Removing Orange Mask from Colour Negative Scans

Forum covering all aspects of small gauge cinematography! This is the main discussion forum.

Moderator: Andreas Wideroe

carllooper
Senior member
Posts: 1206
Joined: Wed Nov 03, 2010 1:00 am
Real name: Carl Looper

Re: Removing Orange Mask from Colour Negative Scans

Post by carllooper » Sat Jun 07, 2014 7:23 am

The one on the left is obtained without removing the orange mask, so it makes it impossible for the technique used to get anything much better than what is presented.

However the colour information is still in there. It's just locked up in a particular way. By removing the orange mask (as prescribed), one is then able to use normal CC controls to properly pop the colours out - in this case, from their imprisonment in purple. While the orange mask removal process is a little involved it can be automated, either by setting up a reusable comp in After Effects or an Action in Photoshop.

Image
Carl Looper
http://artistfilmworkshop.org/

PyrodsTechnology
Posts: 64
Joined: Sat Dec 11, 2010 10:45 pm
Real name: Roberto Pirodda
Location: Italy

Re: Removing Orange Mask from Colour Negative Scans

Post by PyrodsTechnology » Sun Jun 08, 2014 12:47 pm

well Carl, by my taste the left image has the correct look. The right image is over saturated and contrasted. May be the best should be a mix between the two

carllooper
Senior member
Posts: 1206
Joined: Wed Nov 03, 2010 1:00 am
Real name: Carl Looper

Re: Removing Orange Mask from Colour Negative Scans

Post by carllooper » Sun Jun 08, 2014 2:39 pm

Yes I put some extra warmth in it - just my personal taste really.
However the point is not whether it's to your taste or not, but whether you can get the image to a place where your taste can actually be implemented.
Until you've removed the orange mask you are stuck with compromised colour choices.

C
Carl Looper
http://artistfilmworkshop.org/

JeremyC
Posts: 154
Joined: Fri Jun 01, 2012 2:51 pm
Real name: Jeremy Cavanagh

Re: Removing Orange Mask from Colour Negative Scans

Post by JeremyC » Sun Jun 08, 2014 3:44 pm

Carl,

I have a question on your first post as to why you want to remove the image information when removing the orange mask.I can well understand the need to remove the mask and the techniques you have presented are very helpful but isn't this information useful for the final detail and clarity in the scanned image even when such detail is contained in the other layers? Couldn't your techniques incorporate this detail as a Y signal while completely completely getting rid of the colour in the mask so making it available if any of the detail has been 'missed' out of a film's layers?

What you have written has really intrigued me so I'm hoping to quiz one of my engineering colleagues about it who spent over 20 years working on film scanning and telecine machines to see whether this was addressed at a professional level. Reading around, in the meantime, I found this site as a very interesting tutorial on colour masking in negative film: http://www.brianpritchard.com/why_colou ... orange.htm. I am sure there must be other sources of information.

carllooper
Senior member
Posts: 1206
Joined: Wed Nov 03, 2010 1:00 am
Real name: Carl Looper

Re: Removing Orange Mask from Colour Negative Scans

Post by carllooper » Sun Jun 08, 2014 4:21 pm

Hi Jeremy,

that link you posted is spot on. Regarding whether information is being lost is an interesting question. The answer is no. Although one speaks of "removing the mask" it remains the case that given a corrected image (where the mask has been removed) you can actually run the mask removal procedure in reverse and the mask will reappear. From where could it come if it was removed?

So what's happening is more like a rearrangement of the information rather than the removal of anything: rearranging the information for how digital tools otherwise expect the image.

One can appreciate the mask in terms of a film to film pipeline. When printing film one doesn't have to put some special purpose filter in place in order to "remove the mask". In film printing the information in the negative is "programming" two systems at the same time: the printer lights which are in RGB space, and the dyes in the print stock which are in CMY space. The printing is a handshake being done in both RGB space by means of light and in CMY space by means of dyes, where the orange mask acts as a zero sum per 'pixel' switch between the two colour domains.

Now if you remove one part of this system, ie. the print film, you no longer have this zero sum handshake, and so you get this side effect called the "orange mask". The so called "orange mask removal" process is not really removing anything - it's recreating the missing half of the handshake that would otherwise occur in the film domain.

