v1.0.5

PhotoMapAI v1.0.5

This version adds the following significant features:

1. The option to select among three image/text encoders that have different levels of speed and accuracy. See Encoders
2. Integration with the InvokeAI image generation package, allowing you to create albums from InvokeAI image collections, send previously-generated images to a running InvokeAI backend to recall and remix them, and to use an arbitrary image as an InvokeAI reference image. See InvokeAI Integration.
3. Reorganized Settings dialogue, with the ability to control whether search results wrap around or not.

In addition, there are a number of internal bug fixes, performance enhancements, and security patches. See the next section for details.

Installation

For current users, you may do an online upgrade to this version from the About dialogue. Click on the "?" icon to start.
New users see the Installation Guide for a variety of install methods. Executable installers for CPU-only systems can be found at the bottom of this page. Additional installers for Windows and Linux users who have GPU cards can be found here

What's Changed

• Fix artifact upload: create archives before uploading to GitHub Release by @Copilot in #215
• refactor(metadata): finish wiring invoke formatter to pydantic union by @lstein in #216
• feat: support InvokeAI authentication by @lstein in #217
• fix(metadata): handle unknown InvokeAI metadata fields by @lstein in #218
• feat: Improve umap window size and position by @lstein in #219
• feat(invoke): "Use as Ref Image" button + resilient auth fallback by @lstein in #220
• feat(invoke): show "Use as Ref Image" button for any image when InvokeAI is configured by @lstein in #221
• feat(invoke): add image-board selector for reference uploads by @lstein in #222
• feat(invoke): reuse existing InvokeAI images instead of re-uploading by @lstein in #223
• fix(security): close three HIGH-severity API issues by @lstein in #224
• docs: add InvokeAI Integration user guide page by @lstein in #225
• feat: pluggable image/text encoder layer with OpenCLIP and SigLIP backends by @lstein in #226
• feat: offload idle encoder from VRAM to RAM after configurable timeout by @lstein in #227
• fix(frontend): suppress slide animation when entering grid view by @lstein in #228
• fix(frontend): rebuild swiper after single-image delete by @lstein in #229
• feat(frontend): add wrap-around navigation setting by @lstein in #230
• feat(frontend): show spinner during album switch by @lstein in #231

Full Changelog: v1.0.4...v1.0.5