🎬 NVIDIA NIM + Remotion Video Generator

Transform your ideas into stunning AI-powered videos with just a text prompt. This Actor combines the power of NVIDIA's NIM LLM for intelligent script generation with Remotion's professional video rendering capabilities.

What it does

🎯 Generate AI-powered videos from text prompts

This Actor creates professional explainer videos by:

• Understanding your topic: Uses NVIDIA's NIM LLM to generate compelling video scripts
• Creating structured content: Produces titles, subtitles, slides with bullet points, and call-to-action text
• Rendering high-quality videos: Uses Remotion to render videos in multiple resolutions (480p, 720p, 1080p)
• Delivering instantly: Uploads videos to Apify Key-Value Store with public download URLs

How to use

Input

{
  "prompt": "The future of AI and how it's changing our world",
  "quality": "medium",
  "outputFileName": "ai-video",
  "skipRender": false
}

Input fields:

Field	Type	Required	Default	Description
prompt	string	✅ Yes	-	The topic or idea for your video (max 1000 chars)
quality	string	No	medium	Video quality: low (480p), medium (720p), or high (1080p)
outputFileName	string	No	ai-video	Name for the output MP4 file (without extension)
skipRender	boolean	No	false	Set to true to only generate script without rendering

Output

The Actor outputs structured data to the default dataset:

{
  "prompt": "The future of AI and how it's changing our world",
  "title": "The AI Revolution",
  "subtitle": "Transforming Our World Through Intelligence",
  "slides": [
    {
      "heading": "What is AI?",
      "points": [
        "Artificial Intelligence simulates human thinking",
        "Machine learning enables pattern recognition",
        "Neural networks process complex data"
      ]
    },
    {
      "heading": "Impact on Society",
      "points": [
        "Automation transforms industries",
        "Healthcare becomes more precise",
        "Education becomes personalized"
      ]
    },
    {
      "heading": "The Future",
      "points": [
        "AI assistants become ubiquitous",
        "Creative tasks are augmented",
        "Ethical frameworks evolve"
      ]
    }
  ],
  "cta": "Embrace the future today",
  "slideCount": 3,
  "videoUrl": "https://api.apify.com/v2/key-value-stores/xxx/records/ai-video",
  "scriptKey": "video-script",
  "quality": "medium",
  "duration": 540,
  "durationSeconds": 18,
  "renderedAt": "2024-01-15T10:30:00.000Z"
}

Output fields:

Field	Type	Description
prompt	string	Original user prompt
title	string	Generated video title
subtitle	string	Generated subtitle
slides	array	Array of slide objects with heading and points
cta	string	Call-to-action text
slideCount	integer	Number of slides in video
videoUrl	string	Public URL to download the MP4 video
scriptKey	string	Key in Key-Value Store where script is saved
quality	string	Quality setting used
duration	integer	Video duration in frames @ 30fps
durationSeconds	integer	Video duration in seconds
renderedAt	string	ISO timestamp

Key-Value Store Outputs

The Actor saves two items to the Key-Value Store:

1. video-script - The complete generated script (JSON)
2. {outputFileName} - The rendered MP4 video file

Access via API:

curl "https://api.apify.com/v2/key-value-stores/{storeId}/records/{fileName}" \
  -H "Authorization: Bearer {token}"

💰 Pricing

This Actor uses Pay Per Event pricing:

Event	Price	Description
apify-actor-start	$0.00005	Actor startup (free 5s compute provided by Apify)
video-script-generated	$0.05	AI script generation with NVIDIA NIM LLM
video-rendered	$0.15	Complete video rendering (up to 3 slides)
slide-rendered	$0.03	Each additional slide beyond 3

Example pricing:

• 1 video with 3 slides = $0.20
• 1 video with 5 slides = $0.26 ($0.20 + 2×$0.03)
• Script only (skipRender=true) = $0.05

⚙️ Setup Requirements

Environment Variables

Before running this Actor, set up:

Variable	Required	Description
NVIDIA_API_KEY	✅ Yes	Your NVIDIA NIM API key from NVIDIA NIM

