forked from sagnik/Project_Astral
49 lines
3.5 KiB
Plaintext
49 lines
3.5 KiB
Plaintext
1. The Global Navigation (The Sidebar)
|
||
A slim, translucent sidebar (glassmorphism) that keeps the focus on the creative workspace.
|
||
Icon Set: Minimalist wireframe icons.
|
||
Top: Production Hub (The "Main Stage"), Asset Library, Model Forge (Admin only).
|
||
Bottom: User Profile (GAuth avatar), System Status (Green/Red pulse indicating ComfyUI API health).
|
||
2. Page 1: The Production Hub (The Mastermind)
|
||
Inspired by Midjourney’s speed and Veo’s visual clarity.
|
||
Top Section: The Inputs (The "Alchemy" Zone)
|
||
Left Pane (Actor Identity): A large, dashed-border drop zone.
|
||
Dynamic State: When files are dropped, it displays a 3D point-cloud preview (derived from LiDAR) instead of just flat thumbnails.
|
||
Connectivity: A "Link Private Dataset" button that opens a mini-modal to select pre-processed actor LoRAs.
|
||
Right Pane (Product Essence): A smaller, focused drop zone for the object.
|
||
Visual Logic: Once a product is uploaded, a small AI-badge appears: "Extracting Material Properties..." (indicating IP-Adapter prep).
|
||
Middle Bar: The Intent (The Prompt)
|
||
A single, elegant text input field spanning the width.
|
||
Floating Action: As you type, "Ghost Text" suggests stylistic modifiers (e.g., "in the style of a 35mm Leica shot").
|
||
The "Generate" Button: A high-gloss, subtle gradient button that pulses slightly when inputs are ready.
|
||
Lower Section: The Loom (Task Orchestrator)
|
||
Layout: A horizontal scrolling "Film Strip" of tasks.
|
||
Card Design: Each card shows:
|
||
Live Preview: A low-res latent preview (from ComfyUI's K-Sampler).
|
||
Status Stats: ETA, VRAM Load, and Step Count.
|
||
The Finish: When complete, the card expands slightly, and the video begins to loop silently.
|
||
3. Page 2: Asset Library (The Digital Vault)
|
||
Where raw data becomes a trained dataset.
|
||
Dataset Grid: Folders categorized by Actor name or Product Line.
|
||
Dataset "Deep Dive":
|
||
LiDAR Visualizer: A toggle to view the iPhone sensor data overlays on the images.
|
||
Permission Matrix: A simple "Toggle" system to share this specific dataset with other team members or lock it to the current user.
|
||
Processing Status: A "Training" bar for when you are fine-tuning a specific LoRA on a new actor.
|
||
4. Page 3: Account & Identity
|
||
Managed via Firebase + GAuth.
|
||
User DNA: Displays user role (Creative, Lead, or Guest).
|
||
Usage Analytics: Simple, clean charts showing "Minutes of Video Generated" and "NAS Storage Occupied."
|
||
Personal Settings: Theme selection (Midnight Black vs. Studio Gray).
|
||
5. Page 4: Admin & Developer Console (The "Engine Room")
|
||
Hidden from standard users; accessible via specialized admin login.
|
||
Pipeline Configurator:
|
||
JSON Editor: A live editor to swap the "Hidden" ComfyUI Workflow JSON without redeploying the frontend.
|
||
Model Routing: Change which local IP or GPU is handling the LTX-2 vs. the SDXL/Flux image generation.
|
||
Prompt Engineering Lab:
|
||
System Prompt Overrides: Define the "Global Aesthetic" (e.g., "Always ensure skin textures are cinematic").
|
||
Log Stream: A real-time terminal window showing the ComfyUI console output for debugging "broken" generations or VRAM overflows.
|
||
6. The Technical "Secret Sauce" for your Developer
|
||
To make this work like a professional tool, the frontend must use Zustand or Redux to manage the state of the "Task Loom."
|
||
Feature Implementation Logic
|
||
Real-time Updates WebSocket (/ws endpoint) listeners in the Frontend to catch node-execution events.
|
||
LiDAR Data Use Three.js in the frontend to render the .obj or depth maps for a "High-Tech" feel.
|
||
NAS Storage The Save Video node in ComfyUI writes to /mnt/nas/output. The Frontend reads from the same mount via a simple static file server. |