A ComfyUI custom node that lets you load a folder of images, preview each one with its own editable prompt directly on the node, batch process them through any downstream workflow, and save the results using the original filenames.

screenshot

Nodes

Scene Prompt Viewer — main node Scans a folder and shows a card per image. Outputs IMAGE batch, prompts, and filenames. Use slot_count to expose individual image_N / prompt_N sockets for per-image handling.
Scene Output Saver — output node, designed to pair with Scene Prompt Viewer Receives the IMAGE batch and filenames, builds a subfolder from your template, and saves each file using its original name.
Scene Prompt Text — helper node A multiline text input for feeding longer prompts into prompt_in_N overrides.

Use cases

Batch image processing with original filenames preserved

Send a whole folder through any processing node. Results are saved with the same filenames as the source — not sequential numbers.

Scene Prompt Viewer → image processing → Scene Output Saver

When using individual sockets for single-image processing, the built-in Save Image node with a manually typed filename is simpler than Scene Output Saver, which is designed for batch use.

Visual library browser with individual handling

See your entire image folder at a glance without opening multiple Load Image nodes. Use slot_count to pull out specific images into separate downstream workflows — all within the same node.

Tip: prompts outputs all prompts joined by newline. To split them per image, pair with a string-split node from your installed packs.

Install

Install via ComfyUI Manager by searching ScenePromptViewer in the Node Pack tab, or clone manually:

cd ComfyUI/custom_nodes
git clone https://github.com/yixuanzona/ComfyUI-ScenePromptViewer.git

Restart ComfyUI. All nodes appear under image → utils.

Quick start

Drop Scene Prompt Viewer onto the canvas, paste your image folder path, click ↻ Rescan
Write a prompt on each card if needed
Connect IMAGE → your processing nodes → Scene Output Saver; connect filenames directly to Scene Output Saver
Set scene and version in Scene Output Saver, then Queue

Scene Prompt Viewer

Toolbar

| Button | Action | |--------|--------| | ↻ Rescan | Re-scan folder, restore hidden scenes | | 📂 Open | Open folder in file explorer | | Import all prompts | Fill all cards from a pasted block | | Export all prompts | Copy all prompts to clipboard |

Per-card buttons

| Button | Action | |--------|--------| | ↻ | Reload this card's thumbnail | | ✕ | Hide this scene for the session (Rescan to restore) |

Scene Output Saver

Settings

| Field | Default | Description | |-------|---------|-------------| | output_root | ComfyUI output folder | Root save directory — accepts any absolute path | | folder_template | {scene}/{date}_{version} | Subfolder structure — supports {scene} {date} {version} | | scene | scene | Your project or batch name | | version | v1 | Label for this run | | filename_suffix | (empty) | Appended before the extension — e.g. _resize → WinterScene_resize.png |

Example — scene = Scene01, version = v1, filename_suffix = _resize:

ComfyUI/output/Scene01/20260612_v1/
  WinterScene_resize.png
  CityScene_resize.png
  ...

Each run gets its own subfolder by changing version. This keeps iterations clearly separated and prevents previous outputs from being overwritten.

Notes

Batch output handles any number of images — individual sockets (slot_count) go up to 8 for images needing separate treatment
Recommended batch size is around 20 images for smooth performance
Only top-level files are scanned, no subfolders
For VRAM-heavy processing nodes, keep batch sizes manageable to avoid out-of-memory errors

MIT License