ideal grading workflow !!! (luts, colorspace etc) ???

Hi friends,

what is an ideal way to grade digital movies? Generally i get to red epic or red mx footages.
i dont use any lut or change to any other color space.

I want to know an ideal workflow of grading.

what is the input lut to use ???

when is use of rec 709 lut ?

does it make any difference if i grade with lut or without lut???
I mean what is best way to make sure that what i see on my screen should be seen on theatres ????

thanks in advance.

Krish Mehta
Davinci resolve 10

How do you get any jobs as a colorist without knowing the answers to these questions????

How do you get any jobs as a colorist without knowing the answers to these questions????

I think he's asking what random buttons to click inside this buggy, useless program to make it work. In his defense, I know the right answers to these questions but inside Resolve 10 it still makes no difference - you might get lucky and get the footage in, but it's never coming back out in any useful way.

My advice - give up. And for God's sake whatever you do, don't start throwing money at it thinking that's going to solve anything. 10K in a computer, $1000 dongle, lots of reading and fruitless months later I've only just recently realized I've been had.

what is an ideal way to grade digital movies? Generally i get to red epic or red mx footages.

I would do three things:

1) buy and watch one or more of the Resolve Tutorials from excellent instructors like Alexis Van Hurkman or Warren Eagles

2) read the entire 734-page Resolve manual from cover to cover while sitting at the controls and go through every menu and every knob.

3) read Van Hurkman's excellent book Color Correction Handbook (available from many fine online book suppliers) and note in particular the chapters on workflow, LUTs, and signal processing.

There's also some very good advice on this website and the LiftGammaGain.com website.

[insert smartass reply here]

Bob Bozenbeck = Art Moody???

Bob Bozenbeck = Art Moody???

I knew it from the first post of him in the forum

Wow. what a "friendly" forum here ....!!!
What i meant is i dont use Luts....and wanted to know general workflow of other people.

Looks like there are no colorist here with an ideal workflow.

thanks Marc !!!

But i just wanted to know general workflow (not that i dont know one) !!!

"My advice - give up. And for God's sake whatever you do, don't start throwing money at it thinking that's going to solve anything. 10K in a computer, $1000 dongle, lots of reading and fruitless months later I've only just recently realized I've been had".[/quote]

Bob, since you gave up with DaVinci Resolve, please share which grading are you now using and are you pleased with it?

Thanks.

Looks like there are no colorist here with an ideal workflow.

Because there is no "ideal" workflow where there is some kind of magical solution or wand to wave and the software reads your mind and spits whatever out so that it can be used in any context.

Why is it that so many people jump in (wrong-head first) and the first thing they want to do is overcomplicate things so that it looks like they're doing something really difficult, when what they are doing is messing themselves up, probably overtaxing their hardware platform, and in no way understand concepts of colorspaces, gamut transforms, how to read a scope (if they have one) and fundamentally assume that all software is the same, that something like Resolve is just Photoshop for movies.

Rec709 is a description of image media that is suitable for delivering broadcast programming. Resolve works in 709, pretty much by default and does not require any other interventions. A "LUT" is not a "look", at least it did not start out that way. A Look-Up-Table, as has been discussed at great length, is simply a method of mapping one set of values to another. If you are working in 709, you do not require an input LUT if you are dealing with footage shot in 709. If you are working with a calibrated broadcast display monitor you do not require a display LUT. In a sense, every correction is a LUT. There are some tables that have the specific function of transforming the image values found in a particular medium to suitable values for display in another. We used to have daily re-calibrations to apply Look Up Table transforms to final masters for accurate film-out reproduction. The reason for this is that the laser RGB values used to expose Original Camera Negative (various film stocks, and yes, real emulsion) stimulate the dye layers non-linearly, so we have to alter them so that the release prints kind of resemble the digital grade. And still, the lab guys are still tweaking away because of bath variables, concentrations, chemical exhaustion, temperature... stock variance... at once, this is an exact science and a flying-f**k at a rolling donut. Even DCI is not so clear-cut. Clearly defined is not the same as easily achievable.

If you are not operating in a completely controlled environment, calibrating all your references to specific display preferences with clearly defined delivery goals, then it will be extremely difficult to achieve consistent results. Today we have to take an ever-expanding roster of wildly different delivery platforms. I almost pine for the days of NTSC and PAL... almost. But even with such a narrow definition, the multi-million dollar suites of not-so-long-ago required daily attention to stay on the numbers, employing high-price, high-time professionals with proprietary equipment that worked. I have a collection of T-shirts, but it almost seems quaint and faintly ridiculous now (never mind if they still fit).

