[Liuser]

Thank you -- I'm going to revisit this suggestion as it's been brought up by friends as well.

I actually implemented a mobile web version at first with the up-sell of buying high res images. I wasn't happy with the images it was producing though because I forced 1 setting for all images being converted for the sake of ease-of-use for the user.

But this was a mistake. Not all photos are the same: Some photos looked better with smaller/larger cells, as a diamond grid, with different color settings, etc.

Server costs and complexity grew too: (nginx, nodejs, rabbitmq). The algorithm is CPU intensive and not fast enough to my liking yet. So I released the macOS version and put the computational cost on the end user and avoiding the need to support servers.

Sounds like I should make a native mobile version with the option to tweak parameters.

[eigenvalue]

Or just pre-compute low-res previews using a variety of settings and have the user highlight ones that look pleasing. Over time, you can collect this data and find heuristics (i.e., machine learning) for automatically choosing the most pleasing settings for a given image.