Skip to content

Release v1.6.0

Latest
Latest

Choose a tag to compare

View all tags
@github-actions github-actions released this 26 May 00:20
· 3 commits to main since this release
v1.6.0
b47aaf4

Highlights for users

  • Added Smart Fit, a new Auto-Fit mode that fits the current resolution to the closest preset aspect ratio while keeping the final size close to the original.
  • Added Canvas Only compact mode, which hides extra sections below the 2D canvas and gives more space to the visual resolution picker.
  • Added Prioritize Ratio scaling, allowing scale, resolution, and megapixel targets to preserve the current aspect ratio as the main priority.
  • Added Auto-Snap automation, so newly detected or resized dimensions can be snapped automatically to model-friendly multiples.
  • Improved Auto-Detect so connected previews can update live when source images or source-node settings change, without always requiring a workflow run.
  • Added live auto-detect support for LayerForge nodes.
  • Added auto-detect support for LocalImageGallery nodes, including safer handling when no image is selected.
  • Added Flux.2 documentation and preset/latent guidance, including latent_128x16 support information.
  • Redesigned the compact help dialog with clearer shortcut information, icons, usage tips, and a GitHub repository link.
  • Improved tooltip text across the node so common controls are easier to understand.
  • Improved ComfyUI interaction compatibility, especially around pointer capture and Ctrl/Ctrl+Shift drag behavior on the Resolution Master canvas.
  • Moved scaling and auto-detect calculations into backend-supported helpers for more consistent behavior between the UI and serialized workflows.

Auto-Detect and integrations

  • Auto-Detect now tracks source signatures, which prevents stale dimensions after reconnecting or switching source nodes.
  • The node now displays live tracking status for detected input dimensions, such as live frontend updates, backend cache fallback, after-run updates, or no input.
  • Frontend preview watchers now monitor image, path, and related widget changes on connected source nodes.
  • Concurrent dimension checks are guarded so multiple detection passes do not race each other.
  • Auto-Detect state is cleared more reliably when the input is disconnected.
  • LayerForge dimensions can be read directly from its live frontend canvas when available.
  • LocalImageGallery selected images can be detected from the frontend, with fallback parsing where needed.
  • Empty LocalImageGallery selections no longer trigger incorrect backend fallback dimensions.
  • Source detectors were split into pluggable modules, making future custom source integrations easier.

Scaling, fitting, and snapping

  • Smart Fit was added as a dedicated Auto-Fit toggle.
  • Prioritize Ratio was added for aspect-ratio-preserving scaling.
  • Snap automation was added to the Auto-Detect workflow.
  • Auto-Detect automation order was clarified as Fit, Resize, Snap, then Calc.
  • Snapped dimensions now respect minimum constraints and avoid invalid zero-size results.
  • Aspect-locked canvas dragging now uses more robust ratio math.
  • Grid and non-grid aspect-locked resize paths were simplified for more consistent results.
  • Backend calculation support now handles scale, custom value, snap, auto-fit, auto-detected image calculations, and model-specific calculations through a unified API route.

Canvas and interaction

  • Canvas Only mode can collapse the extra controls while keeping the 2D canvas usable.
  • Preferred node height is preserved separately for compact and expanded modes.
  • User-set canvas height is preserved more reliably during automatic layout changes.
  • Canvas grid dots now render at real data-value positions and avoid duplicate dots.
  • Ctrl+Shift drag conflicts with ComfyUI canvas zoom are bypassed when interacting with Resolution Master controls.
  • Pointer handling was updated for newer ComfyUI versions that changed pointer capture internals.
  • Captured pointer drag updates are scheduled more smoothly to reduce UI blocking during drag operations.

UI and documentation

  • Added node tooltips for inputs and outputs.
  • Added and refined README documentation for Prioritize Ratio, Auto-Snap, Auto-Detect automation, Flux.2, latent outputs, and model-specific modes.
  • Translated remaining Polish comments, documentation strings, and user-facing labels to English.
  • Improved visual polish for action buttons, toggle gradients, hover states, snap slider spacing, value text offsets, and calc info message sizing.
  • Aspect ratio and preset dialogs now handle scrollbar gutters and window resizing more consistently.
  • The help dialog was redesigned and scaled down for a better fit.

Stability, diagnostics, and maintenance

  • Added structured JavaScript logging and Python logging with better error handling across the extension.
  • Added optional performance diagnostics for canvas drawing, interaction methods, LiteGraph canvas methods, long tasks, frame gaps, event loop lag, event timing, and JavaScript heap usage.
  • Duplicate CSS stylesheet loading is now prevented.
  • Resource loading was moved away from $el helpers toward native DOM APIs with better error handling.
  • The codebase was reorganized into focused modules for auto-detect, canvas, calculations, drawing, interaction, node lifecycle, presets, styles, utilities, and logging.
  • The old monolithic ResolutionMaster.js was split into smaller maintainable modules.
  • CSS files were moved from js/css into js/styles.
  • Python backend helpers were extracted into the core/ package.

Full Changelog: v1.5.14...v1.6.0

Assets 2
Loading