Debayer is a commandline tool to convert camera raw images into scene-linear exr.
- OpenImageIO - oiiotool is used for converting debayered tif images to exr.
- RawTherapee - Powerful raw development software used to decode raw images. High quality, debayer algorithms, and more advanced raw processing like chromatic aberration removal.
- LibRaw - dcraw_emu commandline utility included with LibRaw. Optional alternative for debayer. Simple, fast and effective.
usage: debayer [-h] [-o OUTPUT] [-w] [-p PROFILE] [-ca] [-r RESIZE]
[-e EXPOSURE] [-f FILTER]
input_paths [input_paths ...]
Debayer is a commandline tool to convert camera raw images into scene-linear
exr.
positional arguments:
input_paths Source(s) to process. Can be one or more images or
directories containing images.
optional arguments:
-h, --help show this help message and exit
-o OUTPUT, --output OUTPUT
Output directory. If not specified, the current
directory will be used.
-w, --overwrite Overwrite existing output files.
-p PROFILE, --profile PROFILE
Optional override to specify a custom rawtherapee pp3
profile to use. If none, the default config file will
be used.
-ca, --aberration Remove chromatic aberration (rawtherapee only)
-r RESIZE, --resize RESIZE
Apply a resize to the output image. Aspect ratio is
not preserved if both width and height are specified.
<width>x<height>. (ML 1x3) 1920x2348 -> 5760x2348 ->
3840x1565 -> 2880x1174 (ML 1x3 2.35) - 1808x2300 ->
(unsqueeze) 5424x2300 -> (0.5) 2712x1150 Or 1280x2160
(1x3) -> 3840x2160 -> 2560x1440 -> 1920x1080 Or
preserve aspect ratio if w or h = 0: e.g. "1920x0" Can
also be a percentage like "50%"
-e EXPOSURE, --exposure EXPOSURE
Raw to scene-linear exposure adjustment. Default is
4.0
-f FILTER, --filter FILTER
Include only files that match regex. Can be comma
separated list.
First thing, you need to configure a few things in the debayer python file. At the top are a few variables you need to customize so debayer knows where to find the right executables.
Most importantly, you need to set the location of the rawtherapee-cli and oiiotool executable files (or dcraw_emu if using this instead of rawtherapee).
RT_BIN = '/usr/bin/rawtherapee-cli'
DCRAW_BIN = '/usr/bin/dcraw_emu'
OIIO_BIN = '/usr/bin/oiiotool'
You also need to specify a temp directory for debayer to use as a location for intermediate files (don't worry they are cleaned up automatically).
CACHE_DIR = '/var/tmp/rt'
Using debayer is very simple if you are familiar with commandline utilities. In the simplest possible form, you could write
debayer rawfile.cr2
This will process the raw file into the directory you are currently in.
If you want to specify a custom directory you could do
debayer rawfile.cr2 -o /path/to/output_dir
You can also process an entire source directory recursively. For example say you have this source directory structure:
/media/footage/20211012
└── dng
├── M22-1558
│ ├── M22-1558.000000.dng
│ ├── M22-1558.000001.dng
│ ├── M22-1558.000002.dng
│ ├── M22-1558.000003.dng
│ ├── M22-1558.000004.dng
└── M22-1600
├── M22-1600_000000.dng
├── M22-1600_000001.dng
├── M22-1600_000002.dng
├── M22-1600_000003.dng
└── M22-1600_000004.dng
Say you want to recursively process all raw files inside /media/footage/20211012/dng, and output the results into /media/footage/20211012/exr. You could run
debayer /media/footage/20211012/dng -o /media/footage/20211012/exr
# or alternatively
cd /media/footage/20211012
debayer dng -o exr
One important parameter when debayering to scene-linear is how much to expose up with the -e parameter. Camera raw files are usually stored in an integer data format, which means the data is contained in a 0-1 container. How many bits of precision that data has within that container varies with the camera and sensor.
To get this 0-1 data range into sensible scene-linear data, we need to expose up the raw data so that an 18% grey diffuse reflector will live at a scene-linear value of around 0.18. How much we need to expose up will of course depend on how the image was exposed in the camera. This also determines how much headroom you will have in highlights. A default value of 4.0 is provided as a starting point, but you will probably want to customize this depending on your source raw images.
You can specify a custom rawtherapee config pp3 file with -p path/to/file.pp3. This is useful if you need to customize whitebalance or sharpening settings for examples.
A default config is included in the repo and is used by default unless you change the RT_DEFAULT_PROFILE variable in the debayer code. This default profile outputs to linear ACEScg, and uses the RCD debayer algorithm, which is the best quality.
If you are curious about the different available debayer algorithms in open software I have put together a big set of comparison images in jpg and exr available at this mega.nz link.
The images are processed with dcraw_emu for AHD, AAHD, DHT and VNG, and with rawtherapee-cli for RCD, DCB and AMaZE. The test images are a selection of 256x128 pixel crops of interesting image regions for judging debayer quality. My subjective ranking would be
RCD > DHT > DCB > AMaZE > AAHD > AHD > VNG
oiiotool can be installed with common pacakge managers on all platforms. See this documentation for help.
Packages are provided for all platforms. However if you are on linux, the appimage won't give you access to the rawtherapee-cli commandline utility. You will need to install or compile it. There is a copr for many fedora and rhel distros here, which might help you.
LibRaw provides packages for Windows and Mac, and for Linux it is easy to compile with minimal dependencies. I would recommend compiling from master to get the latest raw formats like Canon CR3.