LUTs for negative conversion

Started Jan 19, 2018 | Discussions
Ferguson
Ferguson Senior Member • Posts: 1,851
LUTs for negative conversion
1

I'm digitizing years of negatives by photographing them and then variously inverting them to positives.

The photograph aspect is nice - fast, easy; not as good perhaps as a dedicated negative scanner, but it makes up for a lot in terms of speed and file size.

I can then invert and edit in lightroom and things look somewhere between mediocre and awful.   

I can edit in photoshop, invert there and adjust (even bring back to lightroom and edit further) and it looks decent.  Sometimes.  Sometimes not so good.

So... with more and more LUT support present, e.g. in photoshop, is it possible to (a) invert in a LUT, and (b) remove the color cast in a LUT?

In particular (I'm a programmer at heart), could I write a program to generate such a LUT, and then tweak the algorithm as needed for color cast?

In particular, is the math for the LUT something deterministic?  If I had a photo of a blank frame (for color cast) and a negative, is there a simple formula for the inversion and removal of the color cast?

Or is this like asking "is there a formula for happiness"?

And/or, has anyone done this and are such LUT's available?   I found some LUT's for various effects for video, but did not see much about LUT's for still negative conversions.

Linwood

-- hide signature --

Comments welcomed on photos: http://www.captivephotons.com

 Ferguson's gear list:Ferguson's gear list
Sony a7R IV Sony a9 II Sony FE 24-70mm F2.8 GM Sony FE 70-200mm F2.8 GM OSS Sony FE 85mm F1.4 GM +2 more
Sympa Senior Member • Posts: 2,573
Re: LUTs for negative conversion

http://forum.luminous-landscape.com/index.php?topic=87681.0 seems a nice discussion of this subject.

 Sympa's gear list:Sympa's gear list
Sony RX100 Sony E 30mm F3.5 Macro Sony E 50mm F1.8 OSS Sony E 55-210mm F4.5-6.3 OSS Sony Vario-Tessar T* E 16-70mm F4 ZA OSS +4 more
Ferguson
OP Ferguson Senior Member • Posts: 1,851
Re: LUTs for negative conversion

Sympa wrote:

http://forum.luminous-landscape.com/index.php?topic=87681.0 seems a nice discussion of this subject.

I actually read that, and it started with the same question, but then seemed to go off more or less in a more classic invert-and-set-white/black-points direction, with additional discussion about Silverfast and Vuescan and ROC.

Underlying all of these (well, the manual ones) seems to be taking out the color cast by working separately on R, G and B either before or after inversion.

It would seem, at least in principle, for a given negative type, one could use a LUT to do all that at once.   And conceptually allow for different instances (age, etc.) of even the same stock adjust by shooting a blank frame and comparing?

Basically, with Photoshop now supporting LUT's, wondering if there's a better way than what people have been doing for some years.

Linwood

-- hide signature --

Comments welcomed on photos: http://www.captivephotons.com

 Ferguson's gear list:Ferguson's gear list
Sony a7R IV Sony a9 II Sony FE 24-70mm F2.8 GM Sony FE 70-200mm F2.8 GM OSS Sony FE 85mm F1.4 GM +2 more
Sympa Senior Member • Posts: 2,573
Re: LUTs for negative conversion
1

I think the essential bit is to use a linear representation of RGB with indeed no cross-processing applied. Then set white and black points.

If the representation is not linear, the curves for the colors will be different and a whole lot of curvewrangling is needed to compensate afterwards.

Only once the white and black points have been set the gamma is applied. I think this could be an essential difference to 'scan and process' which applies the tone curve immediately.

 Sympa's gear list:Sympa's gear list
Sony RX100 Sony E 30mm F3.5 Macro Sony E 50mm F1.8 OSS Sony E 55-210mm F4.5-6.3 OSS Sony Vario-Tessar T* E 16-70mm F4 ZA OSS +4 more
Wayne Larmon Forum Pro • Posts: 10,666
ColorPerfect Photoshop plugin

Ferguson wrote:

Sympa wrote:

http://forum.luminous-landscape.com/index.php?topic=87681.0 seems a nice discussion of this subject.

I actually read that, and it started with the same question, but then seemed to go off more or less in a more classic invert-and-set-white/black-points direction, with additional discussion about Silverfast and Vuescan and ROC.

A lot of that LuLa thread was me figuring out a workable way of converting negatives.  My final post in that thread lead to a page I derived from participating in the above thread.

