Required skills: Python, maybe Cython

Difficulty: Medium

The openage converter is able to optimize the compression for created PNG files by utilizing a search algorithm inspired by OptiPNG. To do this, the search algorithm first tries 8 different compression settings by creating PNGs with these settings in-memory. The compression settings that produced the smallest PNG are then used to export the image data into a PNG file. Using the optimized compression usually results in a much smaller file size (~50% of unoptimized size). However, the optimized compression is currently much slower as the resulting PNG has to be created 8 times (instead of 1 time with unoptimized compression) before it can be exported.

The way we can speed this process up is to save the best compression settings for each source image in a configuration file and use them as input during the PNG export phase. As a result, the PNG exporter can skip the 8 trials of the search algorithm and directly generate the "best" PNG. Furthermore, using optimized compression would then be as fast as using unoptimized compression.

To do this, we have to do the following:

1. Save the best compression settings for each source image in a configuration file
2. Load the file for media conversion and forward the compression settings to the PNG exporter
3. Use the settings during PNG export (this is already implemented)

For the configuration file, TOML should be used. The compression settings should be accompanied by information to identify the source image file (i.e. filepath and a SHA-256 hash value).

Example:

# Header
# The version of this config file format
file_version = "1.0"
hash_algo = "SHA3-256"

[images]
    [2.slp]
    "filepath" = "graphics/2.slp"
    "hash" = "DEDF4675668799912378178479281"
    "settings" = [4, 5, 6, 7]

    [1080.slp]
    "filepath" = "graphics/1080.slp"
    "hash" = "79244847392374923749273497237"
    "settings" = [4, 5, 6, 7]

Further Reading

• PNG exporter code
• Converter Architecture Overview
• Converter Workflow