OpenLayer v0.6.0-alpha

OpenLayer v0.6.0-alpha
UI polish pass, architecture cleanup, reliable inpaint finishing, and outpaint canvas expansion.

v0.5.5-alpha

v0.5.5-alpha - 2026-07-11
Correctness and trust release: working inpaint uploads, safe cancel, aligned inpaint context, and honest status copy.

Fixed
Fixed ComfyUI uploads losing their filenames in Photoshop UXP. UXP's FormData saved every upload as a file named blob, so the Inpaint mask upload overwrote the source upload and both LoadImage nodes read the mask. Uploads now build the multipart body manually with an explicit filename per file, and inpaint-basic composites over the real captured source again.
Fixed the Photoshop "Make is not currently available" alert during Inpaint Import to Layers. Photoshop's place command clears the active selection, so the layer mask import now restores a selection from the captured Inpaint bounds and falls back to aligned context import cleanly when no selection can be restored.
Fixed inpaint misalignment caused by ComfyUI VAE rounding: the captured inpaint context now snaps to multiples of 8 so the generated result matches the captured context size exactly.
Fixed multipart encoding to work without TextEncoder, which Photoshop UXP does not expose.
Added
Added per-preset recommended generation defaults. txt2img-basic now starts at 20 steps, and switching workflow presets applies that preset's recommended steps and CFG/guidance values.
Added queue-aware Cancel Generation: prompts still waiting in the ComfyUI queue are removed with a queue delete instead of interrupting whatever job is currently running, and diagnostics report whether the prompt was dequeued or interrupted.
Added a 16-megapixel capture guard so very large layer/canvas/mask captures fail fast with a friendly message instead of risking UXP memory failures.
Added the clipboard and launchProcess manifest permissions used by Copy Diagnostics, Copy Prompt, and the footer links.
Changed
ComfyUI WebSocket preview frames are now received as arraybuffer for UXP reliability. Live KSampler step previews appear in the result preview when ComfyUI is started with --preview-method auto.
The Outpaint and Sketch to Image tool cards now show the Experimental badge to match their preset status, and Z_image_Turbo warnings plus the hardware advisor no longer claim those presets are disabled.
Bumped plugin/package metadata to 0.5.5.
Known limitations
Live sampler previews require ComfyUI to be started with --preview-method auto, and the preview panel may flicker between steps until the planned UI polish pass.
Capture is limited to 16 megapixels (4096 x 4096) until a downscale option is added.
Inpaint and Outpaint remain experimental and should be tested on duplicate layers or disposable documents.

OpenLayer v0.5.3-alpha — Public Alpha Release Checkpoint

🦖 OpenLayer v0.5.3-alpha is a public alpha checkpoint for the open-source Photoshop UXP plugin that connects Photoshop to local ComfyUI workflows.

This release is for testing, not production work yet.

Ready for testing

• Text to Image
• Image to Image
• Z_image_Turbo presets
• Experimental Flux1-dev fp8 Text to Image
• Prompt from Layer with local Florence-2 PromptGen setup
• Upscale with local pixel/model upscale models
• Workflow Health diagnostics
• Cancel Generation
• Session History metadata
• Flux Fill Inpaint testing
• Flux Fill Outpaint testing

Experimental / known limitations

• Inpaint remains experimental.
• Outpaint remains experimental.
• inpaint-basic may not match the source as reliably as the Flux Fill path.
• Prompt from Layer requires local Florence-2 custom nodes and model files.
• Upscale is pixel/model upscale only.
• Custom workflow import, LoRA browser, batch variants, ControlNet panels, and true persistent Photoshop layer metadata are future work.
• CI does not run Photoshop, UXP Developer Tool, or ComfyUI integration tests.
  🦖🦖

Package

Attach:

openlayer-v0.5.3-alpha.zip

Suggested tester flow

1. Start ComfyUI on http://127.0.0.1:8190.
2. Load dist/manifest.json in Adobe UXP Developer Tool.
3. Open Photoshop and launch OpenLayer.
4. Run Check ComfyUI and Workflow Health.
5. Test Text to Image, Image to Image, Prompt from Layer, Upscale, and Cancel Generation.
6. Test Inpaint and Outpaint only as experimental workflows.

OpenLayer v0.4.6-alpha

v0.4.6-alpha public release candidate

OpenLayer v0.2.1-alpha - Sketch to Image LINECN Preview

OpenLayer v0.2.1-alpha adds the first working Sketch to Image foundation for Photoshop and ComfyUI.