My page eventually describes the method I found worked  best:  using CF-Systems ColorPerfect Photoshop plugin.  This plugin sidesteps a lot of the contortions you end up doing to null out the negative's orange mask.  ColorPerfect includes a program that makes a more workable linear TIFF file than any of the methods that were described in the LuLa thread.  IMO.

The text on my web page is ambivalent about the CP plugin.  I eventually found that it worked faster and better than the other manual methods I described, so I purchased a CP license.   By that time I finished up converting all the negatives I wanted to convert and never got back to updating my web page.

Wayne

Ferguson
OP Ferguson Senior Member • Posts: 1,851
Re: ColorPerfect Photoshop plugin

Wayne Larmon wrote:

The text on my web page is ambivalent about the CP plugin. I eventually found that it worked faster and better than the other manual methods I described, so I purchased a CP license. By that time I finished up converting all the negatives I wanted to convert and never got back to updating my web page.

Wayne

Thank you. Yes, I read that and was thinking of downloading a trial but then spent more time experimenting with ACR->PS.  I'm tied up today and tomorrow, but then hope to get back to this Monday and may try CP.

I gather no one yet has a packaged approach with LUT's that do everything.  Maybe it's just not possible.  But if LUT's move from PS to LR (as so many things have), it would sure be nice.  I would think the linear aspect becomes relatively moot if you can just do a lookup for any value, as opposed to dragging curves?

Linwood

-- hide signature --

Comments welcomed on photos: http://www.captivephotons.com

 Ferguson's gear list:Ferguson's gear list
Sony a7R IV Sony a9 II Sony FE 24-70mm F2.8 GM Sony FE 70-200mm F2.8 GM OSS Sony FE 85mm F1.4 GM +2 more
Wayne Larmon Forum Pro • Posts: 10,666
Re: ColorPerfect Photoshop plugin

Ferguson wrote:

Wayne Larmon wrote:

The text on my web page is ambivalent about the CP plugin. I eventually found that it worked faster and better than the other manual methods I described, so I purchased a CP license. By that time I finished up converting all the negatives I wanted to convert and never got back to updating my web page.

Wayne

Thank you. Yes, I read that and was thinking of downloading a trial but then spent more time experimenting with ACR->PS. I'm tied up today and tomorrow, but then hope to get back to this Monday and may try CP.

I gather no one yet has a packaged approach with LUT's that do everything. Maybe it's just not possible. But if LUT's move from PS to LR (as so many things have), it would sure be nice. I would think the linear aspect becomes relatively moot if you can just do a lookup for any value, as opposed to dragging curves?

CP is a packaged approach that does a lot of the conversion.  Before using CP, my experience was that every negative tended to be different from other negatives and that any kind of static transform was doomed to fail. Which led to a lot of manual curve twisting. CP reduced the amount of manual curve twisting. For me.

My page on CP has a downloadable RAW file of the negative I worked on. You can download the RAW file and compare your own methods to the methods I described.

Good luck.

Wayne

Ferguson
OP Ferguson Senior Member • Posts: 1,851
Re: ColorPerfect Photoshop plugin

Wayne Larmon wrote:

Ferguson wrote:

Wayne Larmon wrote:

The text on my web page is ambivalent about the CP plugin. I eventually found that it worked faster and better than the other manual methods I described, so I purchased a CP license. By that time I finished up converting all the negatives I wanted to convert and never got back to updating my web page.

Wayne

Thank you. Yes, I read that and was thinking of downloading a trial but then spent more time experimenting with ACR->PS. I'm tied up today and tomorrow, but then hope to get back to this Monday and may try CP.

I gather no one yet has a packaged approach with LUT's that do everything. Maybe it's just not possible. But if LUT's move from PS to LR (as so many things have), it would sure be nice. I would think the linear aspect becomes relatively moot if you can just do a lookup for any value, as opposed to dragging curves?

CP is a packaged approach that does a lot of the conversion. Before using CP, my experience was that every negative tended to be different from other negatives and that any kind of static transform was doomed to fail. Which led to a lot of manual curve twisting. CP reduced the amount of manual curve twisting. For me.

My page on CP has a downloadable RAW file of the negative I worked on. You can download the RAW file and compare your own methods to the methods I described.

Thanks.  Will.

