VPA Logo

Video Analysis

Upload and analyze video files to generate AI-optimized prompts.

Video Analysis is VPA's core feature. Upload any video file and our AI will analyze its visual characteristics to generate prompts optimized for your chosen AI video generator.

Uploading a Video

To analyze a video, navigate to the Analysis page and select the "Upload Video" input type. You can upload videos by:

  • Drag and drop: Drag your video file directly onto the upload area
  • Click to browse: Click the upload area to open your file browser
  • Paste: Copy a video file and paste it into the upload area

Supported Formats

  • MP4 (recommended)
  • WebM
  • MOV
  • AVI
💡

File size limit

Maximum file size is 100MB. For larger files, consider trimming to the most representative section or using YouTube upload and the YouTube Analysis feature.

Frame Count Selection

Choose how many frames VPA should extract and analyze from your video. More frames provide better understanding of motion and temporal changes, but increase processing time and cost.

  • 2 frames: Fast analysis, best for static or simple videos
  • 4 frames: Balanced analysis for most use cases (recommended)
  • 6 frames: Detailed analysis for complex motion
  • 8 frames: Maximum detail for highly dynamic content

Choosing Your Target Generator

Select the AI video generator you plan to use. VPA will optimize the prompt structure, vocabulary, and emphasis for that specific generator:

  • Sora (OpenAI): Up to 1000 characters, technical vocabulary
  • Veo (Google): Up to 800 characters, balanced approach
  • Runway Gen-3: Up to 500 characters, concise and direct
  • Kling: Up to 600 characters, descriptive style

Understanding the Results

After analysis, VPA provides:

  • Generated Prompt: The main output, ready to copy into your generator
  • Visual Analysis Breakdown: Detailed breakdown of detected elements
  • Refinement Options: One-click adjustments to modify the prompt
  • Save to Scene: Option to save this prompt to your project
✨

Pro tip

For best results, use videos that are 5-60 seconds long with consistent lighting and clear subject matter. Avoid videos with rapid scene changes unless you want to capture that style.

Video Preview

After uploading, you'll see a preview of your video with the extracted frames displayed below. This helps you verify the right moments are being analyzed. If the frames don't look representative, try adjusting the frame count or trimming your video.

Best Practices

  • Use high-quality source videos when possible
  • Trim to the most visually interesting section
  • Avoid heavily compressed or low-resolution footage
  • Consider the lighting and color consistency of your source
  • Match your source video style to your intended output style