[No Longer Available] Die Hard With A Vengeance Regrade (35 Project #2)

[CSchmidlapp]

Big thanks to Beber for all the pics and information!

Die Hard 3 suffers from the same shift in colors as True Lies, or Death and the Maiden or even maybe The Empire Strikes Back, as the grey costumes look very much more blue on a photochemically color-timed film.

So is the grade done to accommodate this shift when the release film is printed.
If so the IP needs this process adding in to accomplish the desired look, and scans of IP's would be much more magenta than desired, and if not corrected, would be wrong.

Ive been wondering if this is one of the reasons Old Digital Masters we're often to Red.
I understand that it was most probably to accommodate flaws in the display technology's at the time, but Im just putting it out there
With that I'm also pretty sure if this DHWAV master was just a straight scan if the IP with no tinkering, PDB's idea of using a blanket LUT would have corrected it.

Anyway Brain Fart over

[PDB]

Do you think alot of the grade differences are created during the Release Print process based on stock used?
If so, the characteristics of particular release film stock must get taken into consideration when grading the film.
Do film stock emulation LUTS add this variant?
It seams DHWAV has that 'Teal' look we have been seeing on modern masters, especially the 4K ones.

The further I go, the more I think the film stock used for the release print does have influence over the color of the film.

Certain stocks are know to have color biases. Eastman/Kodak LPP's can be yellow/green. Fujis can be blue/purple. Afga's seem to be blue-ish also. That is, of course, a gross generalization but you get the point.

That's not only cinema film but still film also. I often used different film stocks to get different effects when shooting 35mm still film.

If you go back and read a lot of American Cinematographer you see Directors and DOP's choosing (on big productions) release stock to make sure they get the look they want. So they can and do factor it in. Unless you are low budget, then you get what you get. But there is only so much control they have. The lab doing the processing could screw up in any number of ways and you heard tales of production staff going into theaters to "test" the look of the film and saying it was off or wrong.

And then there is the debate over the projector's influence over color as well as "brightness".

Compounding, when a film goes overseas, it might not be printed as intended depending on the studio and lab in charge. Certainly it might not the stock you might want. For example, a US film printed on Eastman going over to Japan is more likely to be printed on Fuji since that is (was) the dominate film company there.

Now, I am going to walk this back a bit. I don't want to overemphasize release print stocks influence since if it was that big of a deal, why bother color timing at the IP stage at all? So I won't expect this print is bluer and this one redder. More like this one is bluer and this one more sky blue or teal-ish or something like that. It's analog technology, you will never get a perfect match.

So printing variances and film stock could account for some of the difference we see in prints. In the end there are a lot of variables once the film leaves the IP stage that can sometimes change it's look. That is the reason many directors and DOPs embraced DCI/DCP. It's totally control over the picture all over the world (with the only variations being the projector tech used). You don't have to worry about your perfect grade being different in two places.

As for the teal thing (or green). I think it safe to concluded that has been around for years and in not a modern thing. Looking back to surviving LPPs from the 80s we have seen that productions just like that look of teal and orange skin tones. It's just a nice contrast that color guys loved. With computers they pump it up to the nth degree but in the chemical days it was a little more "natural" since the color "settles" differently in the highlights and shadows.

[PDB]

Big thanks to Beber for all the pics and information!

Die Hard 3 suffers from the same shift in colors as True Lies, or Death and the Maiden or even maybe The Empire Strikes Back, as the grey costumes look very much more blue on a photochemically color-timed film.

So is the grade done to accommodate this shift when the release film is printed.
If so the IP needs this process adding in to accomplish the desired look, and scans of IP's would be much more magenta than desired, and if not corrected, would be wrong.

Ive been wondering if this is one of the reasons Old Digital Masters we're often to Red.
I understand that it was most probably to accommodate flaws in the display technology's at the time, but Im just putting it out there
With that I'm also pretty sure if this DHWAV master was just a straight scan if the IP with no tinkering, PDB's idea of using a blanket LUT would have corrected it.

Anyway Brain Fart over

Well the prevailing theory around here is that old HD/DVD masters are red to compensate from the how the digital master would be played on CRTs. It makes sense to a certain point since CRTs would of been they prevailing technology when a lot of older master where made and they would of been used to color time the film (circa late, late 90s and early 2000s)

I've tried a few times but the one LUT for the whole film has never worked for me Although I wish it had. I find I have to break the film up into several pieces, all the individual scenes or (like Blade Runner, Alien, Road Warrior, Evil Dead) go shot by shot. The shot by shot is a pita and why some of my projects take forever.