Getting NVIDIA API Key

1. Visit build.nvidia.com
2. Create an account or sign in
3. Navigate to the API section
4. Generate your API key
5. Copy the key to your Actor's environment variables

🚀 Use Cases

Marketing & Sales

• Create product explainer videos
• Generate social media content
• Build video ads from product descriptions
• Produce pitch deck videos

Education & Training

• Convert lessons into engaging videos
• Create tutorial content
• Generate course previews
• Build training materials

Content Creation

• Transform blog posts into videos
• Create YouTube content
• Generate podcast visualizations
• Produce short-form social content

Business Communication

• Make internal announcements
• Create company updates
• Generate investor presentations
• Build brand storytelling videos

📊 Performance

Quality	Resolution	Avg. Render Time*	File Size
Low	854×480	~30s	~5 MB
Medium	1280×720	~60s	~15 MB
High	1920×1080	~120s	~40 MB

*Based on 3-slide video on 4GB memory

Resource Requirements:

• Minimum memory: 2GB
• Recommended memory: 4GB
• Maximum memory: 8GB
• Timeout: 10 minutes

🔧 Technical Details

How it works

1. Input Processing: Validates user input and sanitizes the prompt
2. Script Generation: Sends prompt to NVIDIA NIM (gpt-oss-20b model) with structured instructions
3. Data Validation: Validates and sanitizes the generated JSON script
4. Video Rendering: Bundles React/Remotion components and renders to MP4
5. Output Delivery: Uploads video to Key-Value Store and pushes metadata to Dataset

Error Handling

The Actor includes comprehensive error handling:

• Input validation with clear error messages
• NVIDIA API retry logic (3 attempts with exponential backoff)
• JSON parsing with multiple fallback strategies
• Video data validation before rendering
• Detailed error logging to Key-Value Store

Retry Behavior

• NVIDIA API: 3 retries with exponential backoff (1s, 2s, 4s)
• Validation: Automatic truncation for text exceeding limits
• Rendering: Timeout protection with cleanup

💡 Tips & Best Practices

Prompt Engineering

Good prompts:

• "The benefits of meditation for mental health"
• "How blockchain technology is revolutionizing finance"
• "10 tips for effective remote work"
• "The history and future of electric vehicles"

Tips:

• Be specific about the topic
• Include the target audience if relevant
• Mention key points you want covered
• Keep prompts under 1000 characters

Quality Selection

• Low (480p): Quick previews, social stories, testing
• Medium (720p): Standard presentations, social media, websites
• High (1080p): Professional content, presentations, YouTube

Memory Optimization

• Use 2GB for script generation only (skipRender=true)
• Use 4GB for standard video rendering (3-5 slides)
• Use 8GB for high-quality rendering with many slides

Cost Optimization

1. Test with skipRender=true first to validate scripts
2. Use low quality for development/testing
3. Limit slides by making prompts more focused
4. Monitor the charging_log dataset for usage tracking

🐛 Troubleshooting

Common Issues

"NVIDIA_API_KEY is required"

• Check that you've set the NVIDIA_API_KEY environment variable
• Verify the key is valid at build.nvidia.com

"Video rendering failed"

• Increase memory allocation (4GB recommended)
• Try low quality setting first
• Check that Remotion components are not throwing errors

"Failed to parse LLM response"

• The LLM may have returned non-JSON output
• Try a simpler, more specific prompt
• Check error details in Key-Value Store (error-details)

"Browser/chromium not found"

• The Actor includes Chrome in the Docker image
• This error shouldn't occur in production
• For local testing, install Chrome or set CHROME_BIN

Getting Help

If you encounter issues:

1. Check the Log tab for detailed error messages
2. Look up error-details in the Key-Value Store
3. Review the Dataset for partial results
4. Contact support with your Run ID

📚 Related Actors

• Text to Speech - Add AI-generated voiceovers to your videos
• Image to Video - Convert images into video content
• Video Transcription - Extract text from videos

🔗 Links

• NVIDIA NIM Documentation
• Remotion Documentation
• Apify Actors Documentation

---

Built with ❤️ using Apify, NVIDIA NIM, and Remotion