1. Call atlas_list_models with type="Video" to see all available video models
2. Call atlas_get_model_info with the chosen model ID to get its parameter schema
3. Then call scripts/generate.py generate with the correct --model and parameters

📦 Installation

1. Add the skill to your AI agent

npx skills add ristponex/nsfw-ai-video-generator-skill

2. Prerequisites

• Python 3.10+ — Download
• uv — Install guide

3. Get your API key

Sign up at Atlas Cloud — free credits included, no credit card required.

4. Set your API key

Mac / Linux:

export ATLAS_CLOUD_API_KEY=your_api_key_here

Windows (PowerShell):

$env:ATLAS_CLOUD_API_KEY="your_api_key_here"

Or use a .env file in the project root:

ATLAS_CLOUD_API_KEY=your_api_key_here

🎬 Popular NSFW Video Models

| Model | Model ID | Type | Resolution | Duration | Price | |-------|----------|------|------------|----------|-------| | Wan 2.2 Spicy | alibaba/wan-2.2-spicy/image-to-video-lora | I2V | 480p–720p | 5s | ~$0.15 | | Wan 2.2 Spicy Extend | alibaba/wan-2.2-spicy/video-extend | Extend | 720p | variable | ~$0.15 | | HunyuanVideo T2V | atlascloud/hunyuan-video/t2v | T2V | 720p | 5s | ~$0.40 | | HunyuanVideo I2V | atlascloud/hunyuan-video/i2v | I2V | 720p | 5s | ~$0.40 | | Jiggle Up | video-effects/jiggle-up | Effect | 720p | 5s | ~$0.10 |

Type key: T2V = text-to-video, I2V = image-to-video, Extend = extend existing video, Effect = apply effect to image/video

These are the most popular uncensored models. Atlas Cloud hosts 50+ video models in total — run uv run scripts/generate.py models to see the full list.

🚀 Usage

Text to Video

Generate a video from a text prompt:

uv run scripts/generate.py generate "a cinematic sunset over the ocean, golden hour lighting" \
  --model hunyuan-t2v

Image to Video

Animate a static image into a video:

uv run scripts/generate.py generate "slow camera zoom with subtle movement" \
  --model wan-spicy \
  --image input.jpg

Video Extend

Extend an existing video clip:

uv run scripts/generate.py generate "continue the scene with smooth motion" \
  --model wan-spicy-extend \
  --extra '{"video_url": "https://example.com/your-video.mp4"}'

List Available Models

uv run scripts/generate.py models

Using with AI Agents

Once installed, just ask your AI agent naturally:

"Generate a 5-second video of waves crashing on a beach using Wan Spicy"
"Create an NSFW video from this image"
"What video models are available?"

The agent will automatically use this skill to call the right model with the right parameters.

💰 Price Comparison

| Platform | Wan Spicy (per video) | Subscription Required? | |----------|----------------------|----------------------| | Atlas Cloud | ~$0.15 | No — pay per use | | PixelDojo | ~$0.25 | $25/mo minimum | | ZenCreator | ~$0.30 | $499/7600 credits | | SoulGen | N/A (no Wan Spicy) | $12.99/mo |

Atlas Cloud offers pure pay-per-use pricing with no monthly commitment. New users receive free credits to get started.

🔧 Advanced Usage

Custom Parameters with --extra

Pass any model-specific parameters as JSON:

uv run scripts/generate.py generate "your prompt" \
  --model wan-spicy \
  --extra '{"seed": 42, "num_inference_steps": 30, "guidance_scale": 7.5}'

Aspect Ratio

uv run scripts/generate.py generate "vertical video for social media" \
  --model hunyuan-t2v \
  --aspect-ratio 9:16

Resolution

uv run scripts/generate.py generate "high quality scene" \
  --model hunyuan-t2v \
  --resolution 720p

Duration

uv run scripts/generate.py generate "extended animation" \
  --model hunyuan-t2v \
  --duration 5

Output Directory

uv run scripts/generate.py generate "your prompt" \
  --model wan-spicy \
  -o ./my-videos

All Options

usage: generate.py generate [-h] [--model MODEL] [--image IMAGE]
                            [--aspect-ratio ASPECT_RATIO]
                            [--resolution RESOLUTION] [--duration DURATION]
                            [--extra EXTRA] [-o OUTPUT]
                            prompt

positional arguments:
  prompt                Text prompt for generation

options:
  --model MODEL         Model ID or shortcut (default: wan-spicy)
  --image IMAGE         Input image for image-to-video
  --aspect-ratio        Aspect ratio (e.g. 16:9, 9:16)
  --resolution          Resolution (e.g. 480p, 720p, 1080p)
  --duration            Duration in seconds
  --extra               Extra params as JSON
  -o, --output          Output directory

🌐 Local Alternatives

If you have a powerful GPU and prefer running models locally, here are some options:

• ComfyUI + Wan 2.2 NSFW LoRA — Free, but requires a GPU with 12+ GB VRAM, plus manual setup of ComfyUI workflows and LoRA weights
• Phr00t/WAN2.2-14B-Rapid-AllInOne — Optimized Wan 2.2 checkpoint on HuggingFace, still requires significant GPU resources
• Civitai workflows — Community-shared workflows for various NSFW video generation setups

This skill is the easiest way if you don't want to set up local GPU inference. No GPU required, no model downloads, no dependency hell. Just an API key and a prompt.

❓ FAQ

What models are available?

Atlas Cloud hosts 50+ video models including Wan 2.2 Spicy, Wan 2.6, Seedance 1.5 Spicy, HunyuanVideo, and more. Run uv run scripts/generate.py models to see the full list.

Do I need a GPU?

No. All inference runs on Atlas Cloud's servers. You just need Python and an internet connection.

What does it cost?

Pay per generation, starting from $0.04. No monthly subscription. New users get free credits. See the price comparison above.

Is my content private?

Yes. Atlas Cloud does not store or monitor your prompts or generated content.

What formats are supported?

Generated videos are saved as MP4 files. Input images can be JPG, PNG, or WebP.

Can I use this without an AI agent?

Yes. The scripts/generate.py script works standalone from the command line. The skill protocol just makes it discoverable by AI agents.

How long does generation take?

Typically 30–120 seconds depending on the model and resolution. The script polls automatically and downloads the result when ready.

What resolutions are available?

Most models support 480p and 720p. Some support 1080p. Check the specific model info with atlas_get_model_info or the models table above.

Can I generate SFW content too?

Absolutely. The uncensored models work for any content — the "uncensored" part just means there are no artificial content restrictions.

What's the maximum video duration?

Depends on the model. Most generate 5-second clips. Use the video-extend models to extend clips. Check individual model specs for details.

How do I pick the right model?

• Wan 2.2 Spicy — Best for image-to-video with NSFW LoRA, good quality at low cost
• HunyuanVideo T2V — Best for text-to-video when you don't have a reference image
• HunyuanVideo I2V — High-quality image-to-video, higher cost
• Jiggle Up — Quick special effects on images, lowest cost
• Wan 2.2 Spicy Extend — Extend existing video clips to make them longer

Can I batch-generate multiple videos?

Not in a single command, but you can script it easily:

for prompt in "prompt one" "prompt two" "prompt three"; do
  uv run scripts/generate.py generate "$prompt" --model wan-spicy --image input.jpg
done

What happens if generation fails?

The script will print an error message from the API and exit with a non-zero status code. Common reasons include invalid prompts, unsupported parameters for the chosen model, or insufficient credits.

🛠 Troubleshooting

"Error: Atlas Cloud API Key not found"

Make sure your API key is set. See Installation step 4.