Linwood

-- hide signature --

Comments welcomed on photos: http://www.captivephotons.com

 Ferguson's gear list:Ferguson's gear list
Sony a7R IV Sony a9 II Sony FE 24-70mm F2.8 GM Sony FE 70-200mm F2.8 GM OSS Sony FE 85mm F1.4 GM +2 more
Ferguson
OP Ferguson Senior Member • Posts: 1,851
Re: ColorPerfect Photoshop plugin

Ferguson wrote:

Wayne Larmon wrote:

Ferguson wrote:

Wayne Larmon wrote:

The text on my web page is ambivalent about the CP plugin. I eventually found that it worked faster and better than the other manual methods I described, so I purchased a CP license. By that time I finished up converting all the negatives I wanted to convert and never got back to updating my web page.

Wayne

Thank you. Yes, I read that and was thinking of downloading a trial but then spent more time experimenting with ACR->PS. I'm tied up today and tomorrow, but then hope to get back to this Monday and may try CP.

I gather no one yet has a packaged approach with LUT's that do everything. Maybe it's just not possible. But if LUT's move from PS to LR (as so many things have), it would sure be nice. I would think the linear aspect becomes relatively moot if you can just do a lookup for any value, as opposed to dragging curves?

CP is a packaged approach that does a lot of the conversion. Before using CP, my experience was that every negative tended to be different from other negatives and that any kind of static transform was doomed to fail. Which led to a lot of manual curve twisting. CP reduced the amount of manual curve twisting. For me.

My page on CP has a downloadable RAW file of the negative I worked on. You can download the RAW file and compare your own methods to the methods I described.

Thanks. Will.

I finally got around to this, and I found the ColorPerfect program somewhere between quite good, and absolutely awful.

It's like something from 1995 or so -- tiny screen, can't resize, awful GUI interface that is arcane.

But the colors sometimes could be quite good; I liked the "10" ring for example, but the size of the GUI on my screen is about 15% of the screen, and the images are just tiny thumbnails.  It's written for 800x600 screens or something similar I think.  It's so small I just can't see well enough what it has done to judge, so I OK out, then have to go start all over if I want to tweak it again.

And their web site is a twisting, rambling, awful stream of consciousness site, with "old" and "new" information and you are supposed to use both, and figure out what applies.  A lot like the GUI -- it's all in there somewhere, but organized so badly it is really tough to find what you want to know.

I think at times I could get better color with it, but especially due to the size of the preview, it is just too hard to use.  So going back to Photoshop (but I'll now experiment with the linear tiff, which may be a better starting point).

Thank you for all your information on the web site, it was a lot easier to actually get an image processed from your concise information, than the mess on their site.  FWIW the product may be MUCH better than I'm giving it credit for, but... it's only as good as one can (with reasonable effort) get it to produce.

Linwood

-- hide signature --

Comments welcomed on photos: http://www.captivephotons.com

 Ferguson's gear list:Ferguson's gear list
Sony a7R IV Sony a9 II Sony FE 24-70mm F2.8 GM Sony FE 70-200mm F2.8 GM OSS Sony FE 85mm F1.4 GM +2 more
jnstovall New Member • Posts: 5
Re: LUTs for negative conversion
1

With the April 2018 addition of profiles, we now have a way to invert and color correct negatives with one click in Lightroom. One Click Inversion of Color Film Negatives in Lightroom

Wayne Larmon Forum Pro • Posts: 10,666
Very good way to convert negatives with Lightroom

jnstovall wrote:

With the April 2018 addition of profiles, we now have a way to invert and color correct negatives with one click in Lightroom. One Click Inversion of Color Film Negatives in Lightroom

I just read your blog and I'm impressed.  I spent a long time figuring out how to do this in Photoshop (described earlier in this thread) and the methods you describe in your blog cover the same issues I found out when I was working out how to do it in just Photoshop.  It is tricky and different shots require different treatment, as you describe.

I don't use Lightroom (and hence, can't completely follow your blog.), but it looks like Adobe made Lightroom much more powerful.

Excellent work.

Wayne

Iliah Borg Forum Pro • Posts: 28,757
Re: LUTs for negative conversion

Negative film doesn't represent a constant colour space, unless it is shot under the studio lights that match the film calibration and developed exactly as intended.

Good approaches to solve this are suggested by Don Hutcheson (his targets for profiling colour scanners are known to be among the best available, if not simply the best) in 2009

http://www.hutchcolor.com/PDF/Cneg.pdf

http://www.hutchcolor.com/CMS_notes.html for additional reading

His Photoshop actions are at http://www.hutchcolor.com/Utilities.html

Things can be simplified if the scan or the shot of the negative is in linear space.

Excellent and very poipular scanning soft https://www.hamrick.com contains carefully created negative inversion transforms and can deal with raw files.

