New CN inversion method for PS

Adrian Gabor · Jan 17, 2018

I've recently developed (for personal use) an advanced action for PS that deals with the inversion of color negatives from linear .tiff scans.

This method is, as far as I am aware, the best yet, and much better than any methods I have tried yet, with results on par with pro lab examples from expensive scanners.

It needs a bit more work, but the results will be worth it. I would like to know if there are people interested in purchasing my action when it would be ready for sharing.

Features (currently working):

One click color mask removal and inversion
Automated initial CC
Simple initial brightness adjustment
Non-destructive layer based process
Adjustable "brightness" control
Adjustable color-neutral deep shadow recovery
Preserves both shadow and highlight detail, favouring highlight detail
Preserves film characteristics and saturation levels
No film profile needed (*neg. base sampling required)
No digital noise introduced in the process

The process itself is designed to mimic RA-4 printing, in a very simple and intuitive way. I'm currently working on a way to finish off the inversion with a basic paper tone curve as well as more accurate preliminary CC.

Adrian Gabor · Jan 20, 2018

I would like to invite photrio members to send me uninverted .tiff scans to do further testing, ideally scans in which the negative base is visible slightly outside of the frame. Scans that were troublesome to invert are also preferred.

I've changed some things and added a much better CC method.

RalphLambrecht · Jan 20, 2018

Adrian Gabor said:
I've recently developed (for personal use) an advanced action for PS that deals with the inversion of color negatives from linear .tiff scans.

This method is, as far as I am aware, the best yet, and much better than any methods I have tried yet, with results on par with pro lab examples from expensive scanners.

It needs a bit more work, but the results will be worth it. I would like to know if there are people interested in purchasing my action when it would be ready for sharing.

Features (currently working):

One click color mask removal and inversion

Automated initial CC

Simple initial brightness adjustment

Non-destructive layer based process

Adjustable "brightness" control

Adjustable color-neutral deep shadow recovery

Preserves both shadow and highlight detail, favouring highlight detail

Preserves film characteristics and saturation levels

No film profile needed (*neg. base sampling required)

No digital noise introduced in the process

The process itself is designed to mimic RA-4 printing, in a very simple and intuitive way. I'm currently working on a way to finish off the inversion with a basic paper tone curve as well as more accurate preliminary CC.

Sorry but, my answer is no; I follow general policy to not purchase any acton, brushes or plug-ins; the current cost of Adobe software is very high already and inverting color files is a very low priority for me; but wish you all the best.

chriscrawfordphoto · Jan 20, 2018

RalphLambrecht said:
Sorry but, my answer is no; I follow general policy to not purchase any acton, brushes or plug-ins; the current cost of Adobe software is very high already and inverting color files is a very low priority for me; but wish you all the best.

What a nasty, ignorant thing to say. The OP is not at fault for Adobe's high prices, and he is obviously putting a lot of work into developing his plugin. Why should he give it to you?

philosli · Jan 20, 2018

chriscrawfordphoto said:
What a nasty, ignorant thing to say. The OP is not at fault for Adobe's high prices, and he is obviously putting a lot of work into developing his plugin. Why should he give it to you?

RalphLambrecht simply, and politely, informed the OP he won't buy it. He never asked it for free.

mmerig · Jan 20, 2018

Your project sounds interesting and useful. Will the action work in Photoshop version 12.0 64 bit? (part of CS5) Any idea of the cost? Are you still looking for uninverted scans, and what is the best way to get them to you?

Thanks

Bob Carnie · Jan 20, 2018

This sounds interesting for those of us who scan large volumes of negatives, a one click action to make web worthy files would be interesting. I do a lot of PS colour correction and file management and it is really a PIA when working on larger groups for presentation purpose.
I have never been a fan of plugins, but rather work in PS for final printing.
Could the OP describe how his plugin would work ie is it a downloadable item we can buy from him.

Bob Carnie · Jan 20, 2018

I could send you some of my files, but I think they would make you bat shit crazy as the colour film is solarized and it would be funny to see how the application try's to figure this out.

Patrick Robert James · Jan 20, 2018

Your project sounds interesting, but more geared toward a point and click mentality. Like Bob mentions, if it is automatic for a lot of images then it may be worth it for that application. Personally I use a LAB based editing method that produces better color than any other method I have ever seen. Just can't automate that. I only bother really working on the ones that are worth it which aren't that many really.

I guess there are quite a lot of people who use other people's actions in Photoshop and Lightroom. Frankly if you can make it work in Lightroom you will do much better I think. If I were you I would concentrate on that. Just my opinion.

Adrian Gabor · Jan 20, 2018

Bob Carnie said:
I could send you some of my files, but I think they would make you bat shit crazy as the colour film is solarized and it would be funny to see how the application try's to figure this out.

