Skip to content
Switch branches/tags

Latest commit


Git stats


Failed to load latest commit information.

Visibility/light raycasting algorithm

Visibility/light-raycasting algorithm python implementation inspired by:

It is a simple window-mode presentation of algorithm. It opens arcade window to draw some poligons (obstacles for light or visibility) and user can move position of "light" source.

It could be easily changed into the FOV script, by diminishing value of arc_angle attribute in the Light class (curently it is 360 degrees to simulate point-source of light, eg. lightbulb hanging from the ceiling).

Installation, running and use instructions:

  1. Download or clone this repository.
  2. Setup new virtual environment with Python 3.6.
  3. Install requirements with pip install -r requirements.txt
  4. Run in any IDE, e.g. PyCharm, or in Terminal.
  5. First, you will see configuration screen with some variables.
  6. You can profile the code using cProfile moule, by setting variable PROFILLING to True. Menu You can change them to set-up your simulation specifications. You can change size and shape of obstacles. For exampole. if you set edges to 3, and size to 150 you will get 4 triangles: Triangles

Setting edges to 6 and size to 50 would produce: Hexagons

  1. Then click "Run" button, and the simulation would begin. You can move "light" source on the screen using your mouse. Left-click allows you to left the light at the current position. Another left-click would pick it up.

With random lights colors and 8 lights: Random colors

It requires more optimizations:

  1. For now it works only for static obstacles - positions of polygons vertices are calculated once at the start. Dynamic updates of obsatcles-corners would cost a lot of computing power which is already entirely consumed by raycasting.
  2. It requires the enclosed FOV (it must treat screen edges as outermost 'walls' encapsulating the scene)
  3. It does not scale-up: with 9 squared-obstacles on the screen it calculates all raycasts for 1 origin in 0.0029 to 0.0105 second, which gives 95-345 FPS. With more obstacles and lights, framerate falls quickly. But I made some progress, by getting rid of shapely library calls with simplier math calculations.
  4. To change it to FOV-simulation, you need to add half-arcs of visible space which would connect raycasts andings. For now, visibility-polygon is cut straightly. Unless your visibility range is larger than screen size, you will get glitches.
  5. Maybe some work could be done with multithreading or multiprocessing, but I did not work it out yet.
  6. Unfortunatelly, my algorithm suffers for the same problems which Amit Patel from Red Blob Games blog reports: robustness problems which causes generation of some invalid, falsely-dark triangles. I was only able to get rid of small part of them. I suppose, there is no easy way to fix this, and I can only send you to the same source about this problem, that Amit reccomends:
  7. Maybe I shgould use another algorithm - triangle-expanding algorithm described in the PDF file added to this repository.

To be done:

  1. Optimizations listed above.
  2. Add option to move position of obstacles when app is running.
  3. Replacing single-colored "lit" polygon with gradient to simulate diminishing power of light. It could force me to replace arcade library with something else, since arcade does not allow draw gradient-polygons in real-time.
  4. Adding this project to my Django blog (when it is also finished).
  5. You can change LIGHTS_COUNT variable to add additional sources of light/FOV shadows effect (although I did not implement alpha transparency so rathar than shadows, you have sharp color-change). Beware of lights number, since it is computation-heavy.
  6. You can randomize colors of each light to better see that there are many, distinct FOV polygons drawn. Set RANDOM_COLORS to True to get random colors, otherwise a visibility/light polygon would by rgb(192, 192, 192).

If anyone has ideas how to make this algorithm better (faster), please, notice me, fork and make pull-requests.

Rafał "Akapkotel" Trąbski


Raycasting algorithm python implementation inspired by:







No releases published


No packages published