Lustre/Fuji - Just can't loose the magenta!!!!!!!!

[Angus Hudson]

I'm in the very early stages of grading a film via a DI (Lustre) and we have come up with a few issues which I would welcome your comments and views.

 

Before getting into the "issues", here are a few details -

 

Stock: 35mm Fuji Eterna 500T, 250T, 250D

Format: 2.35:1 Anamorphic

Scan: arriscanner @ 2k

DI: Lustre

Projector: Barco 2k

Output:Arrilaser 10bit 2k

Print stock: Kodak 2383

 

We are trying to achieve a slightly above average desaturated look (-40%ish) with a silver metallic bleach bypassy feel, with strong but ever so slightly bluish blacks, creamy highlights and soft peachy skin tones. Sounds more extreme than it really is.

 

We came up with a "look" that we really liked in our telecined (Spirit) semi-graded dailies during the shooting of the film and we are keen to try creating a similar if not the same look. We do realise, for obvious reasons that they can never match exactly, but we feel/hope that we can get close.

 

The ""Issues".

 

Whilst grading the images on the lustre, how ever hard we try to stop it happening, once we start disaturating the image to the level we want (which really isn't silly desaturated), the skin tones start going flat and become more than a little magenta, the whites and off-whites start feeling a little green.

 

We could not seem to isolate the magenta and even when we removed the magenta totally via the curves, there was still evidence of it in the image.

 

We did a print out test and found the resulting print was stupidly magenta, as in there was a major calibration issue going on to say the least.

 

I am not an expert in these matters but have a gut feeling that "they" have used the wrong LUT and it is throwing everything out, although i have been assured that the correct LUT's has been used........

 

Other issues include solarisation like artefacts in skin tones (looks like really dodgy make up) - seen more on the 8 bit DLP projected image than on the 10bit test print.

 

Our weekly selected printed rushes are absolutely fine and gorgeous (albeit rather colourful) with no hint of magenta/green and the majority of the time we had healthy printer lights.

 

I have suggested that the post house sort out their shop before we continue and get everything behaving the way it should, at which point we review if it is the grade we are try to achieve that is causing problems or if it is the negative, or if it is a combination of the two. Or if it is down to the equipment or operator inexperience/error or even, God forbid, DP error!

 

The post house is doing the right thing - Lustre suites have been changed, tests are being carried out in our absence to resolve the obvious as well as the less obvious issues and initial reports are that the test prints have improved, although I have not seen anything so can not comment.

 

I should add that this is a large and supposedly reputable company, which for the time being I prefer not to mention by name.

 

A few initial questions come to mind:

 

Will the wrong LUT make it impossible to grade out certain hues (green & magenta) despite the power and tools the lustre has at it disposal.

 

Does destaurating an image too much create problems with greens and magentas as well as artefacts.

 

Surely the Lustre should be able to isolate and remove the colours we do not want.

 

I heard rumour that on the "Paris J'taime" project <http://pro.imdb.com/title/tt0401711/> which was all shot on Fuji with 17 different DoP's, needed a tweak to the LUT had to be carried out in order to get the Lustre and resulting prints doing what they should. Maybe I need a special customised tweak.

 

 

It's getting late now and I'm getting tired and I'm in danger of rambling on and on, so I'll stop. But I would be grateful for any thoughts on these matters. In the meantime I will sleep on the matter and try and follow up soon.

 

Thank you and all the best ... Angus

 

 

Angus Hudson

DP London

[Michael Most]

Will the wrong LUT make it impossible to grade out certain hues (green & magenta) despite the power and tools the lustre has at it disposal.