The harder to work with, the better, as I'm also trying to gather some data to fine tune the process. The only absolute requirement is to not have clipped image data in any of the channels of the file. 16 bit is preferred but 8 bit works too.

The action will use only basic math operations, layer, blending modes and some clever yet simple ways of using what PS can already to to achieve this goal.

As for the cost, to clear up some things, if I fail to deliver on my promise of this being the best action/plugin/method yet, then I will make it free and share it with everybody. It already works internally better than anything I tried, including a very expensive option that I'm sure is known on this forum.

Another thing that I want to do, is provide some concise documentation about how this works, what to do should should things go wrong, why things go wrong, etc.

To send me files, please use any method you want, including Google Drive, wetransfer, and PM me the link. I can then process them with my action and post the result here or as a reply to the message.

Adrian Gabor · Jan 20, 2018

Patrick Robert James said:
Your project sounds interesting, but more geared toward a point and click mentality. Like Bob mentions, if it is automatic for a lot of images then it may be worth it for that application. Personally I use a LAB based editing method that produces better color than any other method I have ever seen. Just can't automate that. I only bother really working on the ones that are worth it which aren't that many really.

I guess there are quite a lot of people who use other people's actions in Photoshop and Lightroom. Frankly if you can make it work in Lightroom you will do much better I think. If I were you I would concentrate on that. Just my opinion.

I'd love to compare my method to yours, with a blind test, in which I only see the negative.

Eric Rose · Jan 20, 2018

I'd be interested as long as the price is reasonable.

RZied · Jan 20, 2018

How about raw files? I am currently scanning my films using my digital camera and made a rig of my own.

Raoul

Lachlan Young · Jan 20, 2018

If it's an automation of the fairly long-established but often poorly publicised method I suspect it is, it works very well and far faster than it takes to type/ explain (critical component is the sampling & dividing out of the orange mask before inversion). However, it seems dependent (in my experience) on 16-bit files with no clipping from scanner shortcomings (Bob and others using high-end CCD, Flextight or PMT scanners should be fine, DSLRs probably too, if they can do 14-bit RAW) and really demands that you work in ProphotoRGB or LAB spaces. I think it could be automated, but that might require a level of machine learning beyond what PS can provide at the current time, unless every frame for every format is set up with a bit of rebate in exactly the same place...

calebarchie · Jan 20, 2018

I just love how everyone in this thread and in general treat scanning as some kind of black magic or art. All the R&D that revolve around these methods has been published in books and online resources freely up to the early 2000s most notably aim-dtp.net before it went down. Just look and you shall find, I suppose if people want to hitch the dslr-scanning wave for $$$ then that is another matter - I agree with Ralph here.

Ted Baker · Jan 20, 2018

calebarchie said:
some kind of black magic or art.

Yes agree, its not magic or an art. However the techniques and physics are poorly documented in comparison with what is available in the pure digital space. If you would like to post any resources I for one would be interested?

BTW I have spent the last few months, as time permits implementing various approaches myself, so it's good to others are interested too. Good luck Adrian!

Adrian Bacon · Jan 20, 2018

Ted Baker said:
Yes agree, its not magic or an art. However the techniques and physics are poorly documented in comparison with what is available in the pure digital space. If you would like to post any resources I for one would be interested?

BTW I have spent the last few months, as time permits implementing various approaches myself, so it's good to others are interested too. Good luck Adrian!

Hey! another Adrian! doing something similar! what are the odds? In all honesty, I'd be much more interested in seeing something like this implemented in something that you can do in Adobe Lightroom. The tool I've been working on (it's called Simple Image Tools, though it's anything but simple) outputs Adobe DNG files that are 32 bit floating point for each channel and since ProPhoto is the built in native color space of Lightroom, also conforms the film to that color space. It's interesting that you're trying to mimic RA4 printing, but in all honesty, I think you'll find that as you try this out on more emulsions, there's a number of variables that change from emulsion to emulsion that you need to be able to control for, like the masking offset isn't the same, the contrast of each channel between different emulsions also isn't the same, etc. It'll pretty quickly turn into providing a way to control those variables more than anything else. That being said, how many unique color negative emulsions are being manufactured today? Compared to BW? Not a lot.

calebarchie · Jan 21, 2018

Ted Baker said:
Yes agree, its not magic or an art. However the techniques and physics are poorly documented in comparison with what is available in the pure digital space. If you would like to post any resources I for one would be interested?

BTW I have spent the last few months, as time permits implementing various approaches myself, so it's good to others are interested too. Good luck Adrian!

As have many with this renewed interested in film. At the turn of the century with the internet still in infancy this information was primarily published in books, articles and academic text etc. That said, I don't really have any links RN unless you want to trawl the wayback machine and old yahoo boards. Look for articles by leading colour theorists and experts from then (Autiokari, Poynton, Hoffmann, Fraser, Cox, Faust, Worthey) otherwise a good primer text ISBN: 978-0-470-02425-6.

