Analysis updated 2026-05-18
Quickly cull hundreds of near-duplicate burst photos down to the best shots.
Automatically filter out blurry, over/underexposed, or closed-eye photos before manual review.
Batch-watermark selected photos with camera and lens EXIF data after culling.
| zhaoyue4810/pianke | kouhxp/yapsnap | s0912758806p/agentic-sop-to-work | |
|---|---|---|---|
| Stars | 169 | 167 | 173 |
| Language | Python | Python | Python |
| Setup difficulty | moderate | easy | easy |
| Complexity | 2/5 | 2/5 | 3/5 |
| Audience | general | developer | ops devops |
Figures from each repo's GitHub metadata at analysis time.
Expert mode downloads about 600 MB of AI models on first use, Tycoon mode requires an external AI API.
Pianke, or "片刻" in Chinese, is a local photo culling tool for photographers. When you shoot a burst or take many similar shots of the same moment, you end up with dozens or hundreds of nearly identical images. Pianke groups those similar photos together and then presents them two at a time in a side-by-side comparison, so you can quickly press a left or right arrow key to pick the better one. The winning photo advances and the discarded one goes to a reject folder. The tool runs entirely on your computer and opens as a web page in your browser. Three modes are available. The Fast mode uses lightweight local algorithms and works on any hardware without downloading anything extra. The Expert mode downloads about 600 MB of AI models once and then runs locally, it is better at grouping photos of people because it uses face recognition to identify the same person in different poses or backgrounds. The Tycoon mode sends thumbnail images to an external AI service such as GPT-4V or a compatible API, in return, the model explains rejections in plain language, for example by noting that a child in the left side of the frame had their eyes closed. Before the head-to-head comparisons begin, the tool automatically filters out technically bad shots: blurry images, overexposed or underexposed frames, and photos where a subject's eyes are closed. You can review any automatically rejected photo and bring it back into the selection before proceeding. RAW files are supported. If you shoot RAW plus JPEG pairs, the tool displays the JPEG for speed while keeping the RAW file paired to it during moves and renames. For pure RAW shooting, it extracts the preview thumbnail embedded in the file rather than doing a slow full decode. Progress is saved automatically to your photo folder, so you can close the app at any time and resume exactly where you left off. A multi-step undo lets you reverse individual decisions or redo an entire group. After culling is finished, a watermark feature can batch-stamp your selected photos with EXIF data including camera model, lens, shutter speed, aperture, and ISO, with brand logos for Fuji, Canon, Nikon, Sony, Leica, and others.
A local photo culling tool that groups similar burst shots and lets photographers quickly pick the best one with side-by-side comparisons.
Mainly Python. The stack also includes Python, Web UI, RAW.
Not stated in the README.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly general.
This repo across BitVibe Labs
Verify against the repo before relying on details.