Here is a raw from Canon 5D MkIII

Here it is opened in Vuescan, with only white balance set by a click on the plaque:

Another pair:

Vuescan allows to select emulsion types, but I used Generic for the above ones.

Here I selected Ektar:

-- hide signature --
Wayne Larmon Forum Pro • Posts: 10,666
Luminous Landscape thread...

Iliah Borg wrote:

Things can be simplified if the scan or the shot of the negative is in linear space.

The Luminous Landscape thread that a lot of this thread is referencing

http://forum.luminous-landscape.com/index.php?topic=87681.0

consisted largely of me asking questions about converting negatives. I was given excellent advice.  I was told early on to use linear space. My final post on March 7, 2014 said

As I said, I am still writing. Here is what I have thus far:
http://www.frogymandias.org/imagery/

Most of this was written while this thread has been progressing. Comments are welcome!

Yes, we are aware of linear space.  And comments are still welcome.

One other part of my camera scanning project is a 3D printed negative carrier

that was partially based on a project by another DPReview poster.  The idea of my project was to create an open source OpenSCAD script that anybody could use and can be easily adapted to work with any lens, and was robust enough that people could produce .STL files for a slide/negative carrier that could be printed locally on a 3D printer or could be printed by a company like Shapeways.

Comments are also welcome for this project.

Wayne

Iliah Borg Forum Pro • Posts: 28,757
Re: Luminous Landscape thread...

Wayne Larmon wrote:

Iliah Borg wrote:

Things can be simplified if the scan or the shot of the negative is in linear space.

Yes, we are aware of linear space.

Not sure I understand your point. We are indeed aware of linear space :)The thing about using linear space for negatives is well-known from the previous century, since subtraction and white balancing are done in linear space in a more natural way

I thought it is useful to bring linear space up since Mr. Hutcheson sort of omits that in his article, that's all.

Kodak and Fuji software for scanners contained recipes for subtracting masks and inversion for different emulsions, and the best thing to do is to reverse-engineer those recipes, as they are based on the real knowledge of emulsions.

-- hide signature --
Wayne Larmon Forum Pro • Posts: 10,666
Re: Luminous Landscape thread...

Iliah Borg wrote:

Wayne Larmon wrote:

Iliah Borg wrote:

Things can be simplified if the scan or the shot of the negative is in linear space.

Yes, we are aware of linear space.

Not sure I understand your point. We are indeed aware of linear space :)The thing about using linear space for negatives is well-known from the previous century, since subtraction and white balancing are done in linear space in a more natural way

I thought it is useful to bring linear space up since Mr. Hutcheson sort of omits that in his article, that's all.

Kodak and Fuji software for scanners contained recipes for subtracting masks and inversion for different emulsions, and the best thing to do is to reverse-engineer those recipes, as they are based on the real knowledge of emulsions.

Sorry, I thought you skipped over the rest of the thread.  Linear space was discussed extensively in the Luminous Landscape thread we had been referring to.

Doesn't RawDigger export TIFFs in linear space?  I don't think that Lightroom or Photoshop ACR do.

Wayne

Iliah Borg Forum Pro • Posts: 28,757
Re: Luminous Landscape thread...

Wayne Larmon wrote:

Doesn't RawDigger export TIFFs in linear space?

Yes, it does; I typically do it like this, from File -> Export TIFF menu:

You can export linear data from Lightroom if you are specifying a linear output profile, due to Lightroom internals linear ProPhoto is the preferred one.

Choosing Lr export "File settings", you can specify "Other" as "Color Space" for JPEG, PSD, TIFF to select a custom colour profile, and navigate to ProPhoto with gamma 1. You will need to find a ready-made profile of this type, or to make your own (you can easily do it in Photoshop, see the inset on the last page of https://www.ledet.com/margulis/Makeready/MA48-Fate_and_False.pdf for the instructions how to create one - for Step 2, select ProPhoto, for Step 3, set gamma = 1).

-- hide signature --
ggbutcher
ggbutcher Senior Member • Posts: 1,140
Re: LUTs for negative conversion
1

This may sound a bit crazy, but after reading all the stuff I could find on color negative conversion, it dawned on me that the first two things that need to be done to any color negative are 1) remove the orange cast, and 2) invert it. After thinking about #1 for a bit, I asserted in my little brain that the color cast should be removable by just aligning the red, green, and blue channels, so I messed with that until I found that being able to set the black and white point for each channel individually did that. #2, inversion, is simple with a rgb curve, just drag the top-right point to the to-left and drag the bottom-left point to the bottom-right, reversing the slope of the line. Ta-da, color positive. Now, what you're left with to correct looks like white balance, which makes sense as film emulsions were mixed to respond to a particular color temperature.

