Welcome to our community

Be a part of something great, join today!

  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

LUT Education

I am drumrolling for some 3D lut extravaganza and ramblings with steve/nick/björn/hrjove as sparring-partners...

Cool. I'll wait for your 3D LUT extravaganza and ramblings party findings
 
Let's talk LUT's!
I'm in

Well ... Steve Shaw eats LUTs for breakfast
Yes, there is obviously something about having the surname Shaw which gives you a predilection for LUT work!

Here's a picture of the three (unrelated) Shaws at IBC – courtesy of Warren Eagles:

three_shaws.jpg


Left to right, Nick Shaw, Steve Shaw and Kevin Shaw
 
Let's start with RED LOOK LUTS that you can create in RCX and load into your camera as a look file.

what sort of shot should one use to create this look? Perhaps a well lit chart with a color balance adjustment at slight curve could be a good starting point.

Battistella
 
Last edited:
Let's start with 1d LUTS that you can create in RCX and load into your camera as a look file.
You effectively already have 1D LUTs in camera with curves. I know they are only five point curves, but frankly I would be wary of using a curve much more drastic than can be defined with five points as a one-size-fits-all in-camera look.

3D LUTs however are a different story…
 
3D LUTs however are a different story…

I know. I get that part completely. Just keeping this to what works inside the camera now.

Have you tested any of the new ACES features?

Battistella
 
Have you tested any of the new ACES features?
Not yet, but I plan to.

I was was doing some other testing at RED Europe the other day, and the subject of testing the ACES proxy output came up. But we didn't have the necessary kit available.
 
I've updated the first post with some 1D camera LUTS (generated from RCX) with stills and a download to the look files.

Maybe these ccouls be renamed as 1D camera LUT instead of LOOK file. Just to standardise things.


Battistella
 
Maybe these ccouls be renamed as 1D camera LUT instead of LOOK file. Just to standardise things.
But they are not 1D LUTs.

While it is true that a great many of the parameters in an RMD look could be concatenated into a 3x1D LUT, that is not the case for all of them. As an obvious example, you cannot desaturate an image with a 3x1D LUT. That requires a 3x3 matrix or a 3D LUT.
 
Thanks Nick.

i know, it's a camera look file, which technically is not a 1d or 3d lut, but a red specific thing. Wouldn't red have to create a curve to which a 1d or 3d lut is applied? This is the function of REDlog film, to provide a log curve to which LUTS can be applied? Doesn't Sony and Arri have to do the same thing?

Battistella
 
Let's start with 1d LUTS that you can create in RCX and load into your camera as a look file.

what sort of shot should one use to create this look? Perhaps a well lit chart with a color balance adjustment at slight curve could be a good starting point.

Battistella

Yeah, that's what I did for this curve. I tried matching the contrast of a Kodak sRGB display LUT. Color accuracy is of course off, but it does give you more contrast in-camera (which would be closer to how your final image would look if you used the display LUT in post):
http://www.reduser.net/forum/showthread.php?105312-Kodak-2393-LUT-emulation&p=1275535#post1275535
 
Yeah, that's what I did for this curve. I tried matching the contrast of a Kodak sRGB display LUT. Color accuracy is of course off, but it does give you more contrast in-camera (which would be closer to how your final image would look if you used the display LUT in post):
http://www.reduser.net/forum/showthread.php?105312-Kodak-2393-LUT-emulation&p=1275535#post1275535

If you use the wheels portion of lift gamma gain in RCX you can start to put a bit of that green in the blacks, etc. But I think it's gotta be a delicate thing. I think the overall point is that LOOK files that emulate LUT's are just ball parking LUT's whereas the same LUT applied, in camera, post, monitoring paths, would be mathematically accurate through those various stages.

The LOOK file kinda just goes away at some point, where the LUT can be a constant.

