Why is the interpositive said to lead to a magenta tint?

[deleted user]

I've read in several threads across the internet and also here that scans from the Interpositive tend to result in a magenta tint.

I don't quite understand why that would be.

Here's an image showing the 4 stages: Original negative, interpositive, internegative (I think) and final print

   

As you can see, the interpositive has an orange base. I am not quite sure why, Wikipedia only states this:

"The orange base provides special color characteristics that allow for more accurate color reproduction than if the IP had a clear base, as in print films."

No idea why that would be, but fair enough. Negatives also come on an orange base and I've never bothered to find out why.

In either way, how do you go from an orange color base and end up with a magenta tint without doing so with deliberation? Is this some kind of historical artifact of some kind of traditional processes or is it somehow related to how the IP is filtered before/after the scan? Does it have to do with fading?

Maybe someone who knows can tune in and enlighten me.

Little update. I found something that appears to be a raw 4k Scan (as jpg, so its not really raw, but you get what i mean) of an interpositive of apparently Dawn of the Dead:


Also very clearly an orange base. How do you go from this to a magenta tint?

[zoidberg]

I think older IP transfers have the red hue, newer transfers/scans less so. Don't Shout use a lot of IP scans? And of course Nolan won't scan the negative.

I think it's partly down to having less control over the colour adjustments during the transfer back then (RGB only, no 'power windows' etc) as well as being mastered for CRT displays (which has weak red phosphors). But mainly the colour correction. IPs have low contrast so their images are kind of a cypher for how they will look when printed to positive stock with it's increased colour/contrast, as well as the colours shifting relative to eachother.

[deleted user]

Yeah, Shout does seem to use those a lot. For example they used it for their Burbs release (which I like except for the low bitrate encode). That one seemed to have a few scenes with magenta cast, but not consistently so. Which of course may just mean that in the process of correcting their scan, they missed a few shots.

Hmmm. When you say their images are a cypher (sorry, not familiar with the figure of speech), do you mean that they don't properly predict the final look? If so, that'd be curious to me, as it seems that the color grading is done before the IP stage, not after. That would suggest to me that a print cast from the IP would have to look *very* close to how the answer print with the final color grading looked (which in itself is a little miraculous, heh). In other words, the IP seems like something of a very precise blueprint when it comes to the colors. Or at least in combination with the same stock used for the final print as was used for the answer print?

Less control over colour adjustments doesn't sound like the answer to me. I'm pretty certain that basic adjustments like hue/saturation/color balance were possible with professional equipment. It also wouldn't explain why the cast seems to be consistenly magenta instead of all over the place.

Color correction seems like it could be responsible and it may have been difficult back then to approximate an actual print's response digitally, but again, I would expect it to be more all over the place then.

I guess that the mastering for CRT does sound like the most plausible explanation. But then again, even those tended to have basic color adjustments iirc. Hmm.

Regarding my idea with fading ... I do think to recall that some old slides of mine tended to turn magenta over time as well. Maybe it's the most stable color layer? Do you by any chance know whether older movies tended to exhibit this more strongly than newer ones? This *might* just explain a little bit if they had, say, a fixed static process for the whole transfer. The idea is, what if they had a process that was designed in such a way to get the original intended colors off the IP based on some scientific principles or whatever, but then, as the IP would age, it would get a magenta cast, and maybe the process just didn't account for that? Though I find it a little hard to believe that they would miss such an obvious defect

Then again, the fading theory would also explain why this is more typical for IPs than for original negative scans, as those, being negatives, would not get a magenta cast as they fade. They would get the complimentary color cast, which would be green. Hmmmm.

[PDB]

Sorry I missed this conversation yesterday. Although I'm no expert (and please someone more knowledgeable correct me) I might be able to help.

Tom you might better be able to find you answer if you look in the world of 35mm still film, in particular the C-41 film process. In fact here is a good article to read:

http://www.brianpritchard.com/why_colour...orange.htm

and for film layers (still 35mm but you get the point):

http://istillshootfilm.org/post/46260540...c-film-101

The short, simplified story is that this type of film process (like C-41 still film) has 3 primary layers: magenta, cyan and yellow (red, green, blue sensitivity) of dyes to absorb each spectrum of color. But the dyes used in these layers are far from perfect in taking in just one narrow band of color. They drift and tend to absorb more then one color. So to help correct color absorption that there are masks (sometimes called couplers, filters, etc) embedded in the sandwich that makes color film. The masks help the layer to absorb just the color that they should. The two masks generally used are yellow and magenta, put those and you have orange. So its those masks that give the negative an orange look. IPs still are orange since they are still trying to protect the precise color of the timed film. But since you can't project a film with that orange color so theatrical prints are clear.

