Color Correction using PR (LT)

[Dustin Pearlman]

I'm about to go through post on a feature shot at 4.5K, REDCODE42, Build 21 and production found a post house that wants to color correct using the 1080p ProRes LT proxies that we made on set by the AC during production using REDRUSHES in Apple Color. They have been told that they can use these files in their current state to color correct at a significant savings, and then once there is interest in the film, go back to the R3D files (presumably down to 2K) and output the film.

What are the issues that you all see in this?

Correct me if I'm wrong (and please add to these issues), but this is not the way to proceed.

1. The grades designed on the LT files will not transfer to the R3D. Therefore every hour spent grading the film will need to be replicated.

1a. Time spent grading is the most costly aspect of color correcting. The hourly rate is highest and percentage of overall time working. If this has to be done twice, there is no reason to proceed. Especially if there is a deadline to meet. It is also the creative aspect of post. Having to do it twice sounds like torture.

2. If going back to the R3D is not feasible, then the very least we can do is re-export everything to ProRes 4444 at 1080p. I know there is concern about re-creating the digital zooms and reframes made by the editor, but helping the color space/compression issue would take minimal effort/time/cost and have a great impact on the end result.

Thanks in advance for your impartial guidance.

 

[Zakaree Sandberg]

sounds like a waste of time

 

[Jim Hoffman]

Seems like your time might be better spent connecting your R3Ds to CC. Hard to know without knowing what your editor did but your quality will be limited if you stick with the PR LTs.

Penny wise pound foolish?

 

[Tom.Wong]

you're comparing an 8 bit 422 highly compressed codec to rgb raw. although color doesn't have the best debayer quality on export (i've never been happy with the debayer quality of of color, all my images always looked way too soft) I'd suggest in the very least re-transcoding your r3d to pro res 4x4 or something else high quality like dpx for a much more happy grading period. even tweaking things in RCX for your online so you can get about 80% of the way before the grade would benefit you if time and cost is an issue. (make sure you're monitoring through a calibrated display out of a rocket)

 

[Patrick Tresch]

I've done a grading in Color of R3D shot in 4.5 (wich is a mistake when finished in Color...).

- We conformed everything to DPX 2048x1744 (scope) through RCX in RC2 RG2
- Color corrected in Color with the 2k (squeezed) files
- Rendered DPX sent for final assembly in AE (titles etc.)
- Made a direct to print with Cinevartor (2K)

Buy Davinci. R3D 4,5k is no hussle.

Patrick

 

[Dustin Pearlman]

Tom, yes, I strongly agree. Nothing could be worse than coloring an 8-bit highly compressed file.

Has anybody ever heard of the workflow proposed by the post house? The thought of spending so much time coloring a PR-LT file and then going back to R3D later sounds like uncharted territory, no?

 

[Ben Brainerd]

This workflow sounds, to put it bluntly, retarded. While, yes, it is theoretically possible that you could do a reconform to your RAW footage and apply the grades, you're still looking at doing the whole thing twice, since a grade that's built off a ProResLT isn't going to look right applied to the RAW. If your colorist says it's a simple matter of dropping in the new files he's either an idiot or trying to scam you into paying him twice.

I don't see how the logic of "cost savings" enters into this. Will you save time on renders? Yes. Should they be charging you for hours of renders if they don't have a rocket? No.

Besides, 4.5k support in Color is... Mediocre. If you're going to grade a 4.5k project, you're better off doing the down-scale in RC-X to DPX/ProRes4 and grading that. If your production is concerned about cost, do those renders *BEFORE* taking it to the post house.

To be perfectly honest, I would go looking for a different post house. The fact that they're suggesting this workflow implies, like I said before, that they're either A) At best marginally competent. (AKA They don't know what they're doing) or B) At worst trying to screw you over by locking you into a deal where you're more or less obligated to come back to them again for a final grade.

 

[Dustin Pearlman]

Yes Ben, it is alarming. The "back-up" color workflow that I had been planning for/expecting in case we had no money for color correction was to re-export all the footage in ProRes 4444 (or even 422 (HQ) since there are no composites) at 2K and grade those in Color. This was never the expectation though, because of the budget of the film...

 

[Rick Turners]

I've done a grading in Color of R3D shot in 4.5 (wich is a mistake when finished in Color...).

- We conformed everything to DPX 2048x1744 (scope) through RCX in RC2 RG2
- Color corrected in Color with the 2k (squeezed) files
- Rendered DPX sent for final assembly in AE (titles etc.)
- Made a direct to print with Cinevartor (2K)

Buy Davinci. R3D 4,5k is no hussle.

Patrick

I'm confused.. you graded with the R3D's but in the description says you used DPX transcodes?