Tservo is a good guy to get into this conversation since he has seen a ton of 35mm print and often points out the strong color difference with the current masters. How even a lot of 80s films are teal and green.

[zoidberg]

As has been mentioned before a straight scan of an IP, even a timed one can look flat and disappointing compared to release prints. Here's Torsten Kaiser:

"TK: Unfortunately, and through no fault of their own, more and more colorists don't know anything about the various photo-chemical color processes and their differences. Now the vast majority of colorists work with new negatives or with new IP [interpositive] material. They work a lot for television - with what we call samples for new projects - any they don't have ties to older material. As a result, errors are being made all the time. If you take a look at the Blu-ray release of Galaxy Quest, for instance, it's clear that the transfer was done from an IP. How do I know this? How can I determine that without having worked on the project? Well, it's pretty easy. An IP always has a much higher black level so that it doesn't crush the film element itself. That can only be achieved if you make it as flat as possible. So you don't have a very contrast-y image, but have all the stops in there. Or at least close to it. That's different than a print, which is very contrast-y – it looks very sharp and very punchy when you hold it against the light. On a telecine, a print would really give you a run for your money because it is very difficult to get all the detail out of that thing. However, with an IP, you can get all the detail out of it, but you have to remember the IP is developed to be very flat. As a result, you have to mimic the process from the IP to the intermediate negative to the print; the print being the ultimate end stage; the answer print, one should say, since that is the one being QC'd by the maker. So if you take an IP with the rather flat imagery and reproduce it only as it is on the IP, it doesn't work because the colors don't register as they should. Whites don't appear as they should, and so forth. And that's why Galaxy Quest doesn't work. The Galaxy Quest prints look stunning. They're an absolute knockout. It's the only picture that I was so frustrated with that I changed the settings on my projector. Fortunately, I have a projector which has different user levels, and that's exactly what I used when watching Galaxy Quest. Otherwise, I would have been like Peter Finch in Network. I would have gone to the window and screamed I was mad as hell and I'm not going to take it anymore. It's really that bad."

I think a lot of films out there are basically scans of IPs with little to no adjustment. As mentioned above a release print is easily two generations separated from the IP which allows the contrast and colour to 'develop' somewhat into what we see on the screen, this would be anticipated by the filmmakers and colour timer.

All of this (and watching 35mm scans) has got me thinking of ways to replicate photochemical timing, mainly the Printer Lights method. If it bears fruit I shall share my findings here. If not, hey at least it keeps me off the streets

[CSchmidlapp]

Well the prevailing theory around here is that old HD/DVD masters are red to compensate from the how the digital master would be played on CRTs. It makes sense to a certain point since CRTs would of been they prevailing technology when a lot of older master where made,and they would of been used to color time the film (circa late, late 90s and early 2000s)

I've tried a few times but the one LUT for the whole film has never worked for me Although I wish it had. I find I have to break the film up into several pieces, all the individual scenes or (like Blade Runner, Alien, Road Warrior, Evil Dead) go shot by shot. The shot by shot is a pita and why some of my projects take forever.

Tservo is a good guy to get into this conversation since he has seen a ton of 35mm print and often points out the strong color difference with the current masters. How even a lot of 80s films are teal and green.

I think a lot of films out there are basically scans of IPs with little to no adjustment. As mentioned above a release print is easily two generations separated from the IP which allows the contrast and colour to 'develop' somewhat into what we see on the screen, this would be anticipated by the filmmakers and colour timer.

All of this (and watching 35mm scans) has got me thinking of ways to replicate photochemical timing, mainly the Printer Lights method. If it bears fruit I shall share my findings here. If not, hey at least it keeps me off the streets

Yes it is certainly true about CRT'S, well more on the consumer market.
I played a few modern Blu-rays on and old CRT was still have floating around the house and they looked completely drained of red. The test was no way scientific though, it was being played out on an old Geforce 8800GT that had a composite out adapter.

If the IP was the source for a old master, the white balance would be off to compensate the introduction of the release prints characteristics. The old video masters would have certainly been white balanced, probably on the transfer process, knocking the original grade way off. This was probably done scene by scene also.
So by that methodology adding a blanket LUT to reintroduce the missing elements of the 35mm would require a scene by scene white balance adjustment on the source! Very time consuming. It's a full on remaster.

When doing your color correcting projects do you add the LUT' first and correct your source to that?
I gather a LUT that replicates the photo-chemical timing could not be touched, and you have to correct what you are feeding it.

