翻訳待ち:Show HN: Pictiur: Image converter, optimizer and resizer, all in one
AI サービスが一時的に利用できないため、復旧後に翻訳を補完します。ソース概要:Disclaimer: This project was entirely vibe coded, this was done by a professional, do not try to reproduce this at home. I also fully understand that this may not match some people's values, so I'd like to point out tha…
AI サービスが一時的に利用できないため、復旧後に翻訳を補完します。
Disclaimer: This project was entirely vibe coded, this was done by a professional, do not try to reproduce this at home. I also fully understand that this may not match some people's values, so I'd like to point out that I built this project for my personal use first and foremost. If my approach still bothers you, well, that's ok. For personal reasons, I've had to do document scans every day for the past month, photos of my ID papers, and I realized something: the online offering for processing images is horrible. Generally these tools focus on one thing, conversion, optimization, or resizing, but rarely 2 tools in one, and even less 3. It could pass if, on top of that, the offering wasn't riddled with sites full of dumbass ads, with "DOWNLOAD" buttons that are a trap, and with data processing that's dubious at best. I've also been running a home server for a while now, on which I install lots of self-hosted projects (with Coolify <3), and I wondered whether I couldn't make my own tool, one that would work completely offline (PWA), and wouldn't transmit any data. I also wanted to see what Deepseek Flash was capable of, beyond the last questions I had asked it, namely how to set up my vegetable garden on 100m², and whether I should put my rainwater collector on a stand or not. So I dove into a benchmark on the best algorithms to use for my idea, how to architect the app, etc... In the end, after about an hour of back and forth, I had a complete implementation plan and a tech stack I already knew, in case I had to debug by hand and use my brain again. A few hours in the IDE and here is the final result: https://pictiur.pages.dev/ The site is 100% local, the whole process runs in the browser, it's a PWA so the app is installable and usable offline, and for now it supports the most well-known formats but I fully intend to add more. The repo is public in case you want to install it on your own server (and feel free to take the opportunity to trash talk the code quality and explain just how much of a disaster AI is, and how your nephew who's currently learning PHP would have structured this in a much more professional way). https://github.com/Etheonor/pictiur And feel free to tell me if you see any improvements to make. As I said at the beginning, it is built for my personal use, but if other ideas can be grafted onto it, I'm all ears.