Battistella
 
I think the overall point is that LOOK files that emulate LUT's are just ball parking LUT's whereas the same LUT applied, in camera, post, monitoring paths, would be mathematically accurate through those various stages.
Actually that's not quite correct. 1D LUTs are very accurate within the limitations of what can be done with them, but 3D LUTs are actually something of a trick to approximate complex transforms, and there is a great deal of interpolation going on, which depending on the implementation may give varying degrees of accuracy.

If the transform you are applying is definable mathematically (think basically a list of all the settings in all the nodes of a Resolve grade) then applying the transform as a GPU shader programme with a series of mathematical operations is the most accurate way to do it. A 3D LUT of the same thing is an approximation, albeit often a reasonably good one. Where you are measuring something like the response of a film stock, you don't know the underlying maths, so you have no choice but to put a finite number of measurements into a cube, and then interpolate.

The RED look parameters give a very "mathematically accurate" result within the constraints of what can be done with them. It's just that if what you want to do is emulate a film stock or target specific colour ranges such as skin tones, there are no parameters in there you can use to do that.
 
Thanks for these insights Nick.

As I start understanding the technical side of it better, I think that Redusers will benefit from you excellent insights as well.

Battistella
 
But they are not 1D LUTs.

While it is true that a great many of the parameters in an RMD look could be concatenated into a 3x1D LUT, that is not the case for all of them. As an obvious example, you cannot desaturate an image with a 3x1D LUT. That requires a 3x3 matrix or a 3D LUT.

Read this statement 10 times!
 
Yup.

Camera look files are not LUT's. But people call them LUT's.
Technically, they would be a Red Camera Look file.

NOT a 1D or 3D LUT.

Battistella
 
David and Nick are both right. I think the best you can hope for in an on-set LUT is just as a temporary look that works for the sequence being shot. Michael Cioni has often said he advocates the simple approach of basically using one LUT setting per day as a general guide. I think you can make a case pro and con for color-corrected dailies. But "best light" dailies have worked for many, many productions for dozens of years.

I have heard some major-name DPs complain that the problem with trying to set up complicated LUTs and looks on set is that it drags a lot of time away from their actual photographic process, and it slows down the whole shoot. I think there's a happy middle ground where you're not worrying about touching every single take but just go for some reasonable overall look, particularly a CDL that can be exported and used in dailies and even as a reference in the final color-correction session. But on the other hand, I've been part of a dozen projects where we had the CDLs and/or a high-def reference temp version to match to, and all of it was thrown out while the director and DP worked out a different approach for the theatrical release. Things change, depending on how much time and money you have in the post department.
 
But "best light" dailies have worked for many, many productions for dozens of years.

And for dozens of years we had no video tap. No monitors for WYSIWYG. For dozens of years we had no cinetape and advanced focus assist tools.

Things CAN and WILL change. Even in the most "preservationist" business that INSISTS on sticking it's head in the sand.

It comes down to what you want to achieve. MOST shows today use CDLs to create "best light" dailies - and 98% of the time - these CDL files are really ONLY used for the dailies because any skilled colorist is going to want to start from scratch - maybe he / she uses the CDL files for one day as a reference - but as soon as the colorist understands the intent of the cinematographer - those CDLs are over and dead.

It's best to think of CDLs as the way we used "printer lights" when we made dailies on film. You certainly have more control than changing the overall color temp to make it warner and cooler, etc. - BUT - you are pretty damn limited. Much more limited than adjusting non-destructive, RAW debayer level parameters on a RED camera for example (R3D, Camera Look File, FOOLCONTROL preset, etc.)

I personally believe that CDLs are going to go out of fashion sooner than people think as the new tools evolve. But, if you are shooting TV series with 3-4 different camera types - which happens more than people think - and you are doing fast turn-around dailies on set - then CDLs are your best friend.

And if all you want to do is match cameras or set ups and make the image a little warmer or cooler - or light the mids a little, etc., etc., etc. - than a CDL will get you there. BUT what you can do with a 3D LUT - more specifically - what you can do with a 3D LUT CREATIVELY is a whole other world.

