Releases & Roadmap

• Added editable HDR bracket fusion and native transparency-based layer splitting.
• Added document color grading, layer color matching, shadow extraction, and improved artboard workflows.
• Added experimental Gemini Web support for Nano Banana 2 using a signed-in Chrome profile.
• Improved assistant reliability, context handling, transparent references, thumbnails, chat persistence, and composer input.
• Strengthened native builds and release packaging across macOS and Windows.

This maintenance release improves reference-based generation and Photoshop capture reliability.

Improvements

• Fixed missing or incorrect references during retries, continuations, batch processing, and multi-document generation.
• Improved Photoshop reference captures across 8-bit, 16-bit, and 32-bit documents while preserving transparency and framing.
• Prevented active selections from unintentionally narrowing whole-document visual analysis.
• Improved GPT Image output size and aspect-ratio adherence by reinforcing target dimensions.
• Added clearer errors when reference images cannot be saved.

• Added an auto-fitting prompt box with manual resizing and quick reset.
• Added recent prompt history with three quick shortcuts and a +N list for older prompts.
• History now shows timestamps, generation details, and reference previews.
• Reusing a prompt restores its settings and references without automatically starting generation.

This release includes:

• Support for Seedream 5 Pro.
• GPT‑5.6 for the AI Agent.
• New Storage Cleanup controls for managing generated images, chat history, and cached data.

This release adds multi-output image generation: one request can now produce multiple photos, with streamed previews appearing sooner while results are still generating.

Highlights

• Added support for 1 request → multiple generated images
• Improved streamed image-set previews and thumbnail stability
• Added provider upload caching for repeated image runs
• Hardened generated image placement in Photoshop, including better support for CMYK, Lab, Grayscale, RGB 16-bit, and RGB 32-bit documents
• Improved generation state handling, logging, and best-effort result summaries
• Fixed image engine defaults and Custom Codex routing issues

• Redesigned Settings for clearer provider and model setup.
• Added API key support for GPT Image 2 and Grok.
• Added Nano Banana 2 and Nano Banana Pro support through Google AI Studio and Vertex AI.

• Improved GPT Image 2 reliability by routing ChatGPT image generation through the hosted Codex Responses flow.
• Updated the reference image add slot to use a clearer plus icon.

• Restored the attach + button in the composer.
• Improved long permission labels so they no longer crowd the input area.
• Hid voice controls when they are not available.

• Added web image reference search in the assistant, with inline previews and one-click use as reference images.
• Added clearer assistant progress for longer research and visual tasks, including task-plan closeout and visual-report tracking.
• Improved Codex/GPT Image 2 reliability with auth recovery, startup recovery, and fallback handling.
• Improved chroma key controls with better sampling, subject-mask support, despill, and matte verification.

• Fixed a bug where changing text fonts could fail.
• More reliable batch processing with durable journals, rolling item windows, item-level retry, and safer progress tracking.
• Better output handling: generated results are preserved, reused when possible, and saved into the requested folder with stable names.
• Safer Photoshop execution with serialized host writes, cleaner document lifecycle handling, and sanitized document names.
• Release packaging now uses the Adobe UXP Developer Tool generated .ccx, reducing install/package compatibility risk.