Welcome to the DeepC, a suite of Deep compositing plugins for Foundry's Nuke. While I'm sure large studios have something similar to these tools, this open-source project makes them available to small and mid-size studios.
What are the tools offered in the DeepC toolkit? The Wiki has a description for each and an overview of all available plugins.
Also, head over to the Examples page for some visuals.
You can read a quick description of some below.
DeepCAdd
DeepCAddChannels
DeepCAdjustBBox
DeepCClamp
DeepCColorlookup
DeepCConstant
DeepCCopyBBox
DeepCGamma
DeepCGrade
DeepCID
DeepCHueShift
DeepCInvert
DeepCKeymix
DeepCMatrix
DeepCMultiply
DeepCPMatte
DeepCPNoise
DeepCPosterize
DeepCRemoveChannels
DeepCSaturation
DeepCShuffle
DeepCworld
Just a toy at the moment which performs a gain on the image, but demonstrates an approach to getting Blink working on Deep images. This approach works for kernels which need only the current pixel to operate. Experimental proof-of-concept, not really "useful" as a node.
DeepC has several advantages over DeepExpression-based solutions to working with Deep data.
World-position noise in Nuke has always been limited by Nuke's 3D noise algorithm. It's fine for a 2D image, but when working with 3D noise, your really need that fourth dimension so the noise can change over time. DeepCPNoise brings 4D noise to Nuke for the first time, by including and extending the open-source FastNoise library. 4D simplex noise in Nuke!
DeepCGrade works just like the regular Grade node. You can mask it with rotos from the side input (incredibly, this is not available in the standard toolset, or with the DeepExpression node). You can use Deep masks coming in from the top. You can easily configure the channels to process and those to use as masks, just as you normally would.
The power of DeepC is limited only by the power of the NDK. Upcoming features like DeepCAddChannels and DeepCShuffle are only possible using the NDK.
Now that basic functionality is implemented, the focus will be on accelerating the speed of the toolset using the new NDK funcitonality available in Nuke 11.3.
Check the releases page for this repo to get the compiled versions of the plugins. Currently, I've only compiled them for Linux, but that's presumably the most useful for folks, anyway. Windows versions as and when I can, if I can. If you'd like to compile for Mac and contribute back, that would be much appreciated, as I don't have a Mac for development.
Coming soon...
Much like how samples can be compressed in the renderer by merging samples closer than a certain threshold, this node will allow merging of samples in the Deep stream to cut down on processing time when the render is too heavy.
Build has been tested on Centos 7 with devtoolset-3 (not Centos 6 with devtoolset-2, as recommended by Foundry - presumably very few people are still using Centos 6 in production, and it seems to work fine). I think the most recent build I did was using Centos 7 and devtoolset-7, but I don't have that VM around anymore so I'm not 100%. Let me know how it goes...
Thanks to Nathan Rusch (https://github.com/nrusch?tab=repositories) for contributing the CMake build setup which will form the basis for building DeepC going forward.
Install prerequisites:
sudo yum install centos-release-scl
sudo yum install devtoolset-3
sudo yum install mesa-libGLU-develClone:
git clone --recurse-submodules https://github.com/charlesangus/DeepCAdd to .bashrc on dev machine, or run before each build:
# enable devtoolset-3
source /opt/rh/devtoolset-3/enableThen, from the DeepC dir:
mkdir build; cd build
cmake -D CMAKE_INSTALL_PREFIX="`pwd`/../install" ..
make -j $(nproc --all) installAnd, of course, update CMAKE_INSTALL_PREFIX to your preferred install location.
Note: additionally you can adjust the Nuke Version via -D Nuke_ROOT="<PATH_TO_NUKE_ROOT_FOLDER>"
Where <PATH_TO_NUKE_ROOT_FOLDER> is the path to the prefered nuke root like C:/Programs/Nuke12.2v2 or /usr/local/Nuke12.2v2
e.g.
cmake -D CMAKE_INSTALL_PREFIX="`pwd`/../install" -D Nuke_ROOT="/usr/local/Nuke13.1v1"
docker run -v /usr/local/Nuke15.1v1:/usr/local/Nuke15.1v1:ro -v $PWD:/opt/aswf -it --rm aswf/ci-base:2024.1 /bin/bashdocker run -v /usr/local/Nuke14.1v1:/usr/local/Nuke14.1v1:ro -v $PWD:/opt/aswf -it --rm aswf/ci-base:2022.4 /bin/bashdocker run -v /usr/local/Nuke13.2v9:/usr/local/Nuke13.2v9:ro -v $PWD:/opt/aswf -it --rm aswf/ci-base:2020.9 /bin/bashThen follow Linux instructions to build and install.
On Windows you need Visual Studio (15) 2017 to compile the plugins. You can use cmake (gui application-recommended) to create the project files for Visual Studio. Note: make sure to use the x64 platform to compile the plugin.
If you want to build the tool via command line (or us a batch script) use:
mkdir build
cd build
cmake -G "Visual Studio 15 2017" -A x64 .. -DCMAKE_INSTALL_PREFIX=install -B build
cmake --build build --config Release
cmake --install build
Note: additionally you can adjust the Nuke Version via -D Nuke_ROOT="<PATH_TO_NUKE_ROOT_FOLDER>"
Where <PATH_TO_NUKE_ROOT_FOLDER> is the path to the prefered nuke root like C:/Programs/Nuke12.2v2 or /usr/local/Nuke12.2v2
e.g.
mkdir build
cd build
cmake -G "Visual Studio 15 2017" -A x64 .. -D CMAKE_INSTALL_PREFIX=install -D Nuke_ROOT="C:/Program Files/Nuke13.1v1" -B build
cmake --build build --config Release
cmake --install build
Unlikely...
We provide an install batch script to compile DeepC for multiple versions of Nuke which works on linux and windows and creates the compiled plugins in the choosen install folder for every version. The script can be executed running batchInstall.sh, you can then enter the Path to search for Nuke Versions e.g. /usr/local/ or C:/Program Files Optionally an install path can be entered aswell. The script then runs and compiles the plugins (as far as the dependencies like compilers can be found).
We created a repository which includes some example deep render scenes to try/test/use this plugin.
In futur we will add nuke project files to show how the plugins work.
https://github.com/charlesangus/DeepCExamples
Currently, DeepC is maintained by me and Falk Hofmann. I'd love your contributions, though!
If you'd like to contribute, please fork the project make a new feature branch for your contribution. Ideally, also let me know what you're up to so we don't duplicate efforts!
Once you're happy with your work, submit a pull request so I can merge your work back in.
- 4-space hard tabs
- private/protected variable names start with '_' (one underscore)
- variable/function names use lowerCamelCase