biagiomaf/smart-comfyui-gallery

SmartGallery DAM – Open‑source Digital Asset Manager for ComfyUI

What it is – A self‑hosted web application that indexes, searches and organizes the huge folders of images, videos and audio that ComfyUI (and other AI generators) produce. It runs independently of ComfyUI, watches any folder you point it at, and adds AI‑aware metadata (prompts, models, LoRAs, workflow hashes) so you can find and remix assets without opening the generator again.

Why it matters – When you generate tens of thousands of AI images, the file system quickly becomes a mess. SmartGallery gives you:

• instant, full‑text and AI‑powered search (OmniQuery can turn a natural‑language question into a SQL query);
• visual clustering of assets that share the same prompt or graph architecture;
• a built‑in “remix” engine that lets you edit the underlying JSON workflow and queue a new generation directly from the gallery;
• client‑facing portals where reviewers can comment, rate, and approve images while the original prompts stay private.

Key features (as listed in the README)

Feature	What you get
Ultra‑fast browsing	Responsive web UI works on desktop, tablet and phone.
Scalable file handling	Handles hundreds to hundreds of thousands of media files, with drag‑and‑drop upload and bulk move/copy/delete.
Deep search & OmniQuery	Search by prompt, model, LoRA, date, rating, or any tag; ask plain‑English questions and an LLM writes the SQL for you.
Smart Asset Clustering	Press Shift + C to group assets that share the same workflow hash or prompt hash; visual badges (#HASH) let you jump into a cluster inspector.
Full Asset Details panel	Press I for a two‑tab inspector showing file metrics, disk paths, symlink targets, and the full ComfyUI node pipeline.
Remix Engine / Nodepad	Press B to open a three‑tier workspace where you can edit raw JSON, build custom dashboards, or auto‑wire compatible LoRAs.
LoRA Synergy™	Offline matcher that reads safetensor headers, guarantees checkpoint compatibility and auto‑adds trigger words.
Client sharing portal	Create virtual collections, attach Markdown briefs, set status tags, and expose a read‑only exhibition mode for stakeholders.
Multimedia support	Images, video (with on‑the‑fly FFmpeg transcoding) and audio (inline waveforms and player).
Cross‑platform	Native Windows portable zip, macOS/Linux binaries, and an official Docker image.
User & role management	Define admins, reviewers, guests; control who can edit, comment or only view.

Installation at a glance

1. Docker – Pull mmartial/smart-comfyui-gallery and run with a bind‑mount to your ComfyUI output folder.
2. Portable app (Windows) – Download the zip, unzip, and run SmartGallery.exe; no Python or dependencies needed.
3. Source – Requires Python 3.10+, install via pip install -r requirements.txt and launch python -m smartgallery. The README provides launch flags for port, DB path, and FFmpeg integration.

Typical workflow

1. Point SmartGallery at the folder where ComfyUI writes its generations.
2. As new images appear, they are automatically indexed (prompt, model, LoRA, seed, etc.).
3. Use the search bar or OmniQuery to locate a specific asset.
4. Press Shift + C to view all variations of that prompt, or B to remix the workflow and queue a new run.
5. Create a virtual collection, attach a Markdown brief, set status tags, and share the collection link with a client. The client can rate and comment without ever seeing the underlying prompts.

Who should use it

• Solo AI artists overwhelmed by thousands of outputs.
• Studios or agencies that need a review loop with clients while keeping proprietary prompts private.
• Teams that want to keep a permanent, searchable archive of all generations across multiple machines or NAS devices.
• Anyone who wants a media manager that understands AI‑generation metadata out‑of‑the‑box.

Where to learn more

• Official website: smartgallerydam.com – docs, wiki, demo video.
• GitHub repo: stars, releases, and the full changelog.
• Video walkthroughs linked throughout the README (e.g., Smart Asset Clustering, OmniQuery).

---

All information above is taken directly from the repository’s README; no additional features have been inferred.