Steve Yedlin, ASC on Resolution

[Aaron Lochert]

I found that quite useful, so thanks for sharing it, Aaron. Now I know what a 3D LUT is and how it's represented. Here's a question I would have asked him: why not take an average of about 6 or so frames of 5219 before taking colour samples to create a LUT?

I'm sympathetic to the notion that digital can look as nice as film, although there are one or two things that digital fails at miserably. But only one or two things. Anyway, that was not the point of the presentation.

Although, if digital can look exactly like film, why haven't I seen it? Perhaps this is on me to figure out why, as I may not have looked at enough material.

What's also interesting to me is that Nuke can be used to create LUTs. I thought it was merely a compositing tool, and I thought that Resolve was what you used to make LUTs.

He's drawing a box on the grey patch, I'm going to assume that is an average of all those pixels, so he's already got plenty of samples. I'm unfamiliar with nuke though so I could very well be wrong.

"Exactly like film" is probably a phrase that Yedlin would stray away from using. He's shown that either medium can look like the other within a reasonable margin that it makes no perceptible difference for the end viewer. You say there's one or two things that digital fails at miserably, but film also has things it "fails" at, for example halation on high contrast edges (that's a loss of information in that area that can never be recovered).

Resolve is a little less mathematician minded when it comes to its tools. VFX editors often need to work with hard numbers, so it makes sense that Yedlin's more complex approach to that is within a program that facilitates that need.

 

[Mike P.]

Doing what Yedlin is doing in that demo should have been priority #1 for RED since the release of Dragon (at the very latest).

I mean, REDs as raw data collection devices obviously get enough info, but less technical users (aka producers/directors/most DPs -- who are typically artists first) don't want to learn this stuff (nor should they have to) in order to get the tried&true/standard result they want (aka film emulation).

I still think if RED had an official "Kodak 5219 pack"/"Fuji Eterna pack" with 1.) a Film OLPF (that resulted in a more even over/under exposure and gave film-like halation) and 2.) sensor-specific filmstock-emulating 3D LUTs (like Yedlin makes), it would've lessened the "Alexa just looks more like film out of the gate" belief that have been biting RED in the ass since the release of the ALEV3. And I'd argue that until they do, Arri is still going to dominate.

 

[Phil Holland]

Hi Mike

 

[Mike P.]

Lol! I know, I know, it's just a little frustrating seeing RED's potential *still* being limited by afterthoughts and/or lack of knowledge. And seeing Yedlin dissect the looks of film and digital to a.) interchangeably match each other so seamlessly using an across-the-board application of his LUT/display prep/process and b.) essentially say 'any of the contemporary digital cinema cameras are capable', just makes it sting a little.

I mentioned it to Jarred in another thread; it just seems like a no-brainer to remove that variable (whilst still allowing the tweakers to dig deep when they want). Again, imagine how much smoother RED adoption would've been had rental houses offered it as 'You want your RED kitted out as Kodak flavour or Fuji flavour? Kodak? Okay just meter at ISO400 and it's what 5219 would look like if it were 4k/6k/8k!'



...And besides, you have to beat the drum in order to be heard... Squeaky wheel and all that.

 

[Phil Holland]

Lol! I know, I know, it's just a little frustrating seeing RED's potential *still* being limited by afterthoughts and/or lack of knowledge. And seeing Yedlin dissect the looks of film and digital to a.) interchangeable match each other so seamlessly using an across-the-board application of his LUT/display prep/process and b.) essentially say 'any of the contemporary digital cinema cameras are capable', just makes it sting a little.

I mentioned it to Jarred in another thread; it just seems like a no-brainer to remove that variable (whilst still allowing the tweakers to dig deep when they want). Again, imagine how much smoother RED adoption would've been had rental houses offered it as 'You want your RED kitted out as Kodak flavour or Fuji flavour? Kodak? Okay just meter at ISO400 and it's what 5219 would look like if it were 4k/6k/8k!'

...And besides, you have to beat the drum in order to be heard... Squeaky wheel and all that.

Film emulation is literally nothing new. It started out of necessity specifically for VFX, eventually to blend other digital acquisition with film, and now more or less one aesthetic.