 

It's highly unlikely that the DI facility is using the "wrong LUT." In almost all cases, the only LUT that's applied in a film originated DI is the print preview LUT, which is applied at the end of the processing chain (or, in many cases, in a separate piece of hardware prior to being sent to the projector - this is usually how Truelight works). Its purpose is to simulate what the log image being presented to it will look like when recorded to a particular intermediate stock and printed to a particular print stock. The source image is not what is being evaluated by the LUT, the LUT is only there to serve as a substitute for the film printing process. Now, to be fair, in some color management systems - specifically Truelight - the characteristics of the camera negative stock is part of the calibration. But it's a very, very minor component - it has to be, because most productions wind up using multiple camera stocks. The more significant components are the projector/screen combination, the film recorder, the intermediate stock (or the negative stock, if you're recording to a camera stock), and perhaps most significantly, the print stock.

 

Surely the Lustre should be able to isolate and remove the colours we do not want.

 

In Lustre, as in a number of other systems, one normally doesn't do an overall desaturation when trying to create a specific "look." You would usually use either basic 6 vector secondary correction or color curves to do this. Lustre has both of those tools, and should be able to do what you're talking about pretty effectively. If it isn't, it's probably not the equipment that's the problem ;-)

 

One other way to do what you seem to be trying to do is to linearize the scans on input to the Lustre, grade in linear space, then convert back to log on output (and view it through their "standard" print preview LUT). What this will do is present the color corrector with more saturation and thus more control of its manipulation, albeit at the expense of a bit of detail and color control in the lower midrange. But for isolations, it might be a better source for the secondary or curve controls you'd likely use to achieve that. It will also likely add some noise, so if you try this, be aware that there's a downside.

 

I heard rumour that on the "Paris J'taime" project <http://pro.imdb.com/title/tt0401711/> which was all shot on Fuji with 17 different DoP's, needed a tweak to the LUT had to be carried out in order to get the Lustre and resulting prints doing what they should. Maybe I need a special customised tweak.

 

Read what I said earlier. This would have to do with accurate print simulation, not color correction. My guess is that a different print stock was being used than the one they were calibrated for. In other words, they were printing on Fuji when the LUT had been derived from Kodak prints.

[Angus Hudson]

Thank you Michael, most helpful.

 

I will report back!

 

Angus

[Angus Hudson]

I found these posts on this "Kodak vs Fuji" thread - http://www.cinematography.com/forum2004/lo...php?t11346.html

 

David Mullen ASC

Dec 26 2006, 03:08 PM

One reason people find a color bias in Fuji is that the post chain is heavily optimized for the particular shade/density of the brick-orange color-mask of Kodak negative. You can neutralize a telecine for Fuji's base, and you can use Fuji lab stocks, etc. but most post people work with Kodak and thus it takes a little more tweaking to get Fuji to be neutral when the machines are set-up for Kodak and all the lab stocks are Kodak. Of course, Fuji is designed to work with either Kodak or Fuji lab stocks, but you know that the neg stocks were designed as part of a Fuji series of neg, intermediate, and print stocks.

 

David Mullen ASC

Dec 26 2006, 04:34 PM

Fuji has been lowering the density of their color mask over the years to optimize them for scanning / telecine. But my only point is that you can get neutral color from Fuji with a minimal amount of effort, but not if you're set-up for a batch of Kodak stock. Then you'll be trying to time out a color bias and thinking something must be off with the Fuji stock.

 

 

Any thoughts?

 

Angus

 

Angus Hudson

DP London

[David Mullen ASC]

Fuji does make a calibration film for telecines / scanners that your facility should be using. But one would think that balancing a simple color chart / grey scale should work too.

 

The question is why they haven't called in a Fuji tech rep to see what's up.

[Michael Most]

Fuji does make a calibration film for telecines / scanners that your facility should be using. But one would think that balancing a simple color chart / grey scale should work too.

 

The question is why they haven't called in a Fuji tech rep to see what's up.

 

Unfortunately, these days there aren't a whole lot of Fuji tech reps.

