kobejohn / polymaze Goto Github PK

A demo proving it is possible to interact with the grid + background image in real time.

For example, as the mouse moves, it changes the properties of the areas/edges, clicking changes the alpha value, etc.

I have the most uncertainty here about what approaches might work. I want it to be performant enough to produce images at video rate (make and display 1530 changes per second).

The time limit needs to apply to the combination of both making changes and producing the final image.

Tentative idea for the app - a whole image is turned into a maze (not sections) but only the current "region" you are in (for example, a letter if it's text) is visible. So the giant maze is effectively partitioned into interconnected mazes where you can never see the whole thing at one time. Maybe you can drop bread crumbs to track your way (ala Labyrinth).

It should be possible to transparently remove the PIL image generation system (it's kind of clunky and generates inflexible huge images) with vector drawing, flexible output and more beautiful output of cairo using cairocffi.

This is the key step that is doesn't make sense with the polymaze-focused implementation. All original image information has been lost by the time a grid reaches the polymaze module. Since the amount of implementation required to support this is non-trivial and it should be reusable, I judged it's worth moving it while implementing it into the core polygrid system.

This is a request from a user. Thanks for the suggestion!

Originally I wanted the output to be SVG but didn't have time or knowledge to do it when I made this. It's the ideal format for people who know what they are doing.

PNG is still the best output for the average user since it retains clean lines and has excellent compression due to the regular nature of the maze.

This first step is probably the biggest so I'm making it very simple. The current system already has visualization for the grid, but it is done in the polymaze module rather than in the core (polygrid). This is not good because polygrid is fundamentally a visual system so any other types of visualization would be mostly duplicate code violating DRY (don't repeat yourself).

The first step is to move the basic visualization into a new visualizer class that enhances the base PolyGrid class.

The target is eventually to allow polymaze to behave exactly as it does now but also enable interactive image overlays which are impossible with the current polymaze-focused implementation.

It might incredibly inflate file sizes, but if it looks good, it would be possible to reduce maze sharpness to make it feasible.

Instead of setting a value to the whole grid, allow color/alpha to be applied to each edge individually.