A key thing you're overlooking here is the goal of Steve's demo is a deconstruction for this specific look and he can and does take it other directions.

The biggest question that occurs mainly for mainstream DPs (and Directors for that matter) these days is "what should this project look like". A print emulation top to bottom is one flavor of ice cream.

I've been working on something for over a year that I'm keeping under wraps for just one more minute that will likely satisfy your desires.

But I will say, Steve and myself share the same quibbles about how "uncontrolled" the film look generally is. I know from my first decade of being in this industry I probably made two dozen hardcore emulations with slight tweaks. In truth, there is no golden target because with film itself you came up with your final image as well.

No matter what even with a "pure emulation" you'll still be taking it from there. Some fall in love with one look, this industry is not an industry that does one look. Not by a long shot.

 

[Mike P.]

I've been working on something for over a year that I'm keeping under wraps for just one more minute that will likely satisfy your desires.

Ahh, so your "Hi Mike " was more of a hold-on-to-your-butts/'wait'll you see what I'm cooking up!' type of thing? Cool! (Is going to be officially RED supplied? Will it work with DSMC1?)

Anyway, I understand what you're saying, but my thinking/frustrations are how it pertains to RED. In my opinion (I could be totally wrong), the industry/decision makers/on-set usually prefer a "film look" as the base, which can then be disabled/tweaked/changed in a million different ways in post. Whereas, it seems RED went with a more *accurate* look, expecting users to know that it could/would be disabled/tweaked/changed, but instead was interpreted as more videoy looking and requiring more post work/nuance/knowledge.

On top of that, any time someone brought up the idea of 'make it look more like film/Alexa out of the gate' RED's response was, 'it's raw; you can do whatever you want!' which, if Alexa's success is any indication, didn't work.

 

[Phil Holland]

Ahh, so your "Hi Mike " was more of a hold-on-to-your-butts/'wait'll you see what I'm cooking up!' type of thing? Cool! (Is going to be officially RED supplied? Will it work with DSMC1?)

Anyway, I understand what you're saying, but my thinking/frustrations are how it pertains to RED. In my opinion (I could be totally wrong), the industry/decision makers/on-set usually prefer a "film look" as the base, which can then be disabled/tweaked/changed in a million different ways in post. Whereas, it seems RED went with a more *accurate* look, expecting users to know that it could/would be disabled/tweaked/changed, but instead was interpreted as more videoy looking and requiring more post work/nuance/knowledge.

On top of that, any time someone brought up the idea of 'make it look more like film/Alexa out of the gate' RED's response was, 'it's raw; you can do whatever you want!' which, if Alexa's success is any indication, didn't work.

Yes indeed. philmColor has been a journey of releasing, listening, developing, and working with a whole lot of people industry wide. Anything IPP2 works with DSMC1 in post as well. IPP2 wasn't around when DSMC1 existed to have the hardware support in camera, nor LUTs for that matter. But anything REDCODE RAW benefits from current and future Color Science Advancements. The biggest thing really there was RED took the task of simplifying the color pipeline away from all the variations while increasing the quality of color and image from your footage.

As for RED supplied, I'll think about that.

To the point of Steve's Display Prep Demo, he's really trying to push home the concept of maintaining authorship, which is a hard concept for the general industry to wrangle. While people certainly respond nicely to say film, Alexa, or whatever; it's important to always remember these are productions with well crafted looks, colorists, etc.... Sure there's straight to air perspective, but that's not narrative or commercial work typically.

The magic bullet is the medium itself, not some specific LUT that may get popular. I'd feel rather ashamed if I just released one look. That's just not how anything should work when it comes to creative intention or possibly direction.

 

[Hrvoje Simic]

l
In my opinion (I could be totally wrong), the industry/decision makers/on-set usually prefer a "film look" as the base, which can then be disabled/tweaked/changed in a million different ways in post.

They prefer "not digital look" avoiding many parameters which scream "digital", and may refer to it as "film look" while the image is lacking image properties of film.

Whereas, it seems RED went with a more *accurate* look,

That cannot be said in general. Red has some parameters more accurate, some are more accurate with Arri CS, some with other camera makers. It is not either/or.

Accuracy in general is not a factor limiting image organicity and the reference of "accurate" is another subject.