F5: LUT's in POST?

Liam made a point recently that he considered proxies having no place in a modern workflow. Certainly a multi hour transcode is not ideal.

To clarify, I'm not anti-proxies per se; there are points in a number of different post workflows where a proxy file is useful. It's more transcoding is an is inefficient step if all you are achieving is an editable format. A modern NLE should work natively with XAVC files. And, in an ideal world, it should be able to read relevant metadata. Of course, we don't work in an ideal world...
 
So how do the RED and the Amira deal with LUT's in Post?

Red doesn't need LUTs. They have endpoint software (the SDK) that applies the "look" to the raw in the SDK controls.

The Amira, as I stated before, has the look (LUT) embedded in the QT metadata, and the NLEs apply it for viewing.
 
Nate - Sorry to just be peppering with questions but it is why I posted this in the first place - so I wouldn't be so dense .

Could you clarify the SDK software a bit? i keep hearing about RED workflows in general terms but its not really clear to me.
When I say a LUT in RRAW exposure.

Re Arri : Is the LUT only QT embedded in the Amira? What did they do before that? Do all the main NLE's support them?
Why can't Sony do this ( or do they) ?
 
Len, RED is a RAW workflow. All the colour parameters travel as metadata. ISO, WB etc, it's just data.

The SDK (software development kit) enables software engineers to make tools to control all the metadata from within a particular program. Think of the basic controls you get in Lightroom or Aperture that control the general look of each shot. Workflows will differ dependent on a number of factors, not least what NLE someone is using or the amount of data in their project.

Controlling the look from set to screen has always been an issue so don't get too wound up about it. The best solution has always been clear communication. Also, remember LUTs are just a guide on set where viewing conditions are less than ideal and often the DIT isn't qualified to make critical colour decisions. LUTs in post offer a number of solutions, not least being able to accurately de-log your Sony footage or apply a predetermined look to the shots.

If you are working with a post team who understand Slog. Ask them how they would want additional info - like colour charts, slates etc - some editors like slates, some loathe them, again just ask.

If you are working with a post team who don't understand Slog, then shoot in Custom mode with your chosen gamma to get as close to a finished look as you can. There'll appreciate that and won't make you look like a muppet when they screw the de-logging process.

What I want to see implemented is to have Catalyst Browse work as a plug-in from within my NLE. That way it would be non-destructive and as close to a RAW workflow as I need.
 
Thanks for all the responses and some of it has been helpful, but to be honest I've only gotten a few hints of the info I was hoping for - so I did start looking around the WEB and am planning to talk with a colorist friend and just bend his ear a bit

I was looking for nuts and bolts - i.e what kind of workflows are there for this stuff - how does it work - not general principles about how LUT's work or how I should work with the post team.

Nate was pretty helpful when he described how the Amira and RED have embedded workflows - but how does this stuff worked when its not embedded? Also I know there is more to the RED color workflow than just to say its embedded.

That's why I keep posting lists of basic questions:

In which system(s) does the LUT (or whatever you want to call the metadata) get embedded into the original camera file.
(i.e I seem to have read that with Alexa they didn't and with Amira they do. Does the Sony CINE EI white balance get embedded in anything?)
Do all the NLE's know how to read that metadata?
If the metadata isn't embedded how is it connected (EDL, slates,?) and when does the editor start working with it.
If you're using proxies , how does the NLE reconnect footage later.
How does a DIT get his custom LUT to travel with the camera files?
is there often a kind of "dailies " stage where the Log or Raw footage is given a basic color and debayer before it is sent to the editor or does the editor just get the files and the metadata takes care of a basic look as he cuts it?
Does the LUT used on set often just get ignored after it goes to Post?

I'm used to getting 709 footage, putting in the NLE and cutting . Make color corrections anywhere along the line - but this is a whole new world.

Thanks anyway . I guess I was just trying to get around doing a lot of homework.
 
Last edited:
Thanks for all the responses and some of it has been helpful, but to be honest I've only gotten a few hints of the info I was hoping for - so I did start looking around the WEB and am planning to talk with a colorist friend and just bend his ear a bit

I was looking for nuts and bolts - i.e what kind of workflows are there for this stuff - how does it work - not general principles about how LUT's work or how I should work with the post team.

The responses are broad because workflows can vary so much. The key is to fully understand the principles.

In which system(s) does the LUT (or whatever you want to call the metadata) get embedded into the original camera file.

All the main cameras embed that info in some form or another the bigger question is how can that info be used in post (hence why I want Cat browse as a plug-in in my NLE).

Does the Sony CINE EI white balance get embedded in anything?)

WB in CineEI is both baked-in to the footage and noted in the metadata.

Do all the NLE's know how to read that metadata?
Nope. Well not all of it and not in a useful way. That is the issue.