And of course - there are workflows that use ONE 3D LUT for a creative look - and then CDLs ride on top for "best light" adjustments for dailies - and this workflow is what you will see more and more until CDLs become less in fashion.

We will often use digital to analog analogies and say "the sensor is the new film stock" or the "lens is now the film stock" - so .... I would say this -

A 3D LUT allows a cinematographer to literally create their OWN, PERSONAL film stock. One that REACTS very specifically the way he/she wants.

There are only FOUR reasons 9 out of 10 cinematographers are not already using 3D LUTs on every job right now:



1) Hardware limitations in current camera systems & processing "boxes"

Not many cameras accept 3D LUTs internally currently - BUT - EVERY SINGLE new cinema camera made from this day forward WILL support 3D LUTs just as EVERY SINGLE new cinema camera made from this day forward will have WIFI.

MANY external LUT boxes that MANY people use today are HIGHLY FLAWED or very low precision. Ask Steve Shaw. This is changing - and ...

OFFHOLLYWOOD is trying to pull of an ultra-precision, wifi controlled 3D LUT hardware side-car module for DSMC cameras - cheer us on and cross your fingers I can show it at NAB 2015. The specs are WAY past anything out there today. I'm just officially fed up waiting for companies to make the stuff everyone actually wants but they just don't know it yet.

You need some real processing power to do high precision (64^3/65^3 point LUTS) - but fortunately ARMs are cheaper and easier now thanks to mobile device development.

2) Lack of easy to use, inexpensive, intuitive 3D LUT creation software tools

Typically, tools for creative LUT creation have been either full grading tools with limited practicality or on set tools with rudimentary picture manipulation options. This is changing. New tools are coming - I have seen them - testing a game-changer right now. It's happening.

3) GIB (General Industry Bullshit)

FUD. Confusion. Lack of understanding. Lack of education. Everyone wants to call themselves a "DIT" and call themselves and "EXPERT" and yet ... many don't really know what the fuck they are talking about and they just spread more confusion and mis-information. Because there is no test to be an "EXPERT" - you can just put it on your website and business card. Big post companies trying to hold onto their shrinking marketshare by "spinning" bullshit and dis-information. People trying to protect their jobs by putting their job security at a higher priority than actually doing their job - which really SHOULD be - but often is not - to support the cinematographers vision.

4) Hiring the wrong person or consultant or company to support a production.

I have heard some major-name DPs complain that the problem with trying to set up complicated LUTs and looks on set is that it drags a lot of time away from their actual photographic process, and it slows down the whole shoot.

See #3 & #4

I think there's a happy middle ground where you're not worrying about touching every single take but just go for some reasonable overall look, particularly a CDL that can be exported and used in dailies and even as a reference in the final color-correction session. But on the other hand, I've been part of a dozen projects where we had the CDLs and/or a high-def reference temp version to match to, and all of it was thrown out while the director and DP worked out a different approach for the theatrical release. Things change, depending on how much time and money you have in the post department.

A CDL should not even be put the same category as a 3D LUT. A CDL should be thought of as "best light" tool for dailies. THAT'S IT.

A 3D LUT can do - and can be - many different things - including a custom "film stock".

A 3D LUT can paint the entire canvas of a camera or footage's color matrix, providing multiple digital signal interpretations and it can provide more elaborate, direct modification abilities of chrominance and luminance without the need of qualifiers - without dismantling the image - and all done in realtime with less destructive manipulation.

Put that on a T-shirt.

Mark my words - there is going to be a paradigm shift to 3D LUT workflows and camera integration.

Think outside the box. The cube will set you free.
 
Last edited:
Mark,

thanks for posting here. It's an important discussion.

The 3d lut will change how we do things. No doubt.

Youve touched on many issues, one important one being that status quo seems to win over innovation far to often.

Battistella
 
Last edited:
Back
Top