I built an image toolbox that keeps common processing inside the browser

In the daily grind of working with product images, a common workflow seemed both frustrating and inefficient. Numerous separate sites were required for tasks such as compression, format conversion, background removal, and even animated images. Most of these tools demanded the original file to be uploaded, causing inconvenience. Frustrated by this cumbersome process, a new solution was devised: Imging, an image toolbox designed to keep common processing operations within the browser itself.

Imging eliminates the need for multiple sites by handling common image work entirely on the device. This includes converting between JPG, PNG, WebP, AVIF, and HEIC formats, compressing images, removing backgrounds, and even editing animated images such as GIF, APNG, and animated WebP. Larger codecs and models are loaded on-demand, only when a specific feature is needed, optimizing resource usage.

The development of Imging revealed that "runs in the browser" is not a single, uniform capability. Various tasks, such as decoding, canvas processing, encoding, WebGPU inference, WASM fallback, memory limits, and file saving, all operate independently. Each capability had to be tested thoroughly rather than assuming support based on the browser's name.

Background removal, for instance, prefers WebGPU processing but falls back to WASM if necessary. Animated images follow a frame pipeline, allowing for precise timing and loop information to be maintained while frames are edited or recompressed.

To handle PDF compression, the tool rewrites the document structure rather than converting every page into a large bitmap, ensuring a more efficient process. Despite the progress made, Imging still has its honest limitations. Some uncommon formats cannot be fully handled within the browser, mobile devices may struggle with large animations due to limited memory, and HEIC files can sometimes be inconsistent.

To inform users about these constraints, Imging provides clear error messages instead of falsely claiming that all operations can be performed locally. The tool is available for free use without the need for an account or any watermarks: https://imging.cn/en/. During its development, the author took advantage of AI-assisted coding, but the core functionality—such as capability checks, processing boundaries, and validation cases—stemmed from the actual needs of a product-image workflow.