Sadly, the orange has nothing to do with fading. Fading is just the breakdown in the dyes in the layers with the magenta/red color being the longest lasting. LPPs and stock post 1982 just had more stable dyes that didn't break down as easily. Any Negs or IPs pre-1982 are subject to fading. Its the reason the industry created separation masters.

Like Zoid said, the IP being orange should have nothing to do with some SD and HD scans being magenta/pink/red. Any scan equipment should automatically remove that orange color from the start. Like Zoid said since the magenta scans come from the late 90s and early 2000s people assume it has to do with color grading on CRTs or since many prints of the time have a green look to them. We really aren't sure.

Side Note: Is that Dawn Of the Dead pic from the Italian restoration of the Argento cut? It was my understanding that restoration came from an IN and not an IP but I could of heard that wrong.

[deleted user]

Thanks for the explanation about the orange base and the links, I'll check those out later! Doesn't yet fully make sense in my head, but it probably will with more detail.

I didn't mean to suggest a connection between the orange base and fading. My thought process was: Base is orange, so that cant be the source of the magenta tint. But fading produces a magenta tint in a positive, hence it could be the source of said magenta tint.

Re side note: Here is where I found that pic: http://dawnofthedeadcollectors.blogspot....ation.html

[PDB]

I didn't mean to suggest a connection between the orange base and fading. My thought process was: Base is orange, so that cant be the source of the magenta tint. But fading produces a magenta tint in a positive, hence it could be the source of said magenta tint.

Re side note: Here is where I found that pic: http://dawnofthedeadcollectors.blogspot....ation.html

No worries I must of conflated the discussion. Here is another good explanation also:

"Dyes are not perfect. They have some unwanted absorptions. The cyan dye absorbs mostly red light, but it also absorbs some green light. To compensate for this unwanted absorption, a colored coupler is added to the red sensitive layer. This colored coupler is magenta to start with, but it loses the magenta color and forms cyan dye during development. This means that while the red layer has a negative dye image in cyan dye it has a positive dye image in the unused magenta colored coupler. This magenta image offsets the unwanted green light absorption of the cyan dye.

The magenta dye in the green sensitive layers absorbs mostly green light but also has some unwanted absorption of blue light. A yellow colored magenta dye forming coupler is added to the green layer.

The yellow dye in the blue sensitive layers absorbs mostly blue light, but also absorbs some UV energy. Since we can't see UV energy, we don't have to compensate for it.

With magenta colored and yellow colored coupler added to the film, the minimum density (Dmin) areas are orange colored. This color is in the emulsion layers, not in the plastic support."

https://www.photo.net/discuss/threads/co...ns.254896/

(Ok I guess DOTD was an IP scan, I thought I heard the IN was the only existing element left for the Argento cut)

[zoidberg]

Hmmm. When you say their images are a cypher (sorry, not familiar with the figure of speech), do you mean that they don't properly predict the final look? If so, that'd be curious to me, as it seems that the color grading is done before the IP stage, not after. That would suggest to me that a print cast from the IP would have to look *very* close to how the answer print with the final color grading looked (which in itself is a little miraculous, heh). In other words, the IP seems like something of a very precise blueprint when it comes to the colors. Or at least in combination with the same stock used for the final print as was used for the answer print?

You are right in the sense that the IP contains within it the correct printer lights values (printer points, themsleves being fractions of exposure stops).

You have to think of photochemical film as a complete system. The action is photographed onto low contrast negative, printed to high contrast positive and projected via a lamp of known quantity spectral output onto a screen. In the stages of production before mass production of release prints the feature has to be timed, every time a pass is made the original negative is run through the optical printer with the corresponding printer lights programmed. Once the timing is approved the IP is struck using the same timing. Now with the timing 'baked' into the IP the other duplication stages can proceed (INs, release prints) using faster 'one light' printing although densities are controlled/maintained by the labs (LAD tests, 'China Girls'). But the actual intended colour is only revealed by the release print. Even in the early days of DIs (before digital projection took over), a LUT was applied when generating the film-out materials to account for the shift when printing release prints. So a small adjustment to a the IP represents a greater adjustment on a positive/release print. Which is why I say the IP is almost like a cypher, or a code as to what the final result will be. My gut tells me that the green/blue shifts are stronger upon printing so the IP has extra red to compensate for this.

It's the extra stuff in the release print chemistry (the gamma, the highlight/shadow roll-offs per individual colour) which gives film it's look, or 'style' or whatever you want to call it.