vslinx-ComfyUI-vslinx-nodes/web/docs/vsLinx_LoadLastGeneratedImage.md

This node loads an image from the output folder via a dropdown, with the newest image pre-selected. It is designed as a replacement for ComfyUI's built-in LoadImageOutput node. It supports auto-refresh after generation, so it automatically picks up newly generated images without manual interaction. The node also supports the MaskEditor (right-click → "Open in MaskEditor"), and painted masks are preserved across workflow executions and tab switches. If no image is available, the node falls back to a 512×512 black image to prevent blocking the workflow.

This node does the following:

• Lists all images in the output folder (optionally including subfolders), sorted by modification time (newest first).
• Shows a dropdown to select any image from the list, with a live preview below the widgets.
• When Auto refresh after generation is enabled, automatically detects newly generated images after workflow execution and selects them. If no new image appeared, the current selection (and any painted mask) is preserved.
• Supports the MaskEditor: painted masks are saved to input/clipspace/ by ComfyUI's MaskEditor. The node detects this and loads the masked image as both the preview and the source for the mask output.
• Loads the selected image, corrects EXIF orientation, converts to RGB, and extracts the alpha channel as a mask (if present). Images with transparency (RGBA or palette-based) produce an inverted alpha mask; images without alpha produce an empty (all-zero) mask.
• Falls back to a 512×512 black image with an empty mask when no valid image is found or the file is missing.

Parameters:

Parameter	Type	Description
image	COMBO (UI-only)	Dropdown listing images from the output folder, sorted newest-first. This widget is not sent to the backend; it controls the hidden image widget.
Auto refresh after generation	BOOLEAN	When true, the node automatically updates the dropdown and selects the newest image after each workflow execution — but only if a genuinely new file appeared. Default: true.

Buttons:

Button	Description
Refresh	Re-scans the output folder and selects the newest image.
Choose file to upload	Opens a file picker to upload an image into the output folder, then selects it.

Outputs:

Parameter	Type	Description
image	IMAGE	The loaded image as a float32 tensor (1, H, W, 3) (RGB, 0–1 range). When a painted mask exists, this is the original image from clipspace (without the mask overlay).
mask	MASK	A float32 tensor (1, H, W). Derived from the alpha channel (inverted) if the image has transparency, or from the MaskEditor's painted mask. All-zero if neither exists.

Node Properties (right-click → Properties):

Property	Type	Default	Description
include_subfolders	BOOLEAN	true	When true, the dropdown lists images from all subfolders inside the output directory. When false, only top-level files are shown. Changing this re-scans and selects the newest image.

Notes:

• The dropdown always shows output-folder images. When you paint a mask via the MaskEditor, the masked image is saved to input/clipspace/ by ComfyUI internally. The node detects this, updates the hidden image widget to point at the clipspace file, and shows the masked preview — but the dropdown still displays the original output image.
• Painted masks are preserved across workflow executions (when no new file is generated), tab switches, and page reloads. The hidden image widget stores the full annotated path (e.g. clipspace/clipspace-painted-masked-xxx.png [input]) which is serialized with the workflow.
• Auto-refresh works by snapshotting the newest file before execution starts. After execution completes, it compares the new file list — only if a different file now sits at position 0 does it switch selection (and clear any painted mask). If no new file appeared, the current selection and mask are untouched.
• If the selected file no longer exists on disk (e.g. manually deleted), the node falls back to a 512×512 black image rather than erroring. This ensures downstream nodes always receive valid tensors.
• Supported image formats: .png, .jpg/.jpeg, .webp, .bmp, .gif, .tiff/.tif.
• The file list is fetched via a custom API endpoint (GET /vslinx/output_images_list?include_subfolders=true|false) to avoid blocking the UI on large output folders.
• IS_CHANGED uses a SHA-256 hash of the file contents, so re-executions are skipped when the selected image hasn't changed on disk.