I wish I had more time and resources to experiment more. It would certainly nice to go 10bit

[PDB]

As has been mentioned before a straight scan of an IP, even a timed one can look flat and disappointing compared to release prints. Here's Torsten Kaiser:

"TK: Unfortunately, and through no fault of their own, more and more colorists don't know anything about the various photo-chemical color processes and their differences. Now the vast majority of colorists work with new negatives or with new IP [interpositive] material. They work a lot for television - with what we call samples for new projects - any they don't have ties to older material. As a result, errors are being made all the time. If you take a look at the Blu-ray release of Galaxy Quest, for instance, it's clear that the transfer was done from an IP. How do I know this? How can I determine that without having worked on the project? Well, it's pretty easy. An IP always has a much higher black level so that it doesn't crush the film element itself. That can only be achieved if you make it as flat as possible. So you don't have a very contrast-y image, but have all the stops in there. Or at least close to it. That's different than a print, which is very contrast-y – it looks very sharp and very punchy when you hold it against the light. On a telecine, a print would really give you a run for your money because it is very difficult to get all the detail out of that thing. However, with an IP, you can get all the detail out of it, but you have to remember the IP is developed to be very flat. As a result, you have to mimic the process from the IP to the intermediate negative to the print; the print being the ultimate end stage; the answer print, one should say, since that is the one being QC'd by the maker. So if you take an IP with the rather flat imagery and reproduce it only as it is on the IP, it doesn't work because the colors don't register as they should. Whites don't appear as they should, and so forth. And that's why Galaxy Quest doesn't work. The Galaxy Quest prints look stunning. They're an absolute knockout. It's the only picture that I was so frustrated with that I changed the settings on my projector. Fortunately, I have a projector which has different user levels, and that's exactly what I used when watching Galaxy Quest. Otherwise, I would have been like Peter Finch in Network. I would have gone to the window and screamed I was mad as hell and I'm not going to take it anymore. It's really that bad."

I think a lot of films out there are basically scans of IPs with little to no adjustment. As mentioned above a release print is easily two generations separated from the IP which allows the contrast and colour to 'develop' somewhat into what we see on the screen, this would be anticipated by the filmmakers and colour timer.

All of this (and watching 35mm scans) has got me thinking of ways to replicate photochemical timing, mainly the Printer Lights method. If it bears fruit I shall share my findings here. If not, hey at least it keeps me off the streets

Well, I agree with Torsten, that too many masters nowadays lack contrast and many prints are way more dynamic then HD masters. I think many HD masters are too flat and lack a dimensional look to them. I think people are afraid of end users not being able to see every last bit of shadow detail or god forbid accusations of black crush.

It gets back to what is the color/look authority on the film: the IP or the release print? A case could be made for why both are right and wrong.

Problem is to get what Torsten wants, companies would have to scan release prints and most rps resolve to at most 2K but generally are lower. Hard to sell BDs and UHDs that way. Although that is want a lot of restorations do; scan the OCN and color correct to prints but that is going to by a shot-by-shot correction.

Scanning the IP is certainly much easier a job. So I wouldn't doubt a lot of masters are from IPs (see below). Scanning off of IPs do require some color adjustments to at minimum remove the orange color base that is normally filtered out in printing process.

Valeyard's pic:
[Image: z70dSy1.jpg]

I guess it is conceivable that a tech was doing their job monstrously wrong and in the case of DH2 and DH3 didn't remove the base correctly, leaving the film(s) with a warm look. But I'm just guessing here....

What are you thinking about doing Printer Lights, Zoidberg?

Yes it is certainly true about CRT'S, well more on the consumer market.
I played a few modern Blu-rays on and old CRT was still have floating around the house and they looked completely drained of red. The test was no way scientific though, it was being played out on an old Geforce 8800GT that had a composite out adapter.

If the IP was the source for a old master, the white balance would be off to compensate the introduction of the release prints characteristics. The old video masters would have certainly been white balanced, probably on the transfer process, knocking the original grade way off. This was probably done scene by scene also.
So by that methodology adding a blanket LUT to reintroduce the missing elements of the 35mm would require a scene by scene white balance adjustment on the source! Very time consuming. It's a full on remaster.

When doing your color correcting projects do you add the LUT' first and correct your source to that?
I gather a LUT that replicates the photo-chemical timing could not be touched, and you have to correct what you are feeding it.  

I wish I had more time and resources to experiment more. It would certainly nice to go 10bit

Interesting observation about the CRTs commodore.

