Color Business Model Changes ... film light tutorial

[PatrickFaith]

I really like this film light tutorial on the changes that have occurred to color workflow and what is likely to occur. These changes make a huge business model change.

Direct video's at https://vimeo.com/119143638

I personally think RED nailed it with the WEAPON, and the people that take advantage of this will have a leg up in the business changes. Also it might make sense for some business, rather then ride the wave, to paddle around it - so a thought could be where would my business need to go in order for me not get wiped out. Here's a example I've been thinking of if RED had some kind of "Kodak'y" old school film camera simulation that could be loaded into a weapon:

"Use a Weapon and load a specific Kodak type LUT into it. This could basically simulate any film camera solution ever made, having a prores baked file that would still have the flexibility of linear compositing. With this type of simulated camera/film setup, you'd film with standard zone based lighting, people the cinematographer back and control. Also the entire edit process would be focused on story and technology, yet if there was a problem with principle photography you could go back to the raw for those shots. With modern software, since the LUT would be reversible, it would still be possible to do "hidden" linear compositing in the post workflow ... making post more focused on story."

[fyi ... there is a mix of very giving people and very selfish people in the business, people need to be more careful with giving out RAW files now with full prores, because huge amounts of research can be lost with reverse transforms ]

Another extreme in workflow direction would be going to a color standardized format that would work across all camera types, still a 3rd totally different workflow would be using a kodaky old school spec(like you used to give to the film processor), and using a mix of camera loaded LUTs and post processed LUT's to stay in a that kodaky format. My preference is there would be a new wavelet "LUT", since that would have spatial operations rather then just point operations - which would allow fully simulated processing all transparent to the film maker.

Anyway I think this video does a good job explaining the problems, and I am ponder how this will change my business model. Good thing RED nailed this, which is allowing me to leverage work from others.

 

[Zeb B]

Good read - thx!

 

[Les Hillis]

Some nice info/explanations in those video's, thanks for the links Patrick.

 

[Marc Wielage]

I worked with Baselight color scientist Peter Postma for a couple of years at Cinesite, and he was (and is) one of the brightest people I know in this area. But I don't believe there's a LUT that will change a digitally-acquired image and make it look like film negative. Too much depends on the nature of the exposure, the quality of the lighting, relative colorimetry, shutter/motion artifacts, and dynamic range. If a LUT could do all that, then it'd be possible to use a LUT to make a 4K iPhone look like a Red, or make a Red look like an iPhone. It's not that easy.

You can use a LUT to simulate what an image will look like when recorded out and printed to (say) a Vision release print, but it just looks like a digital image released on film. There's much more to what film looks like, how it compresses highlights, what happens to saturation, and so on. It's a lot of different little things that all add up to whatever "the film look" is. It's not simple, nor is it easy.

fyi ... there is a mix of very giving people and very selfish people in the business, people need to be more careful with giving out RAW files now with full prores, because huge amounts of research can be lost with reverse transforms ...

Color science is more complicated and more involved than you might believe. Have you read Jeremy Selan's report on Cinematic Color and color management?

http://cinematiccolor.com

Raw files and ProRes don't prove anything or require any research to use. I've worked many times with R3D files and I've also worked with RedLogFilm ProRes files, and there's no appreciable difference except in cases where there are drastic problems with exposure or color temperature. All things being equal, I'd rather have the Raw just in case, but for quick turnaround/tight-budget projects, the ProRes works fine. Roger Deakins' forum had a discussion of this very issue...

http://www.rogerdeakins.com/forum2/viewtopic.php?f=23&t=3121

 

[PatrickFaith]

NP Zeb & Les, as I'm fairly new to cinematography and as I'm trying to catch up I find a lot of good nuggets which are hopefully useful for you guys that are experienced. From my finance background though, I think this change in color workflow is going to hit in a way that many will have a hard time dealing with. The cost of the Weapon upgrade is a good example, and the cost drop of the RAVEN is another, both are disruptive in different ways in how this color processing wave is hitting us.

I worked with Baselight color scientist Peter Postma for a couple of years at Cinesite, and he was (and is) one of the brightest people I know in this area. But I don't believe there's a LUT that will change a digitally-acquired image and make it look like film negative.

You can make it simulate what it will look like when recorded out and printed to (say) a Vision release print, but it just looks like a digital image released on film. There's much more to what film looks like, how it compresses highlights, what happens to saturation, and so on. It's a lot of different little things that all add up to whatever "the film look" is. It's not simple, nor is it easy.

Totally agree with the "point" based LUT just getting it close, it takes full spatial transforms & algorithms to get it there. For many story driven low budget shoots where the cinematographer is hip to these constrains , I think close could be good enough by doing more in camera. In film lights later training video's they make it clear that their solution is using algorithm color spaces and not simple 3D LUT look ups.

I do think wavelet tensor transforms would be a lot more general and cleaner then 3D luts. Since R3D is all wavelet based anyway, wavelet transforms wouldn't be horrible speed wise(my kind of hack to how I do my black shading by pushing the "darker" wavelets down via using the "wrong" fps during black shading is a good example of a lot of this is all ready built into RED processing). I'm kind of oscillating myself on the color correction packages since I'm not seeing a move to wavelet transform nodes. I'm pondering just coding things myself in nuke since i'm already doing the 3D rendering/shader nodes myself to have the color space continuous between r3d footage and vfx processing.

