Convert your digital film negatives (linear TIFF files) to positive images. Other features:
- Gamma correction for adjusting midtones
- Sigmoidal contrast for highlights and shadows adjustment
- Grayscaling: B/W lovers save up to 60% disk space.
- Resizing when megapixels are not everything
- Zipped TIFF or JPEG conversion
What happens inside? ImageMagick does the image editing. GNU Parallel mangles multiple images in parallel, depending on your CPU cores.
These first releases of this package are developed and tested with B/W negatives.
The positive bash script can be installed by a Homebrew formula, which itself is part of the tomkyle/homebrew-negatives tap.
# Variant 1:
# Install tap first, formula second
$ brew tap tomkyle/negatives
$ brew install positive
As “tapping” first is not neccessarily needed, you can install the formula directly:
# Variant 2
$ brew install tomkyle/negatives/positive
Open your terminal application and go to your images directory. positive will work in the current working directory. Run positive --help
or -h
to display help text.
$ positive [options] [--all | file(s)]
See Options · Examples · Gamma correction · Sigmoidal contrast · Workflow Recommendations · Issues and FAQ
Files and Output · Color and Contrast.
All images (batch mode). Process any TIFF file in working directory.
Switch on debugging mode, showing everything that's going on.
Mirror the image vertically and/or horizontally. Possible values are flop
horizontal, flip
vertical or even flipflop
(guess what). Example: -f flop
Save image as JPG with given quality. Example: -j 90
Output directory — default is current working directory. Example: -o results
Resize image — pixel width for larger side, preserving aspect ratio. Example: -r 3000
Verbous mode — show some more information under way.
Files and Output · Color and Contrast.
Desaturate colors, recommended for B/W negatives. Bonus: The TIFF will be converted to 16 bit Grayscale, saving up to 60% in file size; The image will have linear gamma 1.0 ICC profile applied (Gray-elle-V4-g10.icc).
Gamma correction value to apply. It is highly recommended to use this parameter with linear TIFF images, together with -n. Pass a float value like 2.4
or auto
. Auto will apply a calculated gamma adjustment based on the mean values of an image. Examples: -g 2.2
and --gamma auto
Stretch histogram to reach black and white points. Recommended for most images.
Sigmoidal contrast value around 50% middle-gray. Increases the contrast without saturating highlights or shadows. Quoted from ImageMagick docs: “3 is typical and 20 is a lot.” Example: -s 5
See Usage · Examples · Gamma correction · Sigmoidal contrast · Issues and FAQ
Given your photographed negative is a B/W image, so we are using desaturation and grayscaling here. The result will be a grayscale TIFF with approriate ICC profile. Please note that the positive image gets a -positive
suffix when it is stored in the same directory like the original image.
# These are equal:
$ positive --desaturate DSC_0123.tif
$ positive -d DSC_0123.tif
# Output
DSC_0123.tif
Positive ✔ Result DSC_0123-positive.tif
Second example: Again, we use desaturation, and this time the positives go as JPG into a subdirectory. The results are now stored in a foobar
directory (Note: no -positive
suffix).
# These are equal:
$ positive -d --jpg 85 --output foobar DSC_0123.tiff DSC_0124.tiff DSC_0125.tiff
$ positive -d -j 85 -o foobar DSC_0123.tiff DSC_0124.tiff DSC_0125.tiff
# Output
DSC_0123.tiff
Positive ✔ JPG-80 ✔ Result foobar/DSC_0123.jpg
DSC_0124.tiff
Positive ✔ JPG-80 ✔ Result foobar/DSC_0124.jpg
DSC_0125.tiff
Positive ✔ JPG-80 ✔ Result foobar/DSC_0125.jpg
Just pass --all
or -a
option, and the script will mangle everything that has a TIFF extension: tif, tiff, TIF, TIFF
etc.
The batch mode uses GNU Parallel, so every single CPU core will be used. So if you have a Quadcore CPU, four images will be processed at the same time. Great timesaver!
Since normalizing is a good idea, we use the --normalize
or -n
option in addition to desaturation with -d
.
$ positive --all -d -n -o results
$ positive -adn -o results
# Output
Process 3 images, using GNU Parallel:
DSC_0123.tiff
Positive ✔ Result foobar/DSC_0123.tiff
DSC_0124.tiff
Positive ✔ Result foobar/DSC_0124.tiff
(...and many more...)
Some stats:
CPUs used: 4
Elapsed time: 0min 30sec
Done: 16 images
See Usage · Options · Examples · Gamma correction · Sigmoidal contrast · Issues and FAQ
If you are working on linear TIFFs, e.g. as produced by Dave Coffin's dcraw by its -4
flag, both the negative and the positive will look somehow flat, due to their linearity. The dark negative becomes a very light positive.
The positive utility inverses the negative TIFF to positive, and exactly now we are able to perform the very gamma correction that did not take place when dcraw
created the linear TIFF—with the difference that we can choose the gamma as we like.