Here' s one of Iliah's images treated thusly in my hack software:

The top-left pane is the chain of operations. The first is a crop, to isolate the relevant image. Ops 2-4 are blackwhitepoint sets for each of the R, G, and B channels. Op 5 is the invert curve, and 6-7 are single-channel curves to mess with the white balance (I'm too cheap to have a white balance tool... )

For reference, the histogram and the tool panes are showing the last tool in the chain.

Ignore the particularl software for a bit, my conclusion from this is that there is a generic approach to color negative conversion that doesn't depend on particular knowledge of the original film stock. I don't think LUT would be it, because that would have to lock in the transfer functions for the color cast removal, which varies from emulsion to emulsion.

For what it's worth; dpreview posts are cheap!

 ggbutcher's gear list:ggbutcher's gear list
Nikon D50 Nikon D7000 Nikon Z6 Nikon AF-S DX Nikkor 18-200mm f/3.5-5.6G ED VR II Nikon AF-S DX Nikkor 18-140mm F3.5-5.6G ED VR +2 more
Wayne Larmon Forum Pro • Posts: 10,666
Re: Luminous Landscape thread...

Iliah Borg wrote:

You can export linear data from Lightroom if you are specifying a linear output profile, due to Lightroom internals linear ProPhoto is the preferred one.

I'm way behind on this because I don't use Lightroom at all. On the assumption that LR was just Photoshop with a DAM that I didn't want to use. (And because LR never had the content-aware fill tools that are needed to de-ding scanned negatives and old prints.) So now LR can do things that ACR/PS can't? (I dropped out of photography and camera scanning for a few years. I'm back now and have been catching up. Which was why I was excited to see that jnstovall had done new work with negative conversions.)

Back to your original post, I had tried VueScan for negative conversion (when I was asking questions on the 2014 Luminous Landscape thread that was referenced earlier in this thread). But I didn't have good luck with Vuescan negative conversions. Which is why I went with my own method (that was referenced in my March 7 2014 post on the LL thread) that started with a linear TIFF and adjusted in PS from there on out. I found that when I started with a Vuescan conversion that I still had to do much the same amount of curve twisting in Photoshop as I did when I started with a linear TIFF. I didn't get results as good as the ones you posted.  I haven't tried Vuescan for negative conversion since ~2014.

I didn't know of Don Hutcheson's work. I'll study it.

Wayne

jnstovall New Member • Posts: 5
Re: Very good way to convert negatives with Lightroom

Wayne Larmon wrote:

jnstovall wrote:

With the April 2018 addition of profiles, we now have a way to invert and color correct negatives with one click in Lightroom. One Click Inversion of Color Film Negatives in Lightroom

I just read your blog and I'm impressed. I spent a long time figuring out how to do this in Photoshop (described earlier in this thread) and the methods you describe in your blog cover the same issues I found out when I was working out how to do it in just Photoshop. It is tricky and different shots require different treatment, as you describe.

I don't use Lightroom (and hence, can't completely follow your blog.), but it looks like Adobe made Lightroom much more powerful.

Excellent work.

Wayne

Thanks Wayne. If you're not using Lightroom, then the Photoshop action I referenced is all you need to get great color correction of the negatives in Photoshop (or maybe you figured it out in Photoshop for yourself and don't need it). I would prefer to work in Lightroom if possible to avoid creating ginormous TIFF files which is why I'm so excited about the ability to use LUTs.

Your 3d printed negative carriers look to be very handy!

John_a Junior Member • Posts: 25
Re: LUTs for negative conversion
1

ggbutcher wrote:

This may sound a bit crazy, but after reading all the stuff I could find on color negative conversion, it dawned on me that the first two things that need to be done to any color negative are 1) remove the orange cast, and 2) invert it. After thinking about #1 for a bit, I asserted in my little brain that the color cast should be removable by just aligning the red, green, and blue channels

According to this guy the orange mask isn't uniform after development, i.e. it actually contains some of the image. That would mean that just subtracting an orange mask from the negative wouldn't work, because you would be left with a color cast that varies over the image depending on the image content, i.e if you get the color correct in one place it would be wrong in another.

Is he correct? If i google I don't find much about this. I couldn't get his method to work well, but that may be because I didn't use linear space.

Keyboard shortcuts:
FForum MMy threads