It seems almost hilarious that complete neophytes expect the same performance, and with commodity assets, when what is really happening is that the "workflow" is just getting more chaotic, ambiguous and capricious.

Latest pet peeve... what is "4K?" Because one manufacturer says its "this", but you still have to deliver "that", and the commodity suppliers are rubbing their hands together in glee, because they are, in a practical sense, incompatible.

jPo

Bob Bozenbeck = Art Moody???

I thought it was "Mooney"...or "Moony?" Ah, well, what is it about a Rose by Any Other Name?

This is a very old forum-spoofing strategy. I think I had about a dozen identities on an F-1 list, like almost everyone else on it, and eventually a private club formed where we would post acrimonious disagreements, which were often against our proxy-proxies and one proxy would be telling another proxy they were idiots and full of tiot. All the while having a calm discussion by private email. Just to stir the p**s. Anyway, if you check the date of disbarment and re-spawn... [musical note] that's entertainment [/musical note].

jPo

I thought it was "Mooney"...or "Moony?" Ah, well, what is it about a Rose by Any Other Name?

From what I remember it was "Art Looney"

thanks Marc !!! But i just wanted to know general workflow (not that i dont know one) !!!

You have to understand the workflow in context. If you don't understand everything else, knowing the input LUTs, output LUTs, display LUTs, and so on won't help. Just getting the monitor right is a huge part of the equation.

I find often the best approach is to understand where the project is going to end up -- let's say, a network 1080i delivery. Work your way backwards from there, and make the decisions on how the project needs to be edited, conformed, color corrected, and then prepared for final delivery with the final titles and audio.

A big thanks to JPO and marc too.

Guys, I would really appreciate your help on this one,

I am grading on resolve with a calibrated pro plasma. The final target is DCI p.


I have already graded a project {happily approved by DOP}. Since resolve is Rec709 and my target is DCI
what could i do so that what i see on my plasma looks more or less the same on cinema projector.

some options which i can think of.....
1) change settings in plasma to DCI mode (if any) and then grade in TRACK mode, match color correction to one which has been approved.

2} add an input lut to project (rec 709 to linear or anyone...do tell me if you know which lut should be used) and then do correction on track mode to match original CC.


Kmeh

[insert smartass reply here]

Hi Kmeh,

I would recommend to find a lut for your export only, but you need to test the DCP.
You're grading on a rec709 monitor, so if it's well calibrated, find the good lut between your rec709 grading and the dcp software. It can be rec709 to DCI, rec709 to sRGB, rec709 to XYZ...
Are you making the DCP also ? Which software ?

Regards.

Hi Dascolor lab

I am not making DCP myself....we have few companies doing it here. Is there any DIY way of converting dpx to dcp ???? If i can do it myself...it can be cost effective too !!! Easy dcp from resolve....our company is up for it.

As far as lut ......if i add an export lut....it would be difficult for me to match the current grade and would involve lot of guessing game. But i think i will add an input lut and match grade.....and will give for dcp test next week.
fingers crossed.

Kmeh7

If you are working in 709, you do not require an input LUT if you are dealing with footage shot in 709. If you are working with a calibrated broadcast display monitor you do not require a display LUT. In a sense, every correction is a LUT. There are some tables that have the specific function of transforming the image values found in a particular medium to suitable values for display in another.

So, you only need an output lut to fit what the dcp lab want, or they can do it for you if they can process your rec709 files.

Regards.

you are right i guess.

But my only concern is that when i add output lut i dont have any control on output as in how it is going t o look on projector.

Will it be ok if i add 3d lut on the track mode and changes along with it ???

thanks again

kmeh7

"JPOwens wrote:
If you are working in 709, you do not require an input LUT if you are dealing with footage shot in 709. If you are working with a calibrated broadcast display monitor you do not require a display LUT. In a sense, every correction is a LUT. There are some tables that have the specific function of transforming the image values found in a particular medium to suitable values for display in another.[/quote]

If you could explain bit more on "footage shot in 709" ?????

My footage is RED and project settings are
Color Space : RedColor
Gamaa curve : Red Gamma

thanks again !!

