Standalone Database

[Neil Harding]

I've been writing a standalone comic database using the GCD database to automatically identify issues when possible from the filename. It looks for the series name and if there are multiple possible versions it will give you a dropdown, but if it can it will get the data from the GCD database dump. I extract the cover image, but I also take the phash for each cover, and it seems to work well at identifying other copies of the same cover. I have about 200,000 files so once they are all indexed would you like the phashes? I am using Sqlite, Python 3.11, Pony, Bottle, OpenCV, TurboJPG, Scikit Image, Rarlib, 7Z, PDF library, if anyone is interested a copy, and once everything is imported it can also use filehash to identify copies of scans.

[Rich Watkin]

Hi Neil,

Your work extracting the covers and calculating perceptual hashes sounds extremely relevant to something we are building.

Simplist identifies comics from user-uploaded photographs. We currently compare full-cover and regional perceptual hashes, including quadrants, so your approach appears closely aligned with ours. Using a hash reference rather than repeatedly retrieving cover images would also fit GCD’s preference to avoid unnecessary image traffic.

Would you be open to discussing access to the hash data for testing? It would be useful to understand:

- how each hash maps back to a GCD issue or cover ID;
- the precise hash algorithm, image preparation and quadrant layout;
- the current coverage and how updates are handled; and
- whether use and caching within a commercial identification application would be permitted.

We would be happy to credit GCD clearly as a data source and work within any reasonable usage requirements.

Thanks,
Rich
Simplist / Bear Maximum