Open-source PixlStash 1.12 runs ComfyUI workflows from your image library

Infamous_Campaign687 · reddit · 2026-10-19

A developer released version 1.12 of PixlStash, an open-source local image/video manager that deeply integrates your library with ComfyUI workflows:

Workflow grouping: ComfyUI images embed their generating workflow; PixlStash groups identical graphs (different checkpoints/LoRAs) into workflow cards, collapsing thousands of images into a few dozen cards.
Character LoRA generation: right-click a person with a LoRA and choose "Create with LoRA"; pick two people, each with its own strength and trigger word, to generate a two-character image.
Recipe tab: every image exposes its full workflow, prompt, models, LoRA strengths and seed for reproduction.
Repair and versioning: workflows with missing checkpoint/VAE/text encoders can be fixed by swapping models from your shelf; editing in ComfyUI and hitting Ctrl+S stores a new version.
Workflow sets: each checkpoint logs its paired VAEs/encoders/LoRAs and image counts, letting you swap a workflow's models as a unit (e.g. between ZIT and Krea 2).
Everything runs locally; one-button node install links via ComfyUI-Manager.

The author is seeking testing feedback, especially on Windows (full restart of both apps may be needed after linking) and macOS (M1 8GB). Back up your library before upgrading.

Original post →

Open-source PixlStash 1.12 runs ComfyUI workflows from your image library

More from Multimodal