Kmeh7

Hi Kmeh7,

first of all this is a great forum just like the one in http://www.reduser.net/

It's just that all of your questions are have been answered in many forums and tutorials but you have to look for them properly (sometimes you have to pay for the tutorials). It's just a matter of finding them properly. Hence some people put where to look for in the net or video that other people post for free. Trust me there are so many good people out there how are willing to share their knowledge. I think that some people doesn't want to explain all over again about the workflow because it is out there, so it just looks like you asked the questions before you do your research. That's what I do most of the time. I would go on research and look at different forums to look for anyone who has the same questions as I do. Most of the time the answers are always there.

I really suggest that you go through the forums in reduser.net they have all the answers there for you.

If you want to know about workflow for those Hollywood movies here is one example of a Color workflow that was done by the good people of Light Iron on the movie Ender's Game.



Do check out their other videos as well. Very useful information.

As for color space and LUTs and color history, check out FilmmakeIQ



Their other videos are quiet simple and understandable and useful too.

Cheers

f you could explain bit more on "footage shot in 709" ?????

My footage is RED and project settings are
Color Space : RedColor
Gamaa curve : Red Gamma

Okay, so you are shooting with a more or less undifferentiated Bayer-pattern sensor which produces RAW data. The settings you are using are metadata modifiers that are placing the image values in a particular color space and display characteristic -- similar to a LUT. That you are using REDColor places your images in a similar characteristic to Rec709 (Broadcast HD), as regards white point, gamma transfer, etc. If you change those metadata settings in Resolve, choosing some other contrast, gamma transfer, whatever, then you are going away from the decoding/scaling that the nominal image value flow-through is expecting. If you are viewing your graded output on a standard 709 broadcast monitor, you can be satisfied that what you see is what you get. That is why we have transforms that re-place the values for viewing under other circumstances -- simulating film projection or DCP digital projection, because those are obviously very different colorspaces, with a different gamut, gamma transfer and white point.

If you really want confusion, try translating your project through the ACES workflow. It will throw away your metadata settings and make assumptions about your operating and delivery environment. For that, you must specify what you are shooting with, and what you are monitoring with, and if those don't match, you are out to lunch.

jPo

Thanks Jpo and Hendro.

Hendro, i did go through the link you gave me. Some of them really useful.

thanks again.

Kmeh7

Guys,

I read some articles and realised that the safest way to convert from Rec709 to Dci is to ask the lab to do it.

Do most of the colorist do it that way....i mean rely on the lab and let them do it ???? or they do it themselves ???


Kmeh7

There is an investment/confidence level at play here.
Mastering DCP/DCI is a business, and the confidence factor is related to what sort of resources have been brought to bear, and the experience level that goes along with it. Yes, anyone with the software can bang out a DCP package with varying results -- expensive stuff that has a rock-solid XYZ conversion, and freeware that might just work as well - but if you don't have any way of looking at the result, its just a crapshoot.

I do make my own DCPs -- I have a variety of applications to do it -- I get different results from some of them, and there are packages that have known issues depending on the theatre server. If you are going to make the DCP yourself, you must be prepared to back it up, guarantee its compatibility and fidelity. You don't have to be licensed by the state or anything, so it is your choice. If you do not have the technical resources to verify your work, there really is only one way to guarantee the result, and that is to have some other facility take on that burden.

jPo

safest way to convert from Rec709 to Dci

Thinking a bit more about this. The XYZ conversion is likely the least of your worries, if creating your own DCP.
Cooking out a dpx or jpg sequence would be fairly straightforward although your accuracy is dependent on whatever LUT or conversion algorithm has been put in place. But its a well-travelled path by now and just about everybody has it right, whether its out of Resolve, Fraunhofer easyDCP, or nearly whatever else.

The fun begins when you start authoring the package. Flat, full, scope? SMPTE? InterOp? Sound? 23.98? 24.000?
Encrypted? CPL File Name compliance? Linux? NTFS? Is the target server at the theatre compatible with your authoring version?

So how do you play it back to verify the package before it is shipped off to some remote location from where you will (and this is like the two kinds of Indy drivers: the ones who have hit the wall, and the ones who are going to hit the wall) get a last-minute panic phone call from someone "who doesn't understand these technical things" that there is something-I-don't-know-what wrong with your movie and they are cancelling the showing. It might be because the traffic department is utterly convinced they need a KDM to mount the show (and you know you did not send an encrypted package, because you didn't spring for the full authoring suite, so you can't even do that). Or have to contend with re-processing license certificates and show windows for every theatre on the planet in perpetuity.

Present situation is actually ludicrous.

jPo