Skip to content

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Latest commit

 

History

10 Commits

Folders and files

Repository files navigation

Invoke-Favorite-Workflows.kleiner.mp4

invoke_workflow_switcher (⇄ Favorite Workflows) — Stage & Pipeline Switcher for InvokeAI 7 (webv2)

A lightweight, zero-core-patch Custom Node Pack for InvokeAI 7 (webv2) that adds a ⇄ Favorite Workflows form element to the Workflow Form Builder (Edit tab) and Form View (View tab).


What is the purpose of this node?

⇄ Favorite Workflows is a fast, inline workflow switcher that lets you map real-world creative production stages directly at the top of your workflow form.

Take an architectural visualization studio as a practical example — turning an early design sketch, CAD screenshot, or basic clay render into a finished client presentation rarely happens in a single step. Instead, it follows a natural sequence of focused stages:

Example Pipeline A — Architectural Presentation (Image & Cloud Video)

  1. Stage 1 — Image-to-Image (Concept to Photo): Turn a CAD screenshot or hand sketch into a photorealistic architectural rendering (e.g. Nano Banana Img2Img).
  2. Stage 2 — Image Restyle (Lighting & Season Variants): Re-evaluate the rendering and create a night-time or winter mood instead of daylight (e.g. Nano Banana Remix / Restyle).
  3. Stage 3 — Image-to-Video (Initial Motion Sequence): Generate initial 10-second camera flights from the best still renderings (e.g. Omni Image-to-Video).
  4. Stage 4 — Video Extend: Extend the best 10-second clips into 20-second presentation sequences (e.g. Omni Video Extend).

Example Pipeline B — Local Video Production & Restyling (LTX 2.5)

  1. Stage 1 — Base Image / Initial Video Generation: Create the base image or initial video clip (Txt2Vid / Img2Vid).
  2. Stage 2 — Video Pre-Processing (for LTX Control): Extract structural control passes (Depth map, Canny edges, or DWPose) from the video (LTX 2.5 — Vid to Control-Vid).
  3. Stage 3 — Video Restyle & 2× Upscale: Restyle the video while locking its camera motion and geometry via Union IC-LoRA (LTX 2.5 — Restyle Video / 2× Upscale).
  4. Stage 4 — Direct Effects / Extension: Apply direct IC-LoRA effects or extend the sequence.

By placing ⇄ Favorite Workflows at the top of each workflow in a group, you configure the relevant stages once in the first workflow, click ★ Save to Global, and stamp the exact same favorite list into the sibling workflows with a single click (↓ Load Global). You can maintain separate favorite groups for different pipelines (e.g. one group for your Cloud/Nano/Omni pipeline, and another group for your LTX 2.5 pipeline).


Why not put everything into one giant workflow — or just use the Library Dialog?

  • The problem with monolithic "all-in-one" workflows:
    Cramming image generation, video pre-processing, structural restyling, upscaling, and video extension into a single graph produces an overwhelming wall of 40+ sliders, switches, and conditional inputs. Splitting a complex process into smaller, stage-specific workflows keeps each form clean and easy to scan — while ⇄ Favorite Workflows lets you jump to the next stage in 1 click.
  • The problem with the full Workflow Library dialog (This project / Browse / Yours):
    Opening the full library modal interrupts your flow, lists dozens of unrelated workflows, and prompts with copy-choice confirmation dialogs on repeated switches.

Key Features

  • Per-Workflow Favorites stored in the Workflow JSON:
    Each workflow remembers its own curated list of companion workflows.
  • Global Favorites Preset (★ Save to Global / ↓ Load Global):
    Set up your favorite stages once and copy them into sibling workflows with a single click.
  • Direct 1-Click Library Update (💾 Save / Update in Yours):
    Update the current workflow directly in Your Library in place without creating duplicate copies, while automatically preserving InvokeAI's native library link (🔗).
  • 3 Color-Coded Sources & Multi-Family Filter Chips:
    • Your Library (Yours — Cyan): Your saved, fine-tuned personal workflows (default primary source).
    • + Built-in (Browse) (Lavender/Purple): Optional toggle to include InvokeAI's built-in default workflows.
    • + Node Pack Templates (Amber/Gold): Optional toggle to include factory templates shipped inside custom node packs.
  • Compact & Collapsible:
    Folds into a single header bar showing the active workflow badge, with a Colorized toggle (adopts custom accordion/input surface brightness) and an Extra Info toggle (on-demand ⓘ button for workflow descriptions).

Screenshots & Video Demo

Workflow Favorites Switcher Overview (Click to enlarge)

(Click image to open in full resolution)

Installation (InvokeAI 7)

This extension is a pure Custom Node Pack — no InvokeAI core or frontend files on disk are modified.

Option A — Install via InvokeAI 7 UI (Nodes → Add Nodes)

  1. In InvokeAI 7, open Nodes → Add Nodes.
  2. Paste the repository URL into Git URL:
    https://github.com/tschuelke/invoke_workflow_switcher.git
    
  3. Click Install.

Option B — Manual Drop-in Installation

  1. Copy the invoke_workflow_switcher folder into your InvokeAI nodes directory:
    <InvokeAI-Root>/nodes/invoke_workflow_switcher/
    
  2. Restart InvokeAI 7 (or click Reload Nodes in the Node Manager) and refresh the browser (Ctrl+F5).

How to Use

  1. Open Workflow → Edit (Form) and click + Add form element → ⇄ Favorite Workflows.
  2. Drag the ⇄ Favorite Workflows card to the top of the form.
  3. Filter by workflow family chips (e.g. LTX 2.5, OMNI 1.1, Nano Banana), check the workflows that belong to your pipeline, and click ★ Save to Global to reuse the selection in sibling workflows.
  4. Click 💾 Save / Update in Yours to save your workflow in Your Library without creating duplicates.
  5. Switch to View mode to jump between your pipeline stages with 1 click!

License

Apache License 2.0

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages