Best LUTs for Weapon?

[Vadim Bobkovsky]

Moving footage with the same lut.

Is there something with Frame.io? Link to the video is fine, but there's no way to download luts.zip off that site. Weird.

 

[Björn Benckert]

Is there something with Frame.io? Link to the video is fine, but there's no way to download luts.zip off that site. Weird.

aha did not click downloadable... now it works. Sorry what a teaser.

 

[Medavoym]

If you must use an emulation LUT, then you are free to use it in addition to ACES or RCM. Not a problem.
Said that, if you want film look, then shoot film. No LUT can make electronic camera look like film, despite all those claims made by the LUT makers, sorry...

Hi Jake. I understand. And I agree with shooting film for film. Sometimes, we just can't afford it. So I am using the print emulation LUT as a "look". Not as a way to scientifically emulate film.

Question. Since you mentioned using an emulation LUT in addition to RCM.

I am shooting Red Dragon DC4/RLF. I am finishing in REC 709. So, naturally, in Resolve when it comes down to RCM, I will use as input DC4/RLF and as output REC 709.

However, as far as the timeline is concerned... I have CINEON LUTs, not Redlogfilm LUTS. I want to color correct as if I am in Cineon, using the Cineon LUTs.
Do I simply do "Cineon" for the timeline RCM setting?

And - since I am using a Cineon to REC 709 Kodak print emulation LUT - which does the "transform" to REC 709 - do I still need to choose the output as REC 709 in RCM? Or does the LUT take care of that?

THANK YOU!

 

[jake blackstone]

Hi Jake. I understand. And I agree with shooting film for film. Sometimes, we just can't afford it. So I am using the print emulation LUT as a "look". Not as a way to scientifically emulate film.

Question. Since you mentioned using an emulation LUT in addition to RCM.

I am shooting Red Dragon DC4/RLF. I am finishing in REC 709. So, naturally, in Resolve when it comes down to RCM, I will use as input DC4/RLF and as output REC 709.

However, as far as the timeline is concerned... I have CINEON LUTs, not Redlogfilm LUTS. I want to color correct as if I am in Cineon, using the Cineon LUTs.
Do I simply do "Cineon" for the timeline RCM setting?

And - since I am using a Cineon to REC 709 Kodak print emulation LUT - which does the "transform" to REC 709 - do I still need to choose the output as REC 709 in RCM? Or does the LUT take care of that?

THANK YOU!

Cineon is an RLF curve, so you can use it, if you like.

 

[Medavoym]

Cineon is an RLF curve, so you can use it, if you like.

Thanks!

1) So, in RCM, if I use Cineon as the timeline setting with a Cineon to REC 709 LUT, what should I use as the output setting? NOT REC 709, right? Since I am already using a Cineon-to-REC 709 LUT. Wouldn't I "transform" twice then? Or am I getting something wrong...

2) By the way. I heard that RLF is not exactly Cineon (just as LogC and Slog3 are not exactly Cineon Log). In order to squeeze more latitude into the encoding, apparently in all of these cases the 18% and 90% scene reflectance points are pushed further down.

Cineon
0% 95
18% 468
90% 671

LogC
0% 95
18% 400
90% 569

Slog3
0% 95
18% 420
90% 598

I have no idea what the RLF values are.

So applying a Cineon LUT to RLF/LogC/SLOG3 originated material (instead of film scanned to Cineon) is not the same as the LUT adds a s-curve to the display gamma, so if it receives values lower than anticipated it will cause crushing of the data.

For this reason, I want to use the Cineon LUT on material interpreted as true Cineon (even if shot in some other flavor of LOG). Wondering how I can ensure that.

 

[jake blackstone]

Thanks!

1) So, in RCM, if I use Cineon as the timeline setting with a Cineon to REC 709 LUT, what should I use as the output setting? NOT REC 709, right? Since I am already using a Cineon-to-REC 709 LUT. Wouldn't I "transform" twice then? Or am I getting something wrong...

2) By the way. I heard that RLF is not exactly Cineon (just as LogC and Slog3 are not exactly Cineon Log). In order to squeeze more latitude into the encoding, apparently in all of these cases the 18% and 90% scene reflectance points are pushed further down.

Cineon
0% 95
18% 468
90% 671

LogC
0% 95
18% 400
90% 569

Slog3
0% 95
18% 420
90% 598

I have no idea what the RLF values are.

So applying a Cineon LUT to RLF/LogC/SLOG3 originated material (instead of film scanned to Cineon) is not the same as the LUT adds a s-curve to the display gamma, so if it receives values lower than anticipated it will cause crushing of the data.

For this reason, I want to use the Cineon LUT on material interpreted as true Cineon (even if shot in some other flavor of LOG). Wondering how I can ensure that.

RLF is Cineon. Log_C and Slog are not.
We were talking about using creative LUTs in addition to RCM. If you want color space conversion, then use an appropriate tool for that- color space transform.

 

[Marc Wielage]

I spent the better part of my life around Domino, Cineon and Genesis and solitare film printers. I bought my first digital film printer when I was 22 years old and I´m far older now so I been doing it for quite some time. So no this is not just a made up look, its partly made by kodak / cinesite them selves.

I worked for Cinesite/Hollywood for 2 years and knew Mitch Bogdanovich -- the Kodak color scientist who created the Kodak LUTs -- very well. My opinion is that the LUTs were variable at best and required constant adjustment and tweaking. There was almost as much guesswork as there was math in LUTs like this. No need to give me a rundown of how labs worked; I also spent 20 years on and off at Technicolor's digital arm in Hollywood, and over the last 15 years had a myriad of experiences with Technicolor's LUTs. They were all over the place, totally unpredictable, and required lots and lots of trial & error. We had so many, several engineers referred to the "LUT of the Month Club," because just a single LUT could never work for an entire project.

I came away from both experiences feeling that there was more smoke & mirrors with the whole idea than anything else. I do have a lot of respect for the Filmlight people, and I think their Truelight LUTs -- which are real transforms -- are an entirely different thing. Baselight (owned by the Filmlight people) incorporates a lot of those ideas. But they're not using the LUTs as looks; they're actual transform LUTs to get you from one place to another.

My point is that too many people believe that LUTs are a substitute for skill and experience, and they are not. There is no plug-in for that. And there is no way to make a digital camera really look like film. (I would say if there was, then there would be an opposite LUT that would make Film look like Digital.) This is why steam comes out of my ears when somebody asks, "what is the best LUT can I use for my camera?" Because the answer is: you don't need a LUT. You need a colorist -- whether it's you learning how to color-correct, or hiring somebody. An off-shelf LUT is not the answer. By the same token, freeze-dried food cooked in a microwave oven is not going to result in a gourmet meal, either. Is it acceptable to some people? Sure. But not to those who know what the real thing really is.

 

[Marc Wielage]

Cineon is an RLF curve, so you can use it, if you like.

I would say the opposite -- that RedLogFilm is based on the Cineon curve developed by Glenn Kennel and the other Kodak color scientists in the late 1980s and early 1990s (and won an technical Oscar) -- but it's an apt comparison.

 

[jake blackstone]

I would say the opposite -- that RedLogFilm is based on the Cineon curve developed by Glenn Kennel and the other Kodak color scientists in the late 1980s and early 1990s (and won an technical Oscar) -- but it's an apt comparison.

RLF is the same curve as Cineon, therefore they are the same thing. So, what are you objecting to?

 

[Björn Benckert]

I worked for Cinesite/Hollywood for 2 years and knew Mitch Bogdanovich -- the Kodak color scientist who created the Kodak LUTs -- very well. My opinion is that the LUTs were variable at best and required constant adjustment and tweaking. There was almost as much guesswork as there was math in LUTs like this. No need to give me a rundown of how labs worked; I also spent 20 years on and off at Technicolor's digital arm in Hollywood, and over the last 15 years had a myriad of experiences with Technicolor's LUTs. They were all over the place, totally unpredictable, and required lots and lots of trial & error. We had so many, several engineers referred to the "LUT of the Month Club," because just a single LUT could never work for an entire project.

I came away from both experiences feeling that there was more smoke & mirrors with the whole idea than anything else. I do have a lot of respect for the Filmlight people, and I think their Truelight LUTs -- which are real transforms -- are an entirely different thing. Baselight (owned by the Filmlight people) incorporates a lot of those ideas. But they're not using the LUTs as looks; they're actual transform LUTs to get you from one place to another.

My point is that too many people believe that LUTs are a substitute for skill and experience, and they are not. There is no plug-in for that. And there is no way to make a digital camera really look like film. (I would say if there was, then there would be an opposite LUT that would make Film look like Digital.) This is why steam comes out of my ears when somebody asks, "what is the best LUT can I use for my camera?" Because the answer is: you don't need a LUT. You need a colorist -- whether it's you learning how to color-correct, or hiring somebody. An off-shelf LUT is not the answer. By the same token, freeze-dried food cooked in a microwave oven is not going to result in a gourmet meal, either. Is it acceptable to some people? Sure. But not to those who know what the real thing really is.

Hmm, I disagree on some points. Yes it was guess work back with film and yes mimicking film is difficult, film is never the same not in two rolls not two different days at the lab, etc. Why because it´s a seriously analog format. Lab fluids, temperatures, weather, humidity, stock age etc is all playing a part. And all that shit change overtime.

The so called "close loop calibration" of my Dom was at most times quite far from being just that. More of a pipe dream for the sales people. Trail and error would have been a more correct term for the process. Same thing for the Arri, solitare and oxberry gear... simply more of an balancing act then something stable.

But that was back then. When you also more importantly could not see shit until you been trough the lab and looked at the neg and scan that and print it, develop the print and look in the projector... Then go back and alter what ever you though was wrong until getting a copy that was somewhat right which, all took days.

So no need to compare that to where we are now. Having luts built into a very accurate touch screen of the camera with histograms on it (yes the touch is shit more accurate than the video tap or even the colortimers screens at the lab)... Thats the difference now you can kind of apply the lab right there out in the field and see how you could expose to make your "negative" suitable for a INR, Bleach by Pass or what ever print process you prefer... and you can push and pull "your negative" under that lab process. And you can also record processed files with that stuff applied as a guidance for grading and to be used in offline... and in grade you got the used lut and can choose to apply and grade under and above it or toss it away and make your own grade. None of this was possible to get done with even half the accuracy back in days of film.


In the film days for the high budget commercials we normally struck several prints from a neg cut of selected takes and transfered those to get the look of positives to bring to the telecine... and sorry to this date I have not seen a colorist that match the result of those lab processes in full. And yes mimicking the look of film is a difficult process but to me allowing the DP to go to the lab (inside his camera) and take a sneak peak of how that bleach bypass process sort of will look for the shot at hand is not a bad feature.

If Fincher had this for 7even I'm sure he would have used such feature in camera. I have no clue how he is working now but I can only guess is that he got a few luts loaded into his cameras to choose from.

So no there is no need to fiddle with luts in camera or anywhere else. But simply to have three layers of grading in camera is not a bad thing to be able to have. A lut on top, and then be able to select red gama of choice underneath and then the iso and other sliders and curve...

Simply put. to me those are not bad things to have. But will I fiddle with them when sun is about to go down and every second counts... no, then you just push the red button and check your exposure at best, but all shoots are not like that. Some provide more time for the details.

The project I work on now had a big shoot going on for 2 weeks. We had 5 alexas, one phantom flex, a phantom 4k and a weapon. Then we also had a on set colorist that provided luts for the cameras and also live graded the video feeds and together with the DIT also graded the dailies. All that just for the offline edit then Jean-Clément at MPC did his grade and now we do the VFX and after that it probably go back to MPC for some final color tweaks with Jean. So I think no need to worry, none of this is about replacing colorists with one little file. As in the example above all of a sudden there is two colorists which get to work, one for a full 2 weeks and then jean had another 5 days or so, all for a little commercial. So some people will still do proper color sessions. But, like it always have been, some people will not go that route. They want to shoot and be done with. In those cases it´s quite difficult to argue that luts are a bad thing. So why fight it, why all this "leave it to a real colonist" kind of talk all the time?

 

[Phil Holland]

My point

I'm with you 100% on this Marc. There's many different types of jobs out there and turnarounds. One thing I've seen evolve over the years is having a colorist involved ahead of production a bit to help create LUTs to be used on set, various looks for specific episodic productions, as well as generating looks per scene/shot to feed to editorial or whomever along the way. In these cases LUTs are mostly used as a starting place. On well oiled productions perhaps more.