Carl
Last edited by carllooper on Sun Jun 08, 2014 5:06 pm, edited 1 time in total.
Carl Looper
http://artistfilmworkshop.org/

JeremyC
Posts: 154
Joined: Fri Jun 01, 2012 2:51 pm
Real name: Jeremy Cavanagh

Re: Removing Orange Mask from Colour Negative Scans

Post by JeremyC » Sun Jun 08, 2014 4:57 pm

carllooper wrote:. Regarding whether information is being lost is an interesting question. The answer is no. Although one speaks of "removing the mask" it remains the case that given a corrected image (where the mask has been removed) you can actually run the mask removal procedure in reverse and the mask will reappear. From where could it come if it was removed?
I guess it would be this: if you have an image of a face on the neg then the outline or details of that face would be on the orange mask as well, if I understand you correctly, then we remove the orange which includes its masking which is an additive component on the image but we might want the face outline back because of its details. Its a difficult thing because you have to make a choice of what is detail and what is additive i.e. just the mask and there are perhaps a couple of technical ways to decide what is a masking outcome and what is detail.

carllooper
Senior member
Posts: 1206
Joined: Wed Nov 03, 2010 1:00 am
Real name: Carl Looper

Re: Removing Orange Mask from Colour Negative Scans

Post by carllooper » Sun Jun 08, 2014 5:37 pm

JeremyC wrote:I guess it would be this: if you have an image of a face on the neg then the outline or details of that face would be on the orange mask as well, if I understand you correctly, then we remove the orange which includes its masking which is an additive component on the image but we might want the face outline back because of its details. Its a difficult thing because you have to make a choice of what is detail and what is additive i.e. just the mask and there are perhaps a couple of technical ways to decide what is a masking outcome and what is detail.
The mask is just the inverse of the dye image. And vice versa. They interlock like pieces in a jigsaw puzzle. However they contain the same information, just inversely (and in different colours). Both the dye image and the mask image operate on the light passing through the film, but in a way that is specific to the way the print stock will record that light. The issue is that if you scan the colour negative in terms of only RGB light (as we do) we are no longer mediating the information in terms of the CMY contribution the print stock also makes in that process. We have kidnapped the image from a process that would have otherwise restored the image had we not kidnapped it. So the kidnapped information is "out of order". So it's just a matter of putting it back into order, in a way that would be the digital equivalent of the way the film print would have otherwise put it back in order. We speak of "removing the mask" but we're not removing any image - we're only removing the vestige of an algorithm (an analog one) that would have removed itself anyway, had a print been made.

By way of example: think of a number - any number. Think of your number as the image. Now lets propose an algorithm to process your number:

Step1: Add 4.
Step2: Subtract 4.

The result will be your original number. But if we remove from the process, step2, then we'll have both your original number plus an artifact of an incompleted algorithm. However, if we know the algorithm we can complete Step 2. We can subtract the number 4 - and this number is not any information in the image - it is information in the algorithm. By completing the algorithm (subtracting 4) we restore the original number (the image).

So there isn't any details being lost. The process is just making sure that colour information ends up in the right channels, ready for digital CC tools to exploit.

C
Carl Looper
http://artistfilmworkshop.org/

carllooper
Senior member
Posts: 1206
Joined: Wed Nov 03, 2010 1:00 am
Real name: Carl Looper

Re: Removing Orange Mask from Colour Negative Scans

Post by carllooper » Sun Jun 08, 2014 6:55 pm

PyrodsTechnology wrote:well Carl, by my taste the left image has the correct look. The right image is over saturated and contrasted. May be the best should be a mix between the two
I updated the picture to which you referred. So it looks a bit better now. In particular, note the pink light on the back wall and the spot of it down the left in the middle. You can see the pink reflect itself onto the adjacent wall. I didn't put any of that in there. It was already in there, locked up in the image. And quite inaccessible if the orange mask hadn't been removed. But once the mask is removed all you need to use are colour balance controls to flush out all the colours. There's still a bit of work (and art) required in the colour balancing but at least it becomes achievable rather than impossible. For example, I guessed from the angle of the light that it must be sunrise or sunset, so the wall must be warmer than it looked in the original. By assuming such I then went looking for it and found it. I found a distinction in colour between the light and shadow happening, so tweaked the colour balance a tad and it just popped right out. Which was such a delight. It meant I was close to a balance that would unlock the colours across the entire image. And sure enough, a happy by-product (as will happen) is that the flesh tones improved considerably. No longer a ghastly purple!

C
Carl Looper
http://artistfilmworkshop.org/

PyrodsTechnology
Posts: 64
Joined: Sat Dec 11, 2010 10:45 pm
Real name: Roberto Pirodda
Location: Italy

Re: Removing Orange Mask from Colour Negative Scans

Post by PyrodsTechnology » Mon Jun 09, 2014 12:27 pm

Carl, since you have updated/changed those pictures, my previous comment becomes useless

carllooper
Senior member
Posts: 1206
Joined: Wed Nov 03, 2010 1:00 am
Real name: Carl Looper

Re: Removing Orange Mask from Colour Negative Scans

Post by carllooper » Mon Jun 09, 2014 2:49 pm

PyrodsTechnology wrote:Carl, since you have updated/changed those pictures, my previous comment becomes useless
By no means useless, as it inspired me to rework the image to better support the argument about removing the orange mask.

