Manual inversion methods for scanned negatives

[LocalHero1953]

Advertisement (gone after registration)

Following a suggestion in 'I like film', I am starting this thread to share experience of inverting and processing scanned negatives. It is not for discussion of automatic systems, whether Negative Lab Pro, Vuescan or Silverfast. They have their place, but, for working with digital copies of negatives, I have found NLP unsatisfactory and prefer doing it myself.

My starting point has been Alex Burke's recommendations for working in Photoshop. I work in Lightroom Classic.

My workflow is still not settled, but is essentially this:

1. Copy negatives with my SL2-S and Apo-Macro-Elmarit-R 100mm and ELPRO (for 35mm) or Vario-Elmar-SL 24-90 (for 4x5).
2. Copying is done with manual focus and exposure, with white balance set to 5000K (that's what my light source is), with images recorded as DNG.
3. Import to Lightroom.
4. Adjust white balance using the dropper sampler on the film base. I do this as a batch for each film, incase the film base colour changes.
5. Crop all images to show no film base border. This makes the critical next step easier.
6. Examine each red, green and blue tone curve individually. Invert each one by dragging the bottom left point to the top left and the top right point to the bottom right.
7. For each colour tone curve, set the 'black' points and 'white' points by sliding them from the extremes towards the middle until the histogram shows that (e.g.) the red highlights are blocked out; back off a bit till the histogram shows no such blocking. Do this for all three colours.
8. The image should look not too far from a satisfactory positive image, though typically has a strong cyan cast. Adjust the cyan cast by changing the shape of the green and blue curves to reduce their contribution.
9. Export all images as full res tiff files. Put all DNGs aside, and edit the tiff from now on.
10. I can now edit the tiff as a normal image file, with the sliders working the right way. I now adjust exposure, highlights, shadows, contrast, make further adjustments to colours etc.

Edited July 19, 2023 by LocalHero1953

[LocalHero1953]

With my last films I made a change in the workflow above that I thought would make things easier, but hasn't!

• In step 6, I inverted the negative with the global tone curve, not the individual colour tone curves.
• I exported the file as a tiff without further adjustment of colours.
• Working with the tiffs, I adjusted the red, green and blue 'black' and 'white' end points. There is no need to invert the RGB curves, and I thought this order of doing things would be simpler because the tone curve is the 'right' way round in the tiff. It worked as well as I hoped, but.....
• Subsequent adjustment of saturation, both globally and with individual colours in the HSL box did not work as expected: it changed the colour balance in unexpected ways. 

There is clearly something happening in the interaction between the saturation sliders and the colour tone curves that I haven't worked out yet. 

Edited July 19, 2023 by LocalHero1953

[LocalHero1953]

One further processing step that I often take is because even after adjusting the RGB tone curves I find it tricky to avoid either a slight cyan or slight pink/magenta cast. Adjusting the 'Aqua' slider for Hue towards the Blue end of the spectrum can be very helpful with this.

[LocalHero1953]

I don't believe I am near a fixed and reliable workflow, so I look forward to reading other members' ways of doing this.

I have written the posts above for colour negative inversion. B&W negative inversion is far simpler, but please post comments about B&W workflow as well.

Edited July 19, 2023 by LocalHero1953

[Rosebud55]

Paul,

thank you for this info. I have a lot of film and slides at home and if I ever get round to trying to digitise them I know where to start. Now.

 

rosie

[250swb]

In Photoshop you can often get pretty close by opening the image, then press 'Invert' (it usually comes out very unsaturated), then press 'Auto Color' which adds contrast and fixes the colour shift. From there it's down to refining the saturation and contrast.

I mentioned in another thread that I think 'Invert' in Photoshop has been improved over the years, it fixed a negative I had with a bad colour balance (yes I'd left a light yellow filter on the lens, doh!) which Negmaster or Color Perfect couldn't fix automatically, yet it's not so good it's reliable every time. I have no quibbles with NLP as I've never used it 'cos I don't use Lightroom. 

[hansvons]

Advertisement (gone after registration)

8 hours ago, LocalHero1953 said:

They have their place, but, for working with digital copies of negatives, I have found NLP unsatisfactory and prefer doing it myself.

Same thinking here. Thanks for your insights!! And yes, there‘s something going on with the saturation slider. If pulling that slider either way and adding funky colour casts besides the saturation, something is off in the workflow. I’ll come back to that later.

I started my negative conversion journey by checking a colourist friend who regularly scans and restores films. Long story short, there’s no convenient Kodak conversion LUT or something similar from Cintel. So, I had to find my own workflow. (In C1 you can use a LUT as a camera profile.)

First off, my workflow is quite similar to yours, but I use C1. 
 

