This repo aims to implement Hierarchical hp-Adaptive Signed Distance Fields. It extends the presented algorithm with GPU implementations, adds LSQ polynomial fitting and an interactive GUI.
- Download the DragonflyPack.zip
- Find a location and create a virtual T drive, eg.:
subst T: C:\ - Unzip the
DragonflyPackthere. There should be 4 folders inT:\DragonflyPack - Download Dragonfly as a submodule within the cloned project (
--recursive) - Use Visual Studio to build the solution
First set up the desired configuration with the sliders and checkboxes.
- Start degree: the degree of the initial polynomials in the octree.
- Max degree: the degree that the polynomials can reach during octree generation.
- Max level: the deepest level the octree can reach during generation. (Uppermost is 0).
- Error threshold: the maximum the error described in the Hierarchical hp-Adaptive Signed Distance Fields can reach. The lower, the more refined and deep the octree is.
- Use h-adapt: use cell subdivision during generation process.
- Use p-adapt: use polynomial degree incremention during generation process.
- ApproxType: what kind of approximation to use
- Gauss quadrature: the method described in the paper
- LSQ: method based on least-squares
- Gauss GPU: the method described in the paper implemented on the GPU
- LSQ: method based on least-squares implemented on the GPU
- Dropwon: which SDF to use (currently set to Sphere on picture)
- The inputs after this are SDF specific and change based on the dropdown setting.
Second, press the Recalculate octree button and wait for the algrithm to finish.
There is a list of octrees that are loaded into memory. Switching between them comes at a very low cost. Simply:
- Choose the desired octree from the list, thus making it highlighted.
- Click Load selected. This changed the currently loaded octree, and inserts the previous one into the list.
Click either the Load SDF or the Save SDF button. A file explorer comes up, where the desired file/location may be chosen.
Check the Evaluation wiki page.
- Dragonfly Poster
- Dragonfly Project
- Or contact the authors at csabix@inf.elte.hu or bundas@inf.elte.hu
Dragonfly is developed by Csaba Bálint and Róbert Bán at Eötvös Loránd University, Budapest, Hungary.
Supported by the ÚNKP19-3 New National Excellence Program of the Ministry for Innovation and Technology. The project has been supported by the European Union, co-financed by the European Social Fund (EFOP-3.6.3-VEKOP-16-2017-00001).
Dragonfly is licensed under the MIT License, see Dragonfly\LICENSE.txt for more information.