I recently got an old CRT off of Craiglist for basically nothing and I was hoping to test with it also.

It's also important to note a lot of the red/magenta/pink HD masters I have seen are consistently from MGM. MGM did a massive scan of their entire library in the early 2000 and a lot of them are:

1. From IPs and were from a time of majority CRTs
2. Have a red look to them
3. Are extremely flat, un-dynamic, neutrally balanced and bland
4. Made to be shown on TV not necessarily Blu-rays.

That even goes for masters they have given to other companies. Just look at Arrow's UK BD of Buckaroo Banzai. Or look at the difference in Thief. I'm not saying the remaster is right, I don't know if it is but I am saying the "theatrical cut's" color are just off.

I think you are right about TV masters. They are/were white balance for every shot or had an major color bias reduced. Valeyard posted a video to that effect on the shoutbox I think. When I started projects I naively assumed that most would be one major color correction and done. It's not

Not sure I follow what you mean about using a LUT. Like do I apply a film stock LUT and then correct after that?

Side Note: Not that Beber needs to hear this but I know an anonymous source that has a US print of DH3 and basically Beber pics are almost dead on with anonymous' print. The major differences could be attributed to Beber's camera (blown highlights which Beber mentioned and his white balance- the print is a little yellower)

[CSchmidlapp]

Not sure I follow what you mean about using a LUT. Like do I apply a film stock LUT and then correct after that?

Yes.
I'm new to LUT's really, Ive had very little time to play.

I might be talking utter rubbish but I'm gathering the settings of the LUT can not be changed as it would ruin the desired effect. So would shifting the white balance gradually on the source, while monitoring with the LUT applied, be the best approach?

Sorry if this makes no sense lmao

[Beber]

Side Note: Not that Beber needs to hear this but I know an anonymous source that has a US print of DH3 and basically Beber pics are almost dead on with anonymous' print. The major differences could be attributed to Beber's camera (blown highlights which Beber mentioned and his white balance- the print is a little yellower)

Cool! See? I did tell you your first attempt at regrading it could use some golden/yellow hue.

[deleted user]

Not sure I follow what you mean about using a LUT. Like do I apply a film stock LUT and then correct after that?

Yes.
I'm new to LUT's really, Ive had very little time to play.

I might be talking utter rubbish but I'm gathering the settings of the LUT can not be changed as it would ruin the desired effect. So would shifting the white balance gradually on the source, while monitoring with the LUT applied, be the best approach?

Sorry if this makes no sense lmao

Film stock LUTs make very little sense here imo unless your only goal is to get something that might vaguely resemble the look of a certain film stock from the perspective of an amateur who just wants the "style". 

The reason is that a LUT can only be accurate when it is somehow calibrated to a certain kind of source. Apply the same LUT to two different sources with different basic grading or even different cameras and the results will be miles apart. 

For example, many LUTs will be calibrated to a certain standardized Log color space that many cameras offer as raw color space for which the results will be somewhat accurate/predictable. But whatever we have to work with usually (Blu Rays) are in no way raw footage that you can even calibrate to. They are already graded and edited material, so any LUT calibrated to a Log color space for instance would give you crazy and usually unusable results.

You can of course still apply some LUT and it may even look good at times, but the chance that it will be somehow theatrically accurate would have to be close to complete coincidence, as it isn't calibrated to any particular source, much less to any already (and differently) graded source.

To give a very simple (perhaps too simple?) example ... log color space has a rather low saturation. A LUT grading a log color space would increase/transform that low saturation to a normal level. But apply the same LUT to an already normally saturated source like a Blu Ray and you will get a completely oversaturated end result.

[zoidberg]

PDB, I call it printer lights because it's an attempt to at least mimic colour timing but obviously it has nothing to do with printer points as they are based on exposure stops. But basically photochemical colour timing was limited to RGB adjustments and overall density using a Hazeltine optical printer. No shadows/midtones/highlights and definitely no secondary corrections or 'power windows'. A push to blue for example, as TServo mentioned in the T2 thread causes the shadows to turn blue getting progressively paler as you go up to the bright highlights which will still be mostly white. This made me realise that being a subtractive process the negative is essentially acting as a filter or mask, attenuating the light shining through onto the positive film's emulsion (this is also why contrast builds up from o-meg to release print). So I'm basically experimenting with a script that layers an RGB adjusted video over the original video, using the original video's luma as a greyscale mask which decides how much adjustment occurs. Like I said, it's early days and it's just something I'm tinkering away with in the background.