If the metadata isn't embedded how is it connected (EDL, slates,?) and when does the editor start working with it.

It's not connected to EDLs (who uses EDLs anymore?) or slates. Metadata is a separate file that is linked to the media file. It's up to the individual editor when he applies LUTs or looks etc. Most like a basic look to edit with and then finalise the grade once the edit is locked.

If you're using proxies , how does the NLE reconnect footage later.

It's usually a function of the NLE or graphics/grding program or you can do it via an XML file.

How does a DIT get his custom LUT to travel with the camera files?
He/she copies it to the drive or emails it.

is there often a kind of "dailies " stage where the Log or Raw footage is given a basic color and debayer before it is sent to the editor or does the editor just get the files and the metadata takes care of a basic look as he cuts it?

You might do this for clients.

Does the LUT used on set often just get ignored after it goes to Post?

Depends on the LUT and how important it is to the production.

I'm used to getting 709 footage, putting in the NLE and cutting . Make color corrections anywhere along the line - but this is a whole new world.

In Premiere or FCPX or Avid you can still make colour corrections on the timeline at any stage. If you are editing natively, it's not that different tot your traditional workflow really, just one extra step, delog.
 
Len
Im still on holiday so still typing on my phone.

firstly I think you need to have a clear motivation to move from your background of getting it right in camera.

if you paint your camera correctly and light correctly there is no need to go down the route of shooting log.

Prrsonally I am used to painting my colour and tonality in post - having started in a lack and white dark room that is where I 'come from'.

I like "burning" my skies in post for example- because one can burn a sky with more precision than one can use an Nd grad on set.

shioting log opens the door to fully bending an image in post - but unless you or your clients want to do this bending then there is little value to shooting log.

once you have aligned your "desire set" then the workflow will fall in to place - or at least the strength and weakness of different options will become apparent.

-----

Here are three simple workflows.

1) shoot slog in EI mode (not custom) view the stuff in cat browse export all to prores. Deliver.

CB sustaines your EI choice and auto de logs EI footage.
CB does not auto untangle slog custom - but can with mild intervention.

2) shoot log in any mode custom or ei. Load all into resolve add lut, correct exposures (on shots made in EI mode not at 2000) make optional further changes. Export all to pro res. deliver.

3) shoot log in custom mode or EI mode only at 2000, load all into resolve add lut, optionally add further corrections,export to pro res. deliver.

avoiding EI mode not at 2000 circumvents resolves lack of ability to read EI choice.

--
2) is the most complex but gives You the most options in camera and post.
 
Last edited:
..the options above give "edit ready"material to your client but close their options for further deep colouring without round tripping to the original material.

Those options are not too efficient with data management and take transcode time. Neither is ideal.

Liam can probably outline more complex and flexible options that rely on the specific nle of your client, rely on them having a deeper skillset. Those options may be more efficient with both data and time - but may rely on your client having a computer with more power for real time rendering.
 
1) shoot slog in EI mode (not custom) view the stuff in cat browse export all to prores. Deliver.

CB sustaines your EI choice and auto de logs EI footage.
CB does not auto untangle slog custom - but can with mild intervention.

transcode into ProRez -or- DNx for editorial and then send the camera originals to finish / grade, and we re-link the edited timeline to the camera originals, and throw away the transcodes

2) shoot log in any mode custom or ei. Load all into resolve add lut, correct exposures (on shots made in EI mode not at 2000) make optional further changes. Export all to pro res. deliver.

Again as long as the camera originals go to the grade suite, no one would want to grade a show with baked in colour choices, and it will take longer to un-do the accidental damage done, and therefore cost the producer more in the end...


3) shoot log in custom mode or EI mode only at 2000, load all into resolve add lut, optionally add further corrections,export to pro res. deliver.

again along with sending the camera originals is important

Should note that there is another established workflow that is being ignored, that is the CDL, a colour decision list, that sidecars through any modern NLE and grading/finishing system, is software agnostic, is non-destructive and works all day long...

load the camera originals into amy modern DiT systems like Pomfort, Nucoda Look, Scratch, Resolve etc etc, and do a quick grade, send the resulting CDL along with the AFE / AAF / XML and camera originals to editorial and the colour metadata appears in the NLE's colour grading tool.

Send the final show via AAF or XML and export the CDL into any top flight grading system like Baselight, Nucoda, Pableo, Mystika, Resolve and the same colour information now shows up in those systems base colour layer / node

But unlike a LUT, the CDL can easily be used as a base to build upon and is non-destructive
 
Dermot - you use "show" and here is a thing - I'm not really talking "shows" but simple corporate work.

you also mention "accidental damage" - well some might not have accidents :) - or certainly might be able to outdo camera 709 footage very quickly

Of course jobs of different scales should have different approaches - my primers were for getting going as a log shooter looking to deliver to a client who is not a colourist and is used to receiving ready to edit material.