Adjust the midtones: In fact, the light positive must be darkened, using a gamma value lesser than 1. Since such small gamma values are “uncommon” in human-driven image editing, it is easier to talk about “common” gamma values like 2.2 here. Internally, the script then calculates the reciprocal value (0.45) for darkening.
The --gamma
or -g
option carries the Gamma correction value to apply. It is highly recommended to use this parameter with --normalize
or -n
.
positive -adn --gamma 2.2 -o gamma-corrected
positive -adn -g 2.2 -o gamma-corrected
Which gamma value to choose? In fact, the gamma value is completely up to your taste. An appropriate gamma value will be mainly influenced by these two factors:
- The negative density which results from your film development workflow,
i.e. its natural contrast and blackness - The digicam exposure chosen during digitalization,
i.e. how light or dark is your photo from the negative.
Recommendation: Always worth trying is --gamma auto
or -g auto
. This will calculate a mean medium gray which is quite suitable for images taken in ‘real life’ without extreme exposure. Here are some examples for typical values (according to my personal workflow):
gamma | description |
---|---|
auto | Calculated gamma, based on mean values. Implies gamma 2.2 and is recommended for ‘real life images’. |
1.0 | No Gamma for linear TIFFs. This is the default value if omitted. |
1.8 | Moderate darkening, resulting in overall light images |
2.2 | Well-known software standard; This is what common photo software would apply. |
3.6 | Smooth tones, not too light |
4.4 | Somehow darker, still mellow. Not too contrasty. |
... | Try your own! |
Read more about Gamma correction: Wikipedia and ImageMagick: Gamma Adjustments and auto-gamma.
Adjust highlights and shadows: While the gamma correction mainly affects the midtones, the sigmoidal contrast control works on the highlights and shadows, leaving the 50% midtone alone and resulting in a non-linear, s-like curve. Use the --sigmoidal value
or -s value
option to enhance the contrast. Quoted from ImageMagick docs: “3 is typical and 20 is a lot.” – Example:
positive -adn --g 3.6 --sigmoidal 5 -o nice-contrasts
positive -adn --g 3.6 -s 5 -o nice-contrasts
Read more: ImageMagick
See Usage · Options · Examples · Gamma correction · Sigmoidal contrast · Issues and FAQ
When photographing your negatives, you'll probably use a negatives (film) holder.
- The edges of your RAW photo usually will then have a black frame.
- If the window of your negative holder is larger than the negative, your image will also show unlit film areas (i.e., spaces between the single negative frames), resulting in another “white“ frame. These unlit areas will become ‘absolute black’ in your positives version.
The following workflow examples assume your linear TIFFs are framed black and white, having used a negative holder with larger frame window.
Start cropping your images and save them as TIF. Be sure to preserve the ICC profile!
Run positive with gamma and sigmoidal contrast as needed. When using the -n
parameter for normalizing colors, both the black and white points will be determined by the information in your image. Consequences are:
- The higher gamma value or sigmoidal contrast you use, the earlier deepest shadows and lights will start blocking.
- Underexposured or overexposured images are normalized, that is ‘stretched to histogram edges‘, and may loose their generic underexposured or overexposured look. Images being flat in original may show more film grain.
Run positive with gamma and sigmoidal contrast as needed. When using the -n
parameter for normalizing colors, both the black and white points will be determined by the darkest grays of the negative holder edges and the lightest parts in the unlit negativ areas. Deepest shadows and lights are now in the (surfluous) frames. Consequences are:
- You may choose way higher gamma and sigmoidal contrast values.
- The look of underexposured or overexposured images will overall stay natural.
- Long option names for those preferring self-explanatory options like
--resize
,--desaturate
and so on. - Improved code quality: Wrapped main features in single functions; Adhere to Bash best practices and coding standards.
These features go into the current major version 1:
-
Star rating filter: Many photo managers like Lightroom or Bridge let their users reject bad images or rate better ones with ‘stars’. positive should get a new CLI option flag to set a minimum rating level. Feel free to discuss this in issue #6.
-
Custom configuration files: Would it not be fine if users could store their favourite options in a configuration file?
~/.negativesrc
or~/positive.conf
or even an INI, YAML or JSON? Head over to the corresponding issue #7.
-
The -r option will be renamed to -w, as -r is more natural for the upcoming Rating filter, and so is -w for width.
-
The -f option will be renamed to -m, as the actually performed image action is mirroring horizontally or vertically. The option values flip, flop and flipflop will then become something like V, H or VH.
-
New batch mode trigger: New sub-command
all
will replace the current-a
flag, like so:linear-tiff batch <options>
.
To see the full list, head over to the issues page.
Invalid output profile after converting linear Grayscale TIFFs to positive.
Adobe CC Bridge won't render preview image and Photoshop complains about invalid ICC profile. Current workaround is to use --desaturate
also on grayscale linear TIFFs. Stay tuned: issue#8
I get a error message “mogrify: delegate library support not built-in”
ImageMagick must be compiled with litte-cms2 support. See issue#1 for details.
$ git clone https://github.com/tomkyle/negatives-positive.git