance LUT to Cineon. Hue and Saturation (and to a lesser extent value) are driven by a 3D LUT. You still need "REDColor #" even if you are outputing RLF.

 

[Luis Otero]

Andy,

Yes... I do not know of any serious colorist that will start with a backed-in color. Grading rule of thumb: start with desaturated, high dynamic range, low contrast images.

 

[Luis Otero]

Gavin,

Sure: RLF and RC(2 or 3) = Cineon.

 

[Andy Roberts]

Andy,

Yes... I do not know of any serious colorist that will start with a backed-in color. Grading rule of thumb: start with desaturated, high dynamic range, low contrast images.

To say that it is for pre-vis ONLY implies that it isn't usable at all for grading. MANY people grade from RedColor3 and get great results.

And the phrase is "baked in"... as in baking a cake. :smilewinkgrin:

 

[António Fagundes]

I am the first to argue for the merits of digital. Especially in that digital could be made to look exactly like film with the proper LUT. But that's not to say that there isn't a big difference in the look between film and digital.

Compare gatsby to something like Romeo and Juliet:

There is a huge difference in subliminal 'feel' between the two. I don't see why people are so vehemently opposed to striving to offer a more filmic LUT option.

Another two good examples that are more apples to apples:
http://clorstudios.files.wordpress.com/2013/01/storm-copy.jpg
http://www.filmconvert.com/examples/

I wouldn't say that takes a wine snob to tell the difference.

Red looks nothing like film, and thank god for that! Red and Alexa look much much better than film, I personally prefer Red's image, but trying to compare a modern digital image with film is very backwards. Digital looks like digital, film looks like film, full stop. To try and make a digital image look like film with a lut just shows how backwards, uncreative and intimidated by letting go of the past one really is.

Again, Gatsby does not look like Romeo and Juliet and thank God for that. It looks way better!