That would make a lot of sense. It appears that if you want to feed in an image manually rather than using Vision the image must be scaled to a specific size. I bet that’s what Vision is doing behind the scenes anyways, so some quality is definitely lost when it scales down the images. Not sure if this would make an image captured at a lower resolution inherently better or not.

I'm a Harvard student, maker, and radio enthusiast. Check out my book on radio communications at and my website at

Love podcasts or audiobooks? Learn on the go with our new app.