When it comes down to general content from commercials through longer form a colorist is ideal due to the many, many things that can and do happen that need an experienced hand.

And a +1 for Filmlight. At R&H we eventually moved up to the Northlight 6K scanner and to date that's the best designed scanner I've ever used, especially from an operation standpoint. Though back in the day interfacing our stuff with their stuff was a bit tough initially. We also eventually went over the Baselight. They've always done good work.

The sticky widget in the color/film/LUT concept was often the film itself. So many variances of what came into the studio then the variances on top of that when filming out and getting the print back with fingers crossed we were within our LAD tolerance as well as the various stock variances themselves, though that wasn't so bad really. Judicious use of color charts on set helped a great deal, but there was still a wiggling hair that was occasionally annoying.

The biggest crossing point at the studio for me back then was escaping the world "print light limited" CCs, moving on from filming out color wedges, and properly hammering down digital calibration to the laser recorder's output. Color became a much quicker process after that because directors could come straight into my office and make selects of digital wedges or quilts, then we'd film them out to get notes or eventually final approval.

As much as I love film, I certainly enjoy the control and consistency of digital formats a lot more. It tidied up many headaches "eventually".

The extra layer of goodness comes with digital delivery on properly calibrated displays. Then the audiences sees exactly, or as close to exactly as they can get, to what everybody in the chain was going for.

 

[Brett Harrison]

LUTs are per-pixel operation. Demosaick algorithms are iterative and spacial, and much more complex. You can't undo a debayer to get to the original colour filter array image. Just like you can't model the effects of colour film with a LUT - texture is a whole other deal that LUTs don't affect because they're not spacial. Colour film consists of dye clouds. You'd have to use something like LiveGrain to start to get that kind of texture in an accurate way, IMO, because textural processes impart subtle colour changes at the frequency level that can't be affected by a per-pixel lookup.

The standard RED gamut and gamma is a transform from the native camera colour space. It's a 'look', essentially a LUT but in floating space and arrived at directly through the mathematical definitions of the colour space. DC2 and RG4, LogC, ACEScc, ARRI KxSx Rec709 LUTs - all these are per-pixel 'looks' applied to the demosaicked linear image.

 

[Sabovic Adis]

RLF is the same curve as Cineon, therefore they are the same thing. So, what are you objecting to?

Well, Cineon isn't exactly a "development" curve - no "ISO".
Ever tried importing RLF (.dpx) material in ADX 10 color space? :smilewinkgrin:

 

[Björn Benckert]

LUTs are per-pixel operation. Demosaick algorithms are iterative and spacial, and much more complex. You can't undo a debayer to get to the original colour filter array image. Just like you can't model the effects of colour film with a LUT - texture is a whole other deal that LUTs don't affect because they're not spacial. Colour film consists of dye clouds. You'd have to use something like LiveGrain to start to get that kind of texture in an accurate way, IMO, because textural processes impart subtle colour changes at the frequency level that can't be affected by a per-pixel lookup.

The standard RED gamut and gamma is a transform from the native camera colour space. It's a 'look', essentially a LUT but in floating space and arrived at directly through the mathematical definitions of the colour space. DC2 and RG4, LogC, ACEScc, ARRI KxSx Rec709 LUTs - all these are per-pixel 'looks' applied to the demosaicked linear image.

So how would you defined cameraRGB, also a look?

 

[Hrvoje Simic]

LUTs are color tables, that are used to determine a resultant color, i.e this color stimuli on the input should be that resultant color. Transforms are formulas for transformations, like 1+2=3. Not the same thing...

Yes, it's not the same thing.

One are formulas for transformations and other are look up tables for transformations. And they both transform the image.