Good point about not deleting the originals of course. And much to follow up on CDLs

The most valuable part of my post is how the different camera settings interact with various approaches to post.
 
Last edited:
But unlike a LUT, the CDL can easily be used as a base to build upon and is non-destructive

I was never happy with Sony's color "science", and had to resort to ACES and LUTs to get a base look that I liked. Color transforms and highlight roll off a CDL could never do.

I know this is slightly confounding to a colorist, but as DPs, we need to have a camera color/gamma/etc response in our heads (that we like) to visualize.

Now that I have an Amira, I understand that much more of the Amira's look and color response is created in camera, and then transformed to LogC before recording. You can undo the LogC via LUT or by hand and still have it look good.

Much more of making the Sony's look good is a post exercise, and I leaned on LUTs a lot to make it the way I wanted it.

This is another reason why I went Amira...in the hands of almost any colorist, experience level aside, it will look a base level of "great".
 
Thanks guys - i am beginning to get a handle on this better and appreciate the input. Clearly I need to do a lot more research than just this thread. Time for some serious schooling.

Ciouple of questions tho :)

I am not at all clear yet how this process is fundamentally different on RED or Arri and is the Amira a significantly different flow than the Alexa was?
I noticed going on an Arri site there was an Amira flow chart showing white balance as a stage before Log C was added, while my crude understanding is that in CINE EI , most of it is added afterwards. Do I have that right and is that part of what Nate is saying.

Am I also right in gathering that metadata, LUT data , anything DIT might make are carried in separate files alongside the basic XAVC or ProRes files created by the camera? i.e. in an XDRoot folder I see .mxf files, .BIN files and XML files all with the same name for each shot. Where is the LUT and other metadata stored in the MXF itself or only in BIN or XML?

Similarely - if a DIT create a LUT on set what form does it take - the same name/number as the shot with some other file extension?
 
I was never happy with Sony's color "science", and had to resort to ACES and LUTs to get a base look that I liked. Color transforms and highlight roll off a CDL could never do.

I know this is slightly confounding to a colorist, but as DPs, we need to have a camera color/gamma/etc response in our heads (that we like) to visualize.

I was a working DP for more than a decade before turning to post, mainly to stay home with my young family… i and still shoot something every year… i have a clue about what you are saying re; creating the look in camera

CDL is best used to get something close enough to edit with, not finish or be concerned about highlight roll-off, that's where the colourist comes in, with accurate mons and scopes, and works on the shots in context…
 
I was never happy with Sony's color "science", and had to resort to ACES and LUTs to get a base look that I liked. Color transforms and highlight roll off a CDL could never do.

I know this is slightly confounding to a colorist, but as DPs, we need to have a camera color/gamma/etc response in our heads (that we like) to visualize.

Now that I have an Amira, I understand that much more of the Amira's look and color response is created in camera, and then transformed to LogC before recording. You can undo the LogC via LUT or by hand and still have it look good.

Much more of making the Sony's look good is a post exercise, and I leaned on LUTs a lot to make it the way I wanted it.

This is another reason why I went Amira...in the hands of almost any colorist, experience level aside, it will look a base level of "great".

Hi Nate - just a quickie comment to say I clicked through to your website and was really impressed with it - very strong and stylish and not over complicated.
Just out of interest from a current Sony F5 user, I would love you know your thoughts on your Amira and what lenses etc you are using with it. Seems like a very very interesting camera. If you have a blog or something on your experience please share.

Best wishes and happy 2015 , Sparky
 
Len - I for one am not really sure what data is carried or how.

simply I've kept the folder structures intact and kept all those sidecar files for safe use in the future

Relevant to today is that mostly this data is not interpreted by nle's. apart from CB which is miles ahead in this front

catbrowse sees
EIvalue if in EI mode
that you were in EI or custom mode
possibly the lut.

davinci sees
nothing

others?
ive not tested -suspect nothing

with that reality use of a board or some other form of shot notes is prescient

afaik red is "seen" by resolve, red cine x and maybe others - alexa - no idea - not used

--
wbalace is baked in and therefore an inherent portion f the image so not really an issue (unless shooting with the raw box)
 
Last edited:
As long as you shoot EI 2000 or in custom mode with slog3 profile then all you have to communicate for 90% of shots is..
"please delog all shots using slog3 to lc709 lut"
if the editor post people get that you should basically be golden!
 
It only gets a little more complex if you have decided the cam shines brightest when shooting EI 800 where possible.

in that case it is easy for your stuff to hit the nle looking 1.3 stops over exposed!
 
Back
Top