home | download | howto | module docs | git

raw image processing workflow which sucks less

vkdt is a workflow toolbox for raw photography and video. vkdt is designed with high performance in mind. it features a flexible processing node graph at its core, enabling real-time support for animations, timelapses, raw video, and heavy lifting algorithms like image alignment and better highlight inpainting. this is made possible by faster processing, allowing more complex operations.

the processing pipeline is a generic node graph (DAG) which supports multiple inputs and multiple outputs. all processing is done in glsl shaders/vulkan. this facilitates potentially heavy duty computational photography tasks, for instance aligning multiple raw files and merging them before further processing, as well as outputting intermediate results for debugging. the gui profits from this scheme as well and can display textures while they are still on GPU and output the data to multiple targets, such as the main view and histograms.




there are up-to-date packages (deb/rpm/pkgbuild) in the

opensuse build system

also there are nixos packages vkdt and vkdt-wayland which you can use to try out/run on any linux distro, for instance:

  nix-shell -p vkdt

build instructions

you should have checked out this repo recursively. if not, do

  git submodule init
  git submodule update

to grab the dependencies in the ext/ folder. you should then be able to simply run 'make' from the bin/ folder. for debug builds (which enable the vulkan validation layers, so you need to have them installed), try

  cd bin/
  make debug -j12

simply run make without the debug for a release build. note that the debug build includes some extra memory overhead and performs expensive checks and can thus be much slower. make sanitize is supported to switch on the address sanitizer. changes to the compile time configuration as well as the compiler toolchain can be set in config.mk. if you don't have that file yet, you can copy it from config.mk.defaults.


the binaries are put into the bin/ directory. if you want to run vkdt from anywhere, create a symlink such as /usr/local/bin/vkdt -> ~/vc/vkdt/bin/vkdt.

  cd bin/
  ./vkdt -d all /path/to/your/rawfile.raw

raw files will be assigned the bin/default-darkroom.i-raw processing graph. if you run the command line interface vkdt-cli, it will replace all display nodes by export nodes. there are also a few example config files in bin/examples/. note that you have to edit the filename in the example cfg to point to a file that actually exists on your system.


our code is licenced under the 2-clause bsd licence (if not clearly marked otherwise in the respective source files). there are parts from other libraries that are licenced differently. in particular:

rawspeed: LGPLv2 imgui: MIT

and we may link to some others, too.


the full list of packages used to build the nightly appimages can be found in the workflow yaml file.

optional (configure in bin/config.mk):

you can also build without rawspeed or rawler if that is useful for you.


[qvk] dev 0: NVIDIA GeForce RTX 2070

and then place this printed name exactly as written there in your
~/.config/vkdt/config.rc in a line such as, in this example:

strqvk/device_name:NVIDIA GeForce RTX 2070

if you have several identical models in your system, you can use the device number vkdt assigns (an index starting at zero, again see the log output), and instead use

April 2024