C

calebarchie · Jan 21, 2018

I would like to invite the OP and others to simply just post your findings, chances are they all work in a very similar manner (nothing new here!)

Adrian Gabor · Jan 21, 2018

Lachlan Young said:
If it's an automation of the fairly long-established but often poorly publicised method I suspect it is, it works very well and far faster than it takes to type/ explain (critical component is the sampling & dividing out of the orange mask before inversion). However, it seems dependent (in my experience) on 16-bit files with no clipping from scanner shortcomings (Bob and others using high-end CCD, Flextight or PMT scanners should be fine, DSLRs probably too, if they can do 14-bit RAW) and really demands that you work in ProphotoRGB or LAB spaces. I think it could be automated, but that might require a level of machine learning beyond what PS can provide at the current time, unless every frame for every format is set up with a bit of rebate in exactly the same place...

I like the discussion that this thread has generated, very positive (and some negative) things that provide a lot of insight to me.

You are very close in your assessment of how this works, although I suppose the principles are always the same, there's no reinventing the wheel. 16bit files are necessary, but I have done some testing with 8 bit files for this exact purpose, and the results are still good.
ProphotoRGB is not required either. The rebate should ideally be part of the final crop, but the sampling of the color base is manual and only required once, it can be done from a different shot as long as it is the same film and it was scanned with the same settings, including gain.

Adrian and Ted's solutions sound way more advanced than what I'm doing, I'm by no means a very technical person, and I try to find simple solutions to complex problems, and often times that proves possible.

If my colors end up looking like what I need for my own photography, then I consider this to be a success, not very true to scientific principles, I know.
Others have tried scientific approaches that, although (probably) correct in their implementation and logic, fail to deliver in a real world application.

Adrian Gabor · Jan 21, 2018

Here are two inversions done with one click of the action, after sampling the color base of each.
One is Portra400 @ 200 with very dense regions and strong highlights, scanned on my modest Reflecta7200,
the other is a Hasselblad X1 scan of Ektar 100, again with quite a strong range of tones.

bestbefore1978 · Jan 21, 2018

Hi Adrian,
Your method sounds really interesting and I would purchase your PS action if it achieves what it says, My current method involves a combination of the Colorperfect plugin for inversion and a combination of curves adjustment which gives good results, but it is difficult to arrive at consistent results and colour balance frame to frame.

I scan 16-bit linear files on a Flextight Precision III, I can send you some files for testing later today if you like

Adrian Gabor · Jan 21, 2018

bestbefore1978 said:
Hi Adrian,
Your method sounds really interesting and I would purchase your PS action if it achieves what it says, My current method involves a combination of the Colorperfect plugin for inversion and a combination of curves adjustment which gives good results, but it is difficult to arrive at consistent results and colour balance frame to frame.

I scan 16-bit linear files on a Flextight Precision III, I can send you some files for testing later today if you like

Please do!

RZied · Jan 21, 2018

Adrian Gabor said:
View attachment 193834 View attachment 193835

Here are two inversions done with one click of the action, after sampling the color base of each.
One is Portra400 @ 200 with very dense regions and strong highlights, scanned on my modest Reflecta7200,
the other is a Hasselblad X1 scan of Ektar 100, again with quite a strong range of tones.

Hello

Adrian Gabor said:
View attachment 193834 View attachment 193835

Here are two inversions done with one click of the action, after sampling the color base of each.
One is Portra400 @ 200 with very dense regions and strong highlights, scanned on my modest Reflecta7200,
the other is a Hasselblad X1 scan of Ektar 100, again with quite a strong range of tones.

Hello Adrian,
You can try this RAW file of a Fuji Pro 400h 120mm film scanned using Nikon D810. Please let me know. Here's the link: https://d810scan.s3.amazonaws.com/170624-Davis-California-161756.NEF
I am also interested to buy this PS action.

Regards,
Raoul

Adrian Gabor · Jan 21, 2018

RZied said:
Hello

Hello Adrian,
You can try this RAW file of a Fuji Pro 400h 120mm film scanned using Nikon D810. Please let me know. Here's the link: https://d810scan.s3.amazonaws.com/170624-Davis-California-161756.NEF
I am also interested to buy this PS action.

Regards,
Raoul

Hello Raoul,

The results from this are worse than the previous DSLR scan test I did, but I'm not sure why.
I think there might be color information clipped in the raw file. 400H has quite a strong reddish base, did you use any gels over your flash to compensate for this? If not, that might be the reason why. You don't have to completely negate the base, but just enough blue or cyan to get the channels a little balanced.
I first imported in to LR to get a 16bit .tiff file, I assume the dslr was already set to 14bit raw uncompressed.

New CN inversion method for PS

Privacy & Transparency

Privacy & Transparency