 

While I agree that this could potentially be a scanning issue, in my experience this usually isn't the case. While there are certainly differences in negative emulsions, those difference are manifested primarily during exposure, i.e., the way they capture light. Once the negative is developed, those characteristics are "baked in," but the actual densities presented to a scanner are relatively the same. It's not like the difference between, say, the "normal" Kodak Vision stocks and their Primetime stock, in which one was "normal" and the other had no anti halation backing and a completely different density characteristic. So they should check the scanner setup, but my guess is that any custom calibration of the scanner is going to be minor and not related to what's being talked about here.

 

It should also be pointed out that although I don't know what type of scanner was being used, most modern scanners are self calibrating - that is, they look at Dmin (usually by looking at a frame line), measure it, and calibrate themselves accordingly. In some scanners, the calibration is based on a specific exposure curve supplied by the stock manufacturer or the scanner manufacturer, so in that case, they should be using a Fuji characteristic curve. If they're not, that could potentially be one problem source. This is not usually the case with telecines, which are usually "hand calibrated" by the colorist and thus have far less scientific basis for their setup.

[Angus Hudson]

Little update:

 

In absentia, the post house has carried out various tests and they have attributed some of the of the problems to the DLP. We have now been moved from Lustre suite 3, to Lutsre suite 1 and the tests are printing "better", what ever that means. Just wonder why they have substandard suite in the first place. Oh they joys of a low-ish budget film!

 

Apparently still getting some magenta issues when desaturating or if no magenta the skin just goes grey. The tests are on 2 shots, one is on a day exterior shot on Fuji Eterna 250D, with no filters and bounced natural light as fill. The other is an studio interior on Eterna 500T in pure tungsten light. Also slight problems in both tests with tonal artefacts(?) in the skin tones in areas with heavier make up on the actresses slightly less than perfect skin - Seem to remember D.M. mentioning issues with heavier make up and DI.

 

Fuji tech guy didn't have much to offer that hasn't been said here i.e. there shouldn't really be a problem. Although he did suggest having a go with the new Fuji RDI interneg stock.

 

I'm due to review the tests later this week, as well as a few interesting tests involving DI combined with a bit of photochemical shinnagins to help with the desaturation. At least they are trying!

 

Thank you for your comments.

Edited November 6, 2007 by Angus Hudson

[Dominic Case]

Fuji negative does have a different d-min level from Kodak stock, so it's possible that if the original scan was done with the scanner set up for Kodak, you'd have this sort of problem. Essentially, the digital space wouldn't be optimally matched to the film density range, so you'd have something like under- or over-exposed images in some colour layers but not others. Undoubtedly that would look worse in 8 bits than 10. but it's going to limit the degree of perfection you can achieve.

 

If the post house uses stock-specific input LUTs as well, that could present a similar problem if they missed switching to Fuji settings for this job - even if the right LUT is used with the wrong d-min or vice versa.

[Angus Hudson]

UPDATE:

 

After much work we have managed to resolve most of the problems and were getting on the Lustre and are achieving a very similar look, but not exactly the same for obvious reasons, to what was achieved in Spirit. In fact we are getting better look now. Nothing like lavishing a bit of TLC on to a shot ...... and the silver plug-in.

 

We have found some workarounds to help with getting the level and quality of desaturation we are after without getting the magenta-greys we were experiencing before.

 

However I find it interesting that when we do start to desaturate certain cyan greens tend not to be affected and if we go too far magenta starts to appear. All very strange.

 

DLP/Print/Lustre mismatch issues have been resolved with a custom LUT and are now amazingly spot on or should I say spot on... amazingly ; )

 

I think the issues have come from a post-house that were a little lazy and expecting an easy hassle free job with undemanding clients who needed a bit of a kick up the proverbial to get their shop in order (post-house not clients) - combined with a slightly young and inexperienced, albeit it very good and dedicated colourist.

 

Although somewhat delayed we are now getting stuck into the grade with hopefully most things resolved.

 

Thank you for your advice.

 

Angus

[John Holland]

Angus think you should name and shame them please !! make sure i dont go there .