Image Factory isn't a LoRA tester. It's a creative experimentation workspace designed to explore multiple artistic directions from a single generation. Image Factory generates multiple creative directions from a single execution, allowing rapid experimentation without repeatedly reconfiguring your workflow.
Version 2.0 W/L Architecture
Features:
✓ Model Agnostic (Krea2, Illustrious, Pony, ZiT, Flux2, Anima and more)
✓ Wireless Architecture for Modules
✓ AIO & Standard Checkpoint Support
✓ Low VRAM Friendly (Designed for sequential execution)
✓ Modular Variant System
✓ Optional Per-Variant Upscaling
Version 1.0
Features:
✓ Baseline Output
✓ Multiple LoRA Experiment Stacks
✓ Independent Sampling
✓ Chrono Save for CivitAI
✓ Modular Layout
✓ Model Agnostic Design
Intent:
Image Factory is a productivity-oriented workflow designed to generate multiple creative variations from a single prompt in one execution.
Instead of repeatedly changing settings and running the workflow again, Image Factory lets you experiment with multiple LoRA stacks and independent sampling configurations simultaneously.
Core Philosophy:
✓ One Prompt.
✓ One Model.
✓ One Click.
✓ Multiple Creative Directions.
Image Factory minimizes repetitive setup while maximizing experimentation speed.
Features
✓ Model Agnostic
✓ Works with any multiple checkpoints / models (Illustrious / Krea2 / Pony / etc. Just select the appropriate Clip / VAE if not an AIO model and the switch is already there).
✓ Low VRAM Friendly
Designed around sequential execution instead of large parallel batches.
Suitable for 8 GB GPUs and above.
✓ Independent Experiment Branches
Each branch has its own
✓ LoRA Stack
✓ Strengths
✓ Sampler
✓ Scheduler
✓ CFG
✓ Steps
✓ Denoise
Changing one branch never affects another.
✓ Baseline Output
One output intentionally contains no LoRAs.
This acts as the reference image so you can immediately evaluate whether your LoRA stacks improve the result.
✓ Modular Design
✓ Each experiment branch is self-contained.
Additional branches can easily be duplicated or removed based on your workflow.
Common Workflows:
• Compare artistic styles
• Test multiple LoRA combinations
• Compare sampler behavior
• Compare scheduler behavior
• Explore CFG variations
• Explore denoise values
• Build style libraries
• Quickly select the best image for upscaling
Typical Workflow
• Select checkpoint
• Configure prompt
• Configure image size
• Generate baseline
• Configure Experiment #1
• Configure Experiment #2
• Compare results
• Continue from your favorite image
Performance Notes
The workflow is designed to reduce repetitive work.
The checkpoint, prompt encoding and common configuration are reused while each experiment samples independently.
This makes experimentation significantly faster than repeatedly reconfiguring and executing separate workflows.
Hardware Tested:
✓ RTX 2080 (8 GB)
✓ 64 GB System RAM
✓ ComfyUI 0.27.0
Description
🚀 What's New
Introduced the FisherKing IF Engine with wireless Model / CLIP / VAE distribution.
Added modular Variant architecture for cleaner experimentation.
Integrated Krea2 Conditioning directly into the engine.
Added optional per-variant resolution override.
Simplified workflow organization and reduced wiring.
Improved maintainability and future expandability.
Existing users are encouraged to upgrade. Version 1.1 is fully backward-compatible in terms of workflow philosophy but introduces a significantly cleaner engine architecture.
Looks like we don't have an active mirror for this file right now.
CivArchive is a community-maintained index — we catalog mirrors that volunteers upload to HuggingFace, torrents, and other public hosts. Looks like no one has uploaded a copy of this file yet.
Some files do get recovered over time through contributions. If you're looking for this one, feel free to ask in Discord, or help preserve it if you have a copy.