|
2 years ago | |
---|---|---|
data | 2 years ago | |
legacy | 2 years ago | |
web | 2 years ago | |
.gitignore | 3 years ago | |
.prettierrc | 2 years ago | |
README.md | 3 years ago | |
index.html | 2 years ago | |
requirements.txt | 3 years ago |
Utility for quickly finding pokemon by the sprite's "distance" from a given color.
See also explanations of the underlying math and which images are used.
ingest.py
generates database.csv
which is needed for nearest.py
,
database-cam02.csv
which is an alternative database for nearest.py
using
the CAM02-UCS color space, and database.js
which is needed for nearest.html
.
Requires numpy, Pillow (PIL), and colorspacious.nearest.py
provides multiple options for finding pokemon "near" a color.
No external dependencies, but a database must be provided. If CAM02-UCS is
being used, this requires colorspacious and numpy.nearest.html
and nearest.js
allow you to use a front-end in your browser,
by opening nearest.html
directly. The implementation is somewhat lazy to avoid
the use of an actual build system, and statically imports the dependencies
Fuse.js, d3-color,
and d3-cam02. Requires database.js
to be present, meaning ingest.py
must be run first, unless you are using the
included database.explore.py
traverses a subset of the 24-bit RGB color space and finds the
pokemon that most closely match each color, and produces best.csv
and
counts.csv
as results.database.csv
, database-cam02.csv
, and database.js
are included as well,
built from the set of pngs detailed in pngs.md.