1. Neg to pos conversion can be achieved by either turning the RGB curves upside down in the curves tool or by inverting the Levels tool by swapping 255 to 0 and vice versa.

2. Because in C1 the Levels tool is quite powerful, allowing to automatically adjust RGB separately which comes in handy later in the process, I learned that the Curves tool is the more meaningful conversion tool.

3. Now the picture is positive with a strong blue cast. The next step is to select something middle grey in the image and white-balance the shot using the Normalize tool to a middle-ish grey (I use R100, G100, B100). The image looks almost ok now.

4. With the  Levels tool, the next step is to balance the 3 channels individually by pressing the auto icon. It's crucial to set the Leveks tool to individual RGB channels in the app's settings. 

5. Now colours are fine, including neutral shadows and whites. A good way to check that is to crank up saturation. If colours turn funky, redo the process, especially check the WB/Normalize.

6. Next contrast. Rarely, the negs gamma alone matches our eyes preferred contrast. That can be adjusted with the Levels tool and by creating a new filled layer and adding a suitable S-shaped curve.  Also using the HDR sliders make sense, eg reigning in blown highlights with the Shadows sliders (yes, we have to think inverted). Again, working the gamma/contrast should not touch the WB.

7. I adjust the final grade with the 3-Wheels primary grading tool. Eg cooling the shadows a tad or warm/cool the image a bit.

Ok. Sounds easy, but took me weeks to figure that out. I saved that workflow as a Style/preset and things are relatively easy now. BUT only if the neg received the right WB on set. Shooting tungsten-balanced films such a 5219 (love it) in daylight need a more "creative" approach, in terms of WB. Here, I have to WB in the beginning, then level RGB and then WB/Normalize again, and except that colours will be somewhat off regardless of my efforts, most notably in the shadows (back in the day we loved that look for music videos).

Another crucial point is the correct exposure when snapping the negs with the macro/DSLR (SL2S and Sigma 70mm Macro @ f11). As this an inverted process, chances are high that I might end up with too-bright positives. 

Lastly colour negatives tend to need some serious sharping, ideally in the subpixel size. In C1 that would typically be just 0.2 but at a strength of 500 (whatever that measures). Even careful use of clarity can make sense, bringing out the texture.

Hope that makes sense!!

 

[hansvons]

10 hours ago, LocalHero1953 said:

Adjust white balance using the dropper sampler on the film base. I do this as a batch for each film, incase the film base colour changes.

This is what you learn when deep-diving the web. But in my experience, it makes more sense to sample something neutral in the actual picture than sampling the orange base because like digital pictures  film images need to be white-balanced as well (only at 5600K, white is white on regular daylight-balanced films).

Only then, neutrals are neutral and the levelling of the 3 channels will lead to neutral blacks and whites. And only then, increasing the saturation won’t lead to funky colours. Sure, this way you might lose some of the image’s mood but at least you have a correct base for further grading.

I also learned that it's crucial to do first the WB and then the channel adjustments. 

BTW, in C1 the Normalize tool’s picker does a better job than the regular WB picker. 

[LocalHero1953]

2 hours ago, hansvons said:

This is what you learn when deep-diving the web. But in my experience, it makes more sense to sample something neutral in the actual picture than sampling the orange base because like digital pictures  film images need to be white-balanced as well (only at 5600K, white is white on regular daylight-balanced films).

Only then, neutrals are neutral and the levelling of the 3 channels will lead to neutral blacks and whites. And only then, increasing the saturation won’t lead to funky colours. Sure, this way you might lose some of the image’s mood but at least you have a correct base for further grading.

I also learned that it's crucial to do first the WB and then the channel adjustments. 

BTW, in C1 the Normalize tool’s picker does a better job than the regular WB picker. 

Yes, that is what I would do with single 4x5 negatives.

For a 36 exposure film I'm trying to get a happy medium between batch processing and treating each image individually. I use the film base to get within the right ballpark, do a quick inversion, then cull anything I don't want to edit further. After that I might do a WB on an image by image basis. But I can't claim to be very rigorous.

What Lightroom lacks is an automatic 'black' and 'white' point setting tool for the individual RGB curves, as it has in the basic exposure panel (Auto White and Auto Black). It ought to be a simple tool for Adobe to add, and would speed up the process a lot. 

[hansvons]

16 minutes ago, LocalHero1953 said:

What Lightroom lacks is an automatic 'black' and 'white' point setting tool for the individual RGB curves, as it has in the basic exposure panel (Auto White and Auto Black). It ought to be a simple tool for Adobe to add, and would speed up the process a lot. 

That’s what I can do in C1 in the Levels tool. One click and all 3 channels are aligned individually to 0 and 255.

