How to get a super result using Film Negative?

[obe (Olaf Berthelsen)]

I decided to spend some hours testing the new Film Negative tool.

I put together a simple device (see below) to be able to shoot images of old negatives using my camera and a close-up lens.

A workshop lamp (150 W), a piece of white paper as a diffuser and two clothes pegs to hold the negative in position at a distance.

My 20 year old print looks like this:

The first challenge was to shoot a well-exposed photo of the negative. I tested different camera settings and most of the time the output from the Film Negative tool was very blue and dark. Moving the Reference Exponent slider resulted in this somewhat hazy image:

Applying Exposure, Sharpening, Local Contrast, Noise Reduction, Haze Removal and White Balance resulted in this:

It’s amazing that it is possible to convert the negative into this positive but the colors are clearly not super.

How can I get an even better result?

Does anybody know a systematic approach on how to proceed from the first Film Negative output to a super result?DSC_1390.NEF (25.3 MB) DSC_1390.NEF.pp3 (12.2 KB)

[stefan.chirila]

looks like the White Balance could use less yellow and less green. Once I get home today I’ll take a look and try do an edit.

[Thanatomanic (Roel)]

Nice setup and good results! Just wondering: aren’t you worried about uneven illumination by the lamp you’re using, and the produced heat possibly damaging your negative?

[luxapy (lux)]

Hi,
I digitized a lot of old films using the Nikon ES-1 (~90€) duplicator.
Normal shot, in RAW and without special correction.
I had tried to create an inverse filter of this orange found on negatives, just before the inversion, but I never got really correct positives.
The only good results I got were obtained by manually correcting the colors by white balance and curves.

[age]

It’s easier to white balance the photo with only color pickers in the invert negative module but there are strange artifacts at pixel levels

Invert with Film negative

DSC_1390.NEF.pp3 (12.3 KB)

Invert with curves

DSC_1390.NEFb.pp3 (12.5 KB)

[Thanatomanic (Roel)]

I noticed the artifacts too, the red channel kinda goes haywire in the Film Negative module.
@rom9 could you take a look if this is normal behavior?

With minimal edits I see this in the red channel. Making further changes seriously enhances the artifacts. The green and blue channel are not or much less affected.

DSC_1390-1.jpg.out.pp3 (12.0 KB)
My take using the dev RT 5.7 version. I really had to go overboard on the Lab chromaticity and Vibrance to get some colour back.

[martin.scharnke]

@Thanatomanic

I really had to go overboard

[rom9]

Hi all, i tried to improve the result, here’s what i’ve got:

and the .pp3:
DSC_1390.NEF.pp3 (11.6 KB)

Did you use the “Pick white and black spots” tool? You are lucky with this picture, because there are plenty of spots. I picked the black stripe in the foregroud boat, and the white umbrella (…roof? what’s it called? please excuse my english :-D) on the background boat.
This finds the most appropriate exponents for the picture. Most probabily, i’ll be able to use the same values with most other pictures in the same film roll.

@Thanatomanic as for the artifacts, try using RCD demosaic method. I remember @heckflosse suggesting me to use RCD, but i can’t remember why. Anyway, trying it on this picture seems to smooth things out a bit. (There’s no use in choosing a super-sharp method since film grain is the limit here).
Also, the source negative shot is a bit too underexposed… look at the raw histogram, see how it is heavy on low values? This is wasting bit depth without any reason, i think.

@obe, could you try re-shooting the negative with higher exposure? You should definitely see some clipping in the in-camera preview. Then check the raw histogram in RT, that one must not clip.
This could (maybe) also improve the artifact problem.
I noticed you have a large border around the negative picture: try shooting closer and get a bigger negative if your lens allows. That’s because currently, calculations are made on the whole picture, so that border is affecting channel scaling, forcing you to use weird values for exposure and WB.
If you can’t shoot closer, there are cool features coming down the pipe, so stay tuned for next versions

Finally, try to use the specific color profile for your camera, if you have it. With my Sony, i discovered that if i use the standard DCP profile, applying the Look Table helps a lot (exactly as it does for a normal digital picture)

ps.: As @Thanatomanic i would also be worried about the amount of infrared that comes out of that halogen. For what i know, old slide projectors had a special element between the lamp and the film to filter out infrared. And… a huge loud fan

pps.: BTW, do you know what type of film it is? Just for reference. Thanks

[stefan.chirila]

Resizer: DSC_1390-1.jpg.out.pp3 (11.3 KB)
Actual edit: DSC_1390.NEF.pp3 (11.4 KB)

[afre (Alan)]

@obe

Please provide a license for your image.

