PhotoCraft review: a transparent look at the alpha

We do not present this as an independent lab test, do not publish a self-awarded star rating, and do not claim hands-on testing of every competing feature discussed elsewhere on the site. External product claims link to the relevant vendor or project source.

The purpose is brand defense against misleading third-party pages that sometimes rehost or misrepresent PhotoCraft, while giving buyers the limitations they need to make a fair decision.

This is not a paid independent review. We did not fabricate third-party review scores, star ratings, or hands-on conclusions about competing applications. Where behavior has not been verified, it is described as a limitation or test requirement.

The review is best read as a transparency page: it explains what the project claims, how it is tested, what remains missing, and which users should wait. Buyers should still run their own representative file and workflow test.

The editor is written in Rust and uses a native GPU compositor across Metal, Vulkan, DX12, and WebGPU. The same command registry powers the UI, CLI, JSON control channel, and MCP server, which makes repeatable and agent-driven editing possible.

The current release is v0.18.0, with macOS, Windows, Linux, FreeBSD, and Web builds. The macOS DMG is universal for Apple silicon and Intel and is signed and notarized.

The project states clearly that re-saved files are not byte-identical. It rewrites image resources, layer records, and the composite, while carrying through data it does not model where possible. Users should test the features they actually edit, not only whether the file opens.

• Not every Photoshop feature behaves identically, even when a command is present.
• Photoshop plug-ins and Adobe-first team integrations should not be assumed compatible.
• Complex text, print production, and specialized professional workflows need testing.
• Back up important documents and validate every release used in production.

The project’s parity assessment identifies generative AI, plug-ins, typography depth, professional workflows, and missing tools as significant gaps. Some commands are present while particular modes, dialogs, integrations, or behavior remain incomplete.

A production evaluation must exercise the exact task. Opening a layered PSD is different from preserving every smart object, effect parameter, text behavior, and color-management decision through an edit and save.

Do not generalize a benchmark from one machine to every Mac or Linux workstation. Test the largest representative document, the filters used in the workflow, and pen input on the target hardware.

Assign an owner for release updates and retain the prior installer. Record the exact release version with each test because alpha behavior can change quickly.

• Week 1: install from the official release, verify the checksum, and reproduce a simple layered document.
• Week 2: edit a complex PSD with text, masks, effects, smart objects, and the expected color mode.
• Week 3: complete a real deliverable and compare editable files, rendered output, and export settings.
• Week 4: document failures, workarounds, support needs, and whether the editor is safe for the intended scope.

• Good early fit: open-source advocates, Linux users, PSD experimenters, automation-focused users, and people editing local layered documents.
• Wait or keep Photoshop: Adobe plug-in users, large collaborative teams, generative-AI-heavy workflows, and production pipelines that cannot tolerate alpha behavior.
• Consider GIMP or Krita: users prioritizing mature FOSS breadth or painting-first workflows.