Open-Source Guide | Image & Video

Image Generation

Stills • Textures • LoRAs

⚠️ Commercial Safety Guide

🚫

Avoid (Closed/Licensed)

Midjourney
Adobe Firefly (Web)
DALL-E 3

Risk: Hidden fees, data privacy, no raw file control.

➔

✅

Use Instead (Open Source)

Flux.1 (Dev/Schnell)
Stable Diffusion 3.5
Hunyuan-DiT

Benefit: Safe for commercial, fine-tunable, free.

🚀 Where to run them

ComfyUI Local GPU

Full control. Best for complex workflows, Inpainting, & LoRAs.

Flux.1 (Dev/Schnell) SD 3.5 Large SDXL Turbo Playground v3 PixArt-Sigma

Fal.ai Cloud API

Fastest inference. Good for batching & simple prompts.

Flux.1 (All variants) SD 3.5 SDXL Lightning Auraflow

Video Generation

Motion • Animation • T2V

⚠️ Commercial Safety Guide

🚫

Avoid (Closed/Licensed)

Runway Gen-3 (Web)
Sora / Kling AI
Luma Dream Machine

Risk: High cost per second, no keyframe control.

➔

✅

Use Instead (Open Source)

HunyuanVideo
Wan-2.1
LTX-Video

Benefit: Unlimited runs, full motion control.

🚀 Where to run them

ComfyUI Local GPU

Pipeline development. Chain image-to-video, frame interpolation.

HunyuanVideo SOTA quality & coherence.
Wan-2.1 Great prompt adherence.
LTX-Video Fast iteration.
CogVideoX Solid baseline.

Fal.ai Cloud API

Scale production. No setup required.

HunyuanVideo
LTX-Video
Mochi-1
Wan-2.1

Toolkit & Tables

💡

Pro Tip: The "Drag & Drop" Trick

Did you know? You can drag any image generated in ComfyUI directly into your browser window. It will instantly load the entire workflow (nodes + settings) used to create it.

🧠 Designer's Dictionary

Checkpoint The main model (e.g., Flux). The "Brain".

LoRA A style filter. Small file for specific looks.

Seed The DNA. Keep it same = exact same image.

Latent The "sketchpad" before pixels are drawn.

📊 Quick Comparison

Model	Type	License
Flux.1	Image	✅ Open
HunyuanVideo	Video	✅ Open
Wan-2.1	Video	✅ Open
Midjourney	Image	❌ Closed
Sora	Video	❌ Closed

📂 Browse Workflow Templates