Highlights:

• Text to Image with txt2img-basic
• Image to Image with active layer/canvas capture
• New Sketch to Image LINECN starter workflow
• New sketch2img-linecn-basic workflow preset
• Source preview, result preview, and import to Photoshop layers
• Friendly setup validation for required LineArt/ControlNet nodes and models
• Updated README, landing page, changelog, setup docs, and tester checklist

Sketch to Image requirements:

• Recommended checkpoint: epicrealism_naturalSinRC1VAE.safetensors
• Required ControlNet: control_v11p_sd15_lineart_fp16.safetensors
• Required nodes include LineArtPreprocessor, ControlNetLoader, and ControlNetApplyAdvanced

Known limitations:

• Sketch to Image currently targets one SD 1.x LINECN workflow.
• SDXL, SD3, Flux, and Z-Image Sketch workflows need dedicated future presets.
• Image source capture currently uses Photoshop UXP Imaging API JPEG capture.
• Inpainting, masks, upscaling, PNG layer export, and selection-aware workflows are still planned.

Package:
openlayer-v0.2.1-alpha.zip

SHA256:
8DF7D8BFF28A42E4ED49437E818DDD8EA49969590A5DF2FF14985AF69E7F22C3

OpenLayer v0.2.0-alpha - Image to Image Foundation

OpenLayer v0.2.0-alpha introduces the first Image to Image foundation for the local Photoshop + ComfyUI workflow.

Highlights:

• Photoshop-dark Home dashboard with tool cards
• Text to Image workflow with preview and layer import
• Image to Image screen behind the launcher
• Active layer capture for Image to Image
• Canvas capture for Image to Image
• Source preview before upload to ComfyUI
• ComfyUI image upload support
• Experimental checkpoint mode for SD3/Flux-style models
• SD 1.x, SDXL, SD3, and Flux compatibility warnings
• Session History and Settings views
• Responsive panel spacing fixes for narrow and wide Photoshop panels

Known limitations:

• Image to Image is still an alpha foundation.
• img2img-basic is safest with SD 1.x and SDXL checkpoints.
• SD3, SD3.5, and Flux usually need dedicated future workflow presets.
• Source capture is currently JPEG through Photoshop's UXP Imaging API.
• Inpainting, masks, selection workflows, upscaling, and advanced alignment are not included yet.

Package:
openlayer-v0.2.0-alpha.zip

SHA256:
57DAA1FFB15364896222E815E648B686249CA762AE004BC9211405664028322D

OpenLayer v0.1.10-alpha

OpenLayer v0.1.10-alpha is a pre-release focused on usability and reliability for the current text-to-image MVP.

Highlights

• Added persistent local settings for ComfyUI URL, workflow, checkpoint, and generation defaults.
• Added a passive "Find ComfyUI Active Port" helper for common local ports.
• Added session History for recent generated previews.
• Added optional auto-import after generation.
• Added collapsible Negative Prompt controls.
• Improved Text to Image screen flow, Settings diagnostics, and Photoshop-dark panel polish.
• Preserved the existing local ComfyUI text-to-image generation and Photoshop layer import workflow.

Known Limitations

• Text-to-image only.
• Image-to-image, inpainting, masks, layer export, upscaling, and advanced workflows are planned for later versions.
• Very narrow Photoshop panel layouts are usable but still need more UI polish.
• Custom ComfyUI workflows may require node ID updates in the workflow builder.

Package

openlayer-v0.1.10-alpha.zip

SHA256:
01044903751EE86EA500BA3E52F60851F281C8E45411CB41435BDB93271541A5

OpenLayer v0.1.8-alpha - MVP Preview

First public alpha preview of OpenLayer, a local-first Photoshop UXP plugin for connecting Photoshop to a local ComfyUI server.

What works

• Connect to local ComfyUI from Photoshop
• Load available checkpoints
• Generate images with the txt2img-basic workflow
• Preview generated results in the OpenLayer panel
• Import generated output into the active Photoshop document as a new named layer
• Uses the official OpenLayer icon and early Photoshop-dark panel UI

Requirements

• Adobe Photoshop 2024+
• Adobe UXP Developer Tool
• Local ComfyUI running at http://127.0.0.1:8190
• Node.js 18+ if building from source

Known limitations

• This is an alpha MVP, not a production release
• Only text-to-image is supported
• Custom ComfyUI workflows may require node ID edits
• Image-to-image, inpainting, masks, selected layer export, upscaling, and advanced workflow presets are not included yet

Package

Upload asset:

openlayer-v0.1.8-alpha.zip

SHA256:
0042FC9F46C29D738C0B686CF4EE7FF71B32DE40C2D860172543C84BE08E7E67