 

[Nir Shelter]

Sounds very strange, in addition to what has already been mentioned if the movie has a potential film print in its future grading in a linier space i.e QT is a mistake.
The proposed work flow doesn't seam to save time or money except for producing a poor quality grade in the short term. But if this is meant to attract interest to on sell the film it won't really help much when looking far less than ideal.
But if you only want to show the story why waist money and time grading it at all. Just finish off the cut that you are most happy with and try to get money for finishing the movie properly.

 

[Ben Brainerd]

But if you only want to show the story why waist money and time grading it at all. Just finish off the cut that you are most happy with and try to get money for finishing the movie properly.

This +1million. From any perspective besides a post house that's looking to get paid twice there's no sense in any of this.

To continue my "two options" motif, there's a couple of possibilities for you here. Obviously I don't know anything about your project, your budget, or the people involved. This is all based on my interpretation of your situation using your posts as evidence.

1) Final Product: DVD, with the option of a film print. If you're doing a festival-circuit kind of thing, or just planning on selling DVDs, do a transcode to 1080 ProRes4 or DPX and grade from that. This method will produce a finished quality print that you can use to run off DVDs and, should it become necessary, print to film on the cheap. Will it be the highest possible quality film print? No, but remember, people are printing 1080 to film all the time for theatrical release.

2) The "Finishing Funds" method. If your prime intention is to shop the film around in hopes of getting a distribution deal, I would do a quickie one-light grade in RC-X, or do a Clipfinder conform and grade it yourself in Color. Then, once you've drummed up some interest (and some more funds) take it to your post house and do a FINAL grade.

The more I think about it, the more I think you should *run* (Not walk, RUN) away from the post house that made this suggestion. I can't think of any good reasons for them to suggest it, and a LOT of bad reasons. Besides the ones I've already gone over (They don't know what they're doing, they're trying to scam you/charge you twice) here's a few more that come to mind: 1) They're charging by the hour on inadequate equipment. If a ProResLT grade will result in "significant savings", it implies that their hardware isn't up to higher-end grading, so any extra cost is going to be accrued via paying an hourly rate to sit around waiting for their slow hardware to play catch-up. 2) They're charging you for render time. This kinda-sorta makes sense, as obviously 1080 ProResLT files are going to render faster than 4.5K RAW or DPX. But see "inadequate equipment". And personally I think charging for render time on a Colorist session is kinda skeezy. (Mine you, I've not no problem with offering transcodes as a service. But most of them charge by the hour *of footage* not the actual render time. And they've all got Rockets anyways)
3) Have I mentioned it feels like they don't have a clue what they're doing?

Now, it's *POSSIBLE* (Though, honestly, not likely) that they're giving you a lower rate as a sort of "Indie Discount", in that the ProResLT files aren't going to require them to break out the heavy guns and the giant disk arrays. But it feels more like a "Hey, you can pay us twice" sort of deal.

If you're comfortable PMing me the name of the post house I could do a bit of research and give you a better read on what might be going on.

 

[Dustin Pearlman]

Now, it's *POSSIBLE* (Though, honestly, not likely) that they're giving you a lower rate as a sort of "Indie Discount", in that the ProResLT files aren't going to require them to break out the heavy guns and the giant disk arrays. But it feels more like a "Hey, you can pay us twice" sort of deal.

Ben, thanks for your response. While I agree that their merely suggesting of doing color correction twice is reason enough to "RUN away from them," in their defense, they are offering the LT color concept because we edited the film using LT proxies. It was NEVER my intention to use those for anything but the offline edit, but they suggested we use those existing files (only because they are already rendered/linked) for this first round of coloring to avoid going back to the R3Ds and spending money in a transcode.

That being said, the cost/benefit of going and doing a transcode--to something like PR 4444 or DPX--is so blatantly obvious that it makes me severely question their knowledge of color space, bit rate, resolution, compression, etc. All things that I do NOT want to worry about when it comes to a post house! I would be willing to bet that they do NOT have a REDROCKET either...

 

[Ben Brainerd]

That being said, the cost/benefit of going and doing a transcode--to something like PR 4444 or DPX--is so blatantly obvious that it makes me severely question their knowledge of color space, bit rate, resolution, compression, etc. All things that I do NOT want to worry about when it comes to a post house! I would be willing to bet that they do NOT have a REDROCKET either...

Yeah, I would imagine they probably don't either.

A viable alternative, if the budget is tight, might be to do the transcodes yourself and then hand it off to the post house. It will take a LONG time, given that you shot 4.5K, if you don't have a Rocket, but it's an option. If you've got a computer that you can live without for a week or so you could run off your final shot list.