 

[Frank Martin]

got it thanks

 

[Ugo Del]

I really hope Red will work on the colors more than the resolution in the future. When there was the MX, I was reading people here telling the guys-not-100%-happy-with-the-colors that they were wrong, that anything was possible with raw in post anyway and the difference with the Alexa was in their brain. Then came the Dragon. The same people who told that it was already perfect and that the MX would be the only camera they would need and own in their life were jumping all over telling that the colors were improved so much from the MX. Again telling the not-totally-satisfied-ones that they didn't understand anything "because it's raw and Red color science is perfect", and the ones pointing a difference with Arri look were just ignorant. It will be the only camera they will ever own. Now we have Weapon. And it starts again. "Weapon is even better, better colors, better DR, better everything, so easy to match Arri because it's raw". Either they are right and people thinking Arri is more organic with better color separation are just plain stupid (including the Panavision team which is developing another color science with a specific OLPF). Either they are not good enough to notice the difference themselves (which can happen even to guys who earn a lot of money in this industry and think they have the best eye around). And I've seen a good amount of guys having a high opinion of themselves showing some footage with some sort of tasteless instagram color correction applied claiming it was looking like film or like Alexa. Well maybe I'm stupid. Maybe I've got a more trained eye. I don't really know. But IMO Arri has a lot of small tricks that make their image pleasing. Some filtering somewhere, the saturation blocked after 40IRE, the hues, the colors luminosity & saturation, the DR etc etc. I don't think a LUT will be able to mimic all that.

 

[Stacey Spears]

I really hope Red will work on the colors more than the resolution in the future.

Its not like they have one resource who splits time between the two. Chances are the person, or people, working on color have nothing to do with spatial resolution. With the exception being debayer algorithms. Color science is always evolving and, hopefully, improving. It is an interesting struggle of accuracy vs. subjectively pleasing. Something that is subjectively pleasing may or may not be accurate. When you serve multiple markets it is even more difficult as different customers have different requirements. Panavision probably has a more focused market than Red and can tune the color for that market.

When a new sensor comes along, they basically start over again on color working to get the most out of it.

At the end of the day, a highly skilled colorist will produce better results than the starting color space. LUTs help people achieve a certain look with minimal effort or experience, especially with a limited budget. In a perfect world we would all hire a Marc or Jake to make our content the best it can be.

In the world of movies and television, the color is not about reproducing the real world as we see it.

Cameras in cell phones actually tune the color based on country or region. In one country, an iPhone image may be more blue and in another country more red. BTW, the two countries were the US and Japan. I just don't recall which one preferred blue and which one preferred red. This is based on data they have collected. I find it interesting. Years ago a company added a sharpness control to their product. To me, it look awful in any position other than off. The feedback they received from Asia is that they wanted the control to go higher than it did. At the time the scale was 1-10. While this is not color, it does show how we all have our preferences.

 

[jussi rovanpera]

I really hope Red will work on the colors more than the resolution in the future. When there was the MX, I was reading people here telling the guys-not-100%-happy-with-the-colors that they were wrong, that anything was possible with raw in post anyway and the difference with the Alexa was in their brain. Then came the Dragon. The same people who told that it was already perfect and that the MX would be the only camera they would need and own in their life were jumping all over telling that the colors were improved so much from the MX. Again telling the not-totally-satisfied-ones that they didn't understand anything "because it's raw and Red color science is perfect", and the ones pointing a difference with Arri look were just ignorant. It will be the only camera they will ever own. Now we have Weapon. And it starts again. "Weapon is even better, better colors, better DR, better everything, so easy to match Arri because it's raw". Either they are right and people thinking Arri is more organic with better color separation are just plain stupid (including the Panavision team which is developing another color science with a specific OLPF). Either they are not good enough to notice the difference themselves (which can happen even to guys who earn a lot of money in this industry and think they have the best eye around). And I've seen a good amount of guys having a high opinion of themselves showing some footage with some sort of tasteless instagram color correction applied claiming it was looking like film or like Alexa. Well maybe I'm stupid. Maybe I've got a more trained eye. I don't really know. But IMO Arri has a lot of small tricks that make their image pleasing. Some filtering somewhere, the saturation blocked after 40IRE, the hues, the colors luminosity & saturation, the DR etc etc. I don't think a LUT will be able to mimic all that.

If you refer to the "saturation blocking" that Art Adams has been writing about, it only applies when recording 709 straight out of the camera. You wouldn't want that kind of processing on log/raw footage.

 

[Brett Harrison]

So how would you defined cameraRGB, also a look?

Camera RGB is the result of sensor design and demoasick method. It would be interesting to see what ARRI colour scientists and developers could do with the colour filter array channels from RED, for example. It would look different. There's lots of things that are akin to demosaick algorithms that can be done on the image after debayer. For example, denoising is closely related to demosaic.