Lovely image BTW. I wonder if this ship still exists and where it is moored.

Looked up Sadko, not knowing the reference. Good name…?

[obe (Olaf Berthelsen)]

@Thanatomanic

aren’t you worried about uneven illumination by the lamp you’re using, and the produced heat possibly damaging your negative?

The setup is just “quick and dirty” for testing. The lamp produces a lot of heat so it can be turned on for very short periods (15 sec. or so) only.

It turned out that the close-up lens’ shallow depth of focus evens out any uneven illumination caused by the lamp and the white paper when the negative is placed at a distance (6 cm) from the white paper used to diffuse the light.

[obe (Olaf Berthelsen)]

@luxapy

The only good results I got were obtained by manually correcting the colors by white balance and curves.

MeToo and manual correction requires some effort adjusting each individual photo in order to get a good result.

[Sunhillow (Hansgeorg Zauner)]

This one is it:

SADKO (Sailing Vessel) Registered in Spain - Vessel details, Current position...

Really a nice photo, Olaf!

For digitizing negatives I somewhere read the advice to use a blue filter to avoid oversaturation of the red channel. Maybe this would help in this case? You used incandescent light which makes it even more red.

[obe (Olaf Berthelsen)]

@age

It’s easier to white balance the photo with only color pickers in the invert negative module but there are strange artifacts at pixel levels

I like your result using film negative. The yellows are much better. The result is apparently obtained in a simple way: a standard tone curve, white balance and film negative.

I understand your use of the film negative tool as follows: You pick pairs of black and white spots several times until you are satisfied. Only one pair of spots can be active at a time. Correct?

What was your sequence of tuning? 1) film negative, 2) tone curve and 3) white balance?

[obe (Olaf Berthelsen)]

@Thanatomanic

My take using the dev RT 5.7 version. I really had to go overboard on the Lab chromaticity and Vibrance to get some colour back.

This is also a good result and obtained in a very simple way: general lab sliders and the vibrance slider. No curves!

I tend to use more and more tools hunting a good result. But that is not necessary……

[agriggio (Alberto)]

I tried to match your old print – just for fun, not because I have any suggestion on how to actually do this efficiently (sorry). Anyway, here’s what I could get, using ART:

DSC_1390.NEF.arp (10.8 KB)

[obe (Olaf Berthelsen)]

Hi’ @rom9

Thank you for a thorough feedback and good advices.

Yes, I tried to use “Pick white and black spots” but I was not satisfied with the results. Maybe I should have tried some more spots

@rom9

could you try re-shooting the negative with higher exposure

Just to make sure. You think that the image is too dark from the outset?

@rom9

I noticed you have a large border around the negative picture: try shooting closer and get a bigger negative if your lens allows

I have one shot a little closer to the negative with the same camera settings but it turned out to be a bit unsharp, so I didn’t use it. When I invoked the film negative module the default result was much better. Now I understand that this was due to the narrower border. I think that the calculation should be changed to work on the cropped area only. After all, how close can you get to a 24x36 negative? Yes, I will certainly stay tuned……
I will try to follow your other advices. I’m leaving for a short holiday so I will have to wait until next week. Sorry !

The negative are marked as follows: FUJI S-400 G79

[obe (Olaf Berthelsen)]

Hi’ @stefan.chirila

Also a fine result created in yet a different way!

I understand your pp3 as follows: film negative (default settings), local contrast (default settings), white balance and an extremely steep tone curve 1 (film-like).

[obe (Olaf Berthelsen)]

@afre

Please provide a license for your image.

How do I provide a license?

The photo was shot 20 years ago in Barcelona using my old Nikon FE.

[obe (Olaf Berthelsen)]

Hi’ @agriggio

The result is very, very close to the original print!
What is ART and how does ART differ from RT 5.7?