My aunt runs a small side business selling handmade figures, which means she takes a lot of photos. I became her photo editor, and the job was always the same: fix dark images, remove background clutter, find the best shot from a group of duplicates, and get them ready for her online store.

The manual work was slow. It wasn't just photo editing, it was also getting the images ready for the web. I had to resize and compress them, then create SEO-friendly filenames (slugs) and product descriptions to help her store show up in search results.

I started with Photoshop and Lightroom, but the process was slow. I tried open-source tools like GIMP and wrote my own scripts with ImageMagick, but it was still not fast or easy enough. I needed one tool for everything: batch processing, AI-powered improvements, image optimization (compression + resizing), and generating SEO content. So, I built BatchShots.

Here's a quick look at how it was built:

- I began with the backend, experimenting with various open-source models for identifying duplicates, removing backgrounds, and generating SEO-friendly text. I chose to use ONNX to run small models directly in the browser.

- I used opencv.js for the image editing features, all the standard tools you would expect.

- As a backend developer, I've vibe-coded the UI components :) After a lot of tries, I got a UI that I liked.

A few things I learned:

- Localization sucks, it just makes everything more complicated

- Always double-check the model license

- OpenCV.js is complex: At first, I added too many features and had to remove some to make the tool easier to use

- Mobile is hard: Running AI models in a mobile browser took a lot of work to optimize and stop it from crashing

Most of BatchShots is free. If you need to edit more photos at once or want all the AI and SEO features, there's a one-time payment of $19.99.

I'm here to answer any questions and would love to hear your feedback.