[LocalHero1953]

1 minute ago, hansvons said:

That’s what I can do in C1 in the Levels tool. One click and all 3 channels are aligned individually to 0 and 255.

😬

But I'm going to stick with LR for the moment.

[hansvons]

I’ve never used LR on real-life projects, because time and consistent results are vital to me, and I have an established and satisfying workflow using C1. But I will give it a try because I may find time this week (summer holidays) and I’m curious how your workflow works on my images.

---

In the beginning, I tested the Negative Lab Pro workflow in LR, but it didn't work for me because I like to understand what's going on. The results were ok, but I got better results doing it the hard way in C1, which I now find relatively convenient. 

I totally agree on not jumping ship when there’s no real reason.

[LocalHero1953]

1 hour ago, hansvons said:

In the beginning, I tested the Negative Lab Pro workflow in LR, but it didn't work for me because I like to understand what's going on.

Exactly. And I always found the need to do some more editing after NLP had done its bit, which made it pointless.

[hansvons]

I checked LRC and missed C1's Level tool badly. And LRC's WB tool's scope is too narrow to deal with tungsten-based colour negative shots in daylight without the 85-filter. (Shooting on 5219 without the 85 conversion filter is common practice in cinematography).

I would give C1 a shot. In terms of tools, it's distinctly more powerful. I think there's a reason to test it. Below is a screenshot of the reason ;).

 

 

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[LocalHero1953]

I tried C1 many years ago, and couldn't get on with its interface; LR was better as a DAM and, certainly then, for batch processing. I am also heavily committed to distributing photos to others via Adobe's online galleries. And my website is linked via a plug-in to LR collections. But I suppose I could give it another go!

[hansvons]

This is how the pipeline is set up in C1.

 

First, the inversion:

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

 

Then adjusting to the correct exposure:

 

After that, the white balancing. In C1, you'll find a regular WB tool and a targeting WB tool called Normalize. They work together. Normalize does an excellent job, even in almost black areas. Don't check the Exposure box.

 

And lastly, the mighty Levels tool. It's indispensable once you get the knack of it. IMO, easily worth the yearly extra costs. And, BTW, even the most advanced gradings systems for film productions don't have that in their arsenal. It's crucial to set up the correct settings in the settings pane. I also like to set the target levels to dark grey and off-white, giving me headroom and the option to adjust to the desired black and white levels later. See below.

[hansvons]

14 minutes ago, LocalHero1953 said:

I tried C1 many years ago, and couldn't get on with its interface;

I have a history as a part-time finishing artist in the adverts films sector. Interfaces and applications in that area offer an abyss of options and tools and an incredibly steep learning curve with the danger of self-annihilation. From that perspective, C1 is as easy as pie ;). I agree; LRC is much easier to access in the beginning. At some point, C1 can be really fast, especially when you take advantage of the Speed Edits option and the new Smart Adjustment tool, the latter one only meaningful if you were shooting reportage-style stuff or weddings. 

[105012]

Levels tool in Apple Photos seems to work similarly to the C1 above?

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

A less expensive option if you have a Mac, though I’m sure there are specialised features of value in the 3rd party software.

[John Robinson]

On 7/19/2023 at 8:24 AM, LocalHero1953 said:

I look forward to reading other members' ways of doing this

I tried the various manual inversion options in LR and C1 but after a while I find reading the DNG from the camera (SL on a BEEON in my case, with Micro-Nikkor 3.5/55 at about 5.6) into Vuescan with some basic settings for colour negative and output as TIF produces quite acceptable files that can then be imported into C1.

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[LocalHero1953]

On 7/19/2023 at 6:53 PM, hansvons said:

This is what you learn when deep-diving the web. But in my experience, it makes more sense to sample something neutral in the actual picture than sampling the orange base because like digital pictures  film images need to be white-balanced as well (only at 5600K, white is white on regular daylight-balanced films).

Only then, neutrals are neutral and the levelling of the 3 channels will lead to neutral blacks and whites. And only then, increasing the saturation won’t lead to funky colours. Sure, this way you might lose some of the image’s mood but at least you have a correct base for further grading.

I also learned that it's crucial to do first the WB and then the channel adjustments. 

BTW, in C1 the Normalize tool’s picker does a better job than the regular WB picker. 

Thanks for this comment. I agree, it makes a big difference: the closer one can get the white balance before adjusting the RGB curves, the easier the latter is, and you're right, the film base is not a good sampling point. Sometimes it is difficult to see a suitable target for WB sampling. Then I have to iterate: approximate a good WB > adjust RGM curves > take a new WB sample > adjust RGB curves again. But the extra steps are worth it in avoiding colour casts.