 

[Phil Holland]

Just something interesting I'm experimenting with Marc. Taking a .R3D or ProRes file and doing something drastic like a +/- 3-5 EV correction on it results in something. However, now that Weapon has ProRes on board, something I've "enjoyed" is pushing or pulling via the REDCODE RAW metadata using ISO or FLUT as that maintains the general unique color science that's tuned there. That's a fairly large correction, but it's where I've been seeing some of the very beneficial sides of REDCODE RAW. Especially in a tonally rich image.

There's lots and lots of emulation LUTs out there. Some good, some bad. I don't know if there's any true benefit providing a said LUT out in the box Patrick. There's so many options out there and people can put whatever they want in there. Dragon and Weapon's color science is pretty banging right now as well. You can pretty much make it look like anything if you need to do that.

 

[Marc Wielage]

LUTs are very problematic. I have no problem with using custom LUTs designed for a specific project, but I'm very leery of the companies out there that sell $39.95 LUTs that make vast claims over what they can and cannot do. My take is that they change the picture, but I'm not convinced they improve the picture to the extent a good colorist can.

Hitler apparently agrees...

My opinion is that for a large percentage of the LUTs out there, it's possible to recreate their looks within the existing controls of any good color-correction system, and I would count Baselight, Lustre, Mistika, Nucoda, Pablo, and Resolve in that list.

 

[Hrvoje Simic]

"Use a Weapon and load a specific Kodak type LUT into it. This could basically simulate any film camera solution ever made, having a prores baked file that would still have the flexibility of linear compositing.

If it was that simple, rest assured Arri, Sony and Red would already have those transforms.

Film has one type of response to light, digital camera A another, digital camera B another, etc.
The response is not consistent in any of those, it changes throughout many shooting related factors, plus with film the look depends on how it is processed. There isn't one look, there is a whole art in treating it to get a specific look before it reaches the digital grading suite. A LUT cannot fixate one, let alone a bunch of moving targets, it can only transform one set of fixed values to another.

With simulating film, one would also have to simulate texture and acuity. Texture is practically gone with digital and through resolution increase acuity has gone up significantly past film. So those two look defining parts are gone. Another one is the look of motion.

So, no. A LUT would not simulate any film stock ever made. It can only get close to mimicing a part of one type of response in specific circumstances, the transform which is useful until the capture medium/tool response offsets enlarge and/or until the less experienced user manipulates the image in a way it negates the effect of the emulation LUT. LUTs have a future but that future depends on understanding and accepting the need for a colorist in cases where specific image aesthetic results are desired.

There is no plugin which makes every music composition sound good, no plugin which makes any composited FX sequence fit all together properly, and no plugin or a LUT which makes any image look good. Creative human possessing technical knowledge, sensibility to aesthetics and understanding of the context is a prerequisite to achieve that goal.

Ever since the first rudimentary "film look" plugins for first era of digital cameras, the "film look" was and still is - just a terminology for the rational part of the mind, which uses film as a reference, the terminology actually representing "less digital look".

 

[Elsie N]

LUTs are very problematic. I have no problem with using custom LUTs designed for a specific project, but I'm very leery of the companies out there that sell $39.95 LUTs that make vast claims over what they can and cannot do. My take is that they change the picture, but I'm not convinced they improve the picture to the extent a good colorist can.

Hitler apparently agrees...

My opinion is that for a large percentage of the LUTs out there, it's possible to recreate their looks within the existing controls of any good color-correction system, and I would count Baselight, Lustre, Mistika, Nucoda, Pablo, and Resolve in that list.

:smilielol5::smilielol5::smilielol5::smilielol5::smilielol5:

 

[Marc Wielage]

There is no plugin which makes every music composition sound good, no plugin which makes any composited FX sequence fit all together properly, and no plugin or a LUT which makes any image look good. Creative human possessing technical knowledge, sensibility to aesthetics and understanding of the context is a prerequisite to achieve that goal.

Very well said. I wish it were easy.

I can recall similar conversations being done among sound engineers wondering if it were possible to make one microphone sound like another by emulating and adding the differences. It's not, because there are things beyond what we can measure, along with a lot of unpredictable, non-linear reactions that make these things what they are.

And when it comes to cameras, in truth... it all boils down to the lighting, the exposure, and the lenses, much of which are affected by the skill of the cinematographer.

 

[jake blackstone]

In the old days there were Kodak, Fuji, Agfa, Ilford stocks as well as many others. If people wanted image to look like Kodak, they used Kodak stock, for Fuji, they used Fuji. Every film stock had unique characteristics and it was mostly chosen because of that. Obviously, some were also used because of their lower cost. Working with modern electronic cameras is not that much different. I see no problem with different cameras having a unique look signature. In my opinion, various looks should be celebrated and encouraged. I'd hate to see the world, where every camera would look the same...