略读预计 2 分钟

Amélie (2023)

摘要

这是一个名为 Am\xe9lie 的自拍打印机艺术装置，用户按下按钮后，摄像头拍照并通过热敏打印机即时输出照片，照片会随时间褪色。作者分享了三个版本的技术迭代：v1（2018 年，树莓派 3B+、Adafruit 热敏打印机、树莓派摄像头 V2）、v2（鞋盒外壳，增加 USB 存储和配置文件）、v3（2023 年，树莓派 4、摄像头 V3，支持自动对焦和实时预览，透明亚克力面板）。代码用 Python 编写，使用 OpenCV 库，v3 的源码和零件清单已在 GitHub 开源。

荐读理由

虽然这不是AI工程或创业内容，但文中给出了完整的硬件清单、Python代码结构和版本演进思路，对想快速搭建实体交互装置或学习树莓派外设集成的开发者有直接参考价值，可迁移到类似项目。

原文

Amélie

Amélie is a selfie printer that prints photos on cheap thermal paper, reminiscent of old instant photos. The user can tear the print off and take it home, as a souvenir. Over time, the print fades, leaving behind a white piece of paper. The installation combines the ordinariness of a receipt printer with the singularity of the captured moment.

☥

This installation is named after the movie of the same name from 2001. All the components are built into a transparent plexiglass panel, so the user can see the individual modules that make up the installation. The code is written in Python and runs on a Raspberry Pi 4, with a thermal receipt printer and some additional components, like a screen, a camera, and a button with a built-in LED ring. After pressing the button, the LED ring begins to flash and a countdown appears on the screen; when the countdown reaches 0, a photo is taken and printed instantly. Optionally, a small illustration or text segment can be printed above or below the photo. For New Year's Eve, a random text from a large list of fortune cookie sayings gets picked and printed below the photo, so a New Year's prediction accompanies the selfie. Because the images are printed on thermal paper, no ink is required.

v1 and v2 were built in 2018 with a Raspberry Pi 3 B+, an Adafruit Mini Thermal Receipt Printer, a Raspberry Pi Camera Module V2, a small Pi monitor, various buttons and cables and stuff. Coded in Python with the OpenCV Library. Version 2 got a shoebox enclosure with a status LED and some additional functionality, like saving all images to a usb thumb drive and changing the configuration via a config file. Version 3 was built in 2023 and runs on a Raspberry Pi 4. It uses the Raspberry Pi Camera Module V3, this allows for faster live preview as well as autofocus and better exposure measuring. It also has a really nice button with LED ring and is built into a plexi panel.

The source code and parts list for v3 can be found on GitHub.