Step-by-step workflow
1
Install ComfyUI
Download ComfyUI Portable — no Python setup needed.
# Windows: Download from https://github.com/comfyanonymous/ComfyUI/releases
# Mac/Linux:
git clone https://github.com/comfyanonymous/ComfyUI
cd ComfyUI && pip install -r requirements.txt
2
Download LTX-Video model
Place files in correct ComfyUI folders.
pip install huggingface_hub
huggingface-cli download Lightricks/LTX-Video ltx-video-2b-v0.9.5.safetensors
# Place in: ComfyUI/models/video_models/
3
Install LTX nodes via ComfyUI Manager
Manager → Install Custom Nodes → search "LTX-Video" → Install → Restart.
4
Write your video prompt
Formula: [Camera movement] + [Subject] + [Action] + [Environment] + [Style]
"Slow zoom in on a cup of masala chai steaming on a wooden table, morning light, cinematic"
"A young woman walking through a busy Indian street market, golden hour, shallow depth of field, 4K"
# Settings: Resolution: 768x512, Frames: 97, Steps: 30-40, CFG: 3.5
5
Upscale and export
Use built-in video upscaler node or export and upscale with free tools.
Pro tips
→
LTX = fastest free model, 30 seconds on RTX 3090
→
Start with a Flux-generated image as first frame for better quality
→
ThinkDiffusion.com has LTX in browser if you have no GPU
Why this matters for India
// india context
Create product and food videos for social media — free replacement for Sora or Runway subscriptions