But yes, without the image to which you referred, your comments look incorrect. But you were exactly right. The one I did before looked terrible in exactly the way you described. I must have created it while my screen was in a desaturated mode. Anyway I intended to post the updated image in a new post but when preparing the image for the post I inadvertently overwrote the old one on the server, and therefore the image in the previous post. But I no longer had the original one that was there - so thought it best to just clarify the situation in a subsequent post. And more so here.

cheers
C
Carl Looper
http://artistfilmworkshop.org/

milesandjules
Posts: 252
Joined: Wed Mar 09, 2005 11:22 am
Location: brisbane australia
Contact:

Re: Removing Orange Mask from Colour Negative Scans

Post by milesandjules » Tue Jun 10, 2014 1:22 am

I liked the one you had originally, cause in that the soldiers pants looked blacker now I think they look to blue like the original.

carllooper
Senior member
Posts: 1206
Joined: Wed Nov 03, 2010 1:00 am
Real name: Carl Looper

Re: Removing Orange Mask from Colour Negative Scans

Post by carllooper » Tue Jun 10, 2014 1:46 am

milesandjules wrote:I liked the one you had originally, cause in that the soldiers pants looked blacker now I think they look to blue like the original.
Ha ha.

In terms of orange mask removal, the important point is not how it looks, as that is a creative decision that each and every artist/critic has the freedom to make.

Removal of the orange mask does not create any particular look. Rather it's a technical preparation of the image in such a way that gives the artist (but not the critic) a broader range of colour choices. It is not in itself a decision. It is that which enables decisions.

One can say the same thing about Pianos. Pianos don't create a particular note. They enable a full range of notes. It is up to the piano player to select which notes strike a chord.

C
Carl Looper
http://artistfilmworkshop.org/

milesandjules
Posts: 252
Joined: Wed Mar 09, 2005 11:22 am
Location: brisbane australia
Contact:

Re: Removing Orange Mask from Colour Negative Scans

Post by milesandjules » Tue Jun 10, 2014 2:02 am

He He yep :P ….Hey looks like this will only work in full version of photoshop….good old aftere effects doesn't support cmyk :cry: …..i know i have ps full on an old computer will dig that out and try.

carllooper
Senior member
Posts: 1206
Joined: Wed Nov 03, 2010 1:00 am
Real name: Carl Looper

Re: Removing Orange Mask from Colour Negative Scans

Post by carllooper » Tue Jun 10, 2014 2:30 am

milesandjules wrote:He He yep :P ….Hey looks like this will only work in full version of photoshop….good old aftere effects doesn't support cmyk :cry: …..i know i have ps full on an old computer will dig that out and try.
Yes, I came across the same problem in After Effects. However you can set up effects to do the CMY separation.

QUICK NOTES

I used: Effect > Channel > Shift Channels

So for the Cyan separation composition I used the following 'shift channel' on the source:

Take Alpha from: Alpha
Take Red from: Red
Take Green from: Red
Take Blue from: Red

And similarly for the magenta separation, I used Green for all the channels, and for the Yellow separation, I used Blue for all the channels.

So one has three compositions, Cyan, Magenta, and Yellow, all of which are versions of the original scan, in black and white, representing the CMY components of the film.

From there it's the same process as the Photoshop method. Derive Magenta Repaired and Yellow Repaired. [ Note that this 'repair' process actually emulates the subtractive process that print film otherwise performs when operating on the RGB light from the negative. The mask is actually being subtracted from light rather than from any image. However it's only one part of the emulation. The rest of the subtractive print emulation is done in the conversion to RGB]

For conversion back to RGB (where normal CC occurs) one puts the following comps into a new comp called RGB:

Cyan
Magenta Repaired
Yellow Repaired

The effect to apply to each is to invert them, and to have the Blend mode on the first two set as 'Screen'.

Carl
Carl Looper
http://artistfilmworkshop.org/

milesandjules
Posts: 252
Joined: Wed Mar 09, 2005 11:22 am
Location: brisbane australia
Contact:

Re: Removing Orange Mask from Colour Negative Scans

Post by milesandjules » Tue Jun 10, 2014 4:59 am

Hi Carl
Genius…going to try that tonight . :P

Post Reply