ShotSieve

ShotSieve is a desktop tool for narrowing large local photo folders with learned image-quality scores and practical review filters. It suits photographers and hobbyists who want culling help without sending a library to a hosted service. Its open-source, local-first design is compelling, provided you are comfortable downloading model weights and checking dependency licences before commercial work.

Overview

ShotSieve analyzes photos on Linux, macOS, and Windows, helping you decide which images to keep or reject. The review interface runs on your computer’s loopback address, and the library remains on the machine. That makes it a sensible fit for privacy-conscious photographers, offline-oriented workflows, and anyone handling a large folder without adding a cloud upload step.

The application is free and open source under AGPLv3 or later. That removes subscription cost, but it does not make every component unrestricted: the pinned PyIQA dependency uses the PolyForm Noncommercial licence, and model checkpoints have separate terms that deserve review before commercial use.

Key features

Learned quality scoring

TOPIQ is the default model, while CLIPIQA is intended for quick comparisons and Q-ReAlign Mini provides another option. You can compare models on the same library before committing to a larger culling pass. This is useful when rankings from one model do not match your preferred shooting style, although choosing among models adds a step that fully automated culling products may avoid.

Filtering and batch review

Sort or filter by score, format, resolution, file size, and metadata completeness, then apply batch actions. These controls let you combine an AI opinion with concrete file properties—for example, isolating high-resolution images with complete metadata—rather than accepting a score blindly. RAW support, blur detection, and the ability to work through folders locally make it relevant to photography workflows beyond ordinary JPEG sorting.

Runtime and setup

Packaged releases cover Windows, Linux, and Apple Silicon macOS, with CPU and supported accelerator runtime choices. Accelerator performance depends on matching hardware, drivers, operating system, model, and runtime; CPU is the fallback when that combination is unavailable. Release archives do not include model weights, so preparing or first using a model requires a separate download and storage in the configured upstream caches.

Pricing

There is no free trial because the product itself is free. The trade-off is operational and legal rather than financial: you supply model downloads, meet the runtime requirements for acceleration, and account for the separate terms attached to PyIQA and checkpoints.

Platforms

ShotSieve runs on Linux, macOS, and Windows. Packaged downloads include Apple Silicon macOS builds, and releases offer CPU or supported accelerator runtimes. The default review server binds to 127.0.0.1:8765, keeping the interface local unless you deliberately change the setup.

Who it's for

Choose ShotSieve if you are a photographer or hobbyist working through substantial local folders and want learned scoring, blur detection, RAW handling, and batch decisions while retaining the files on your own computer. It is less suitable if you need bundled models, a managed cloud workflow, or a single licence covering every dependency for commercial deployment.

Pros and cons

• Pros: local processing keeps the photo library on the machine; multiple learned models can be compared; filters cover scores and useful file and metadata properties; RAW support and blur detection target real culling tasks; Linux, Windows, and macOS releases are provided; the software costs nothing.
• Cons: model weights are separate downloads rather than included in release archives; accelerator speed is hardware- and driver-dependent; there are no listed third-party service integrations; commercial users must examine PyIQA and checkpoint terms in addition to the AGPLv3+ application licence.

Alternatives

For a broader category overview, see AI Photo Culling Software. Pick Zno instead if you want a freemium product spanning Android, iOS, macOS, web, and Windows, with a free start and optional paid upgrades. Facet is the closer choice for a free local workflow explicitly offering no cloud, accounts, or subscriptions across desktop, web, and self-hosted setups.

PixCull suits readers who prefer MIT-licensed open source and optional API or cloud model calls, while Project Kestrel offers a free desktop application with Windows and macOS downloads plus a Linux-from-source route. Choose Optyx for Lightroom compatibility and paid Pro billing, or AccioPix Photo Culling for a browser tool with no sign-up and no photo limit beyond your computer’s capacity.

Futura Photo is an alternative with a free Community tier limited to 100 images per analysis and one concurrent installation. Kepla is another option for macOS users, offered through a private preview.

Verdict

ShotSieve is a strong choice for photographers who want free, local AI-assisted culling with model comparison and detailed file filters. Look elsewhere if you need integrated cloud services, bundled weights, or simpler commercial licensing across the whole stack.