Video Generation
Create AI-generated videos with Sora and Veo models
Video Generation
Video generation in TractionDesk leverages cutting-edge AI models to create short video clips from text descriptions. The platform supports two world-class video generation models: OpenAI's Sora 2 (default) and Google's Veo 3. These models can create product demonstrations, animated concepts, motion graphics, lifestyle scenes, and more—all from simple text prompts. Video generation opens up possibilities that were previously impossible or prohibitively expensive, letting you create professional video content without cameras, actors, or video editing software.
Sora 2 is the default model and produces high-quality, realistic videos with excellent motion coherence and temporal consistency. It excels at creating smooth, natural movements and can generate complex scenes with multiple subjects and dynamic camera work. Veo 3 offers alternative visual styles and is particularly strong for certain types of creative or stylized content. Most users start with Sora 2 and only switch to Veo 3 for specific use cases where its characteristics better match their needs.
Video generation is the most credit-intensive operation in TractionDesk, costing 20 credits per video. This reflects the computational complexity—each second of video requires generating multiple high-resolution frames and ensuring temporal coherence between them. Generation time ranges from 60 to 120 seconds depending on video length and model load. Because videos are resource-intensive, we recommend testing with shorter durations (5 seconds) initially, then increasing length once you've refined your prompts and presets.
Creating Your First Video
Navigate to the Videos section and click "Generate New Video." Select a video preset if you have one, or enter a custom prompt describing the video you want. Specify the duration in seconds (typically 5-10 seconds for social media clips) and select an aspect ratio: 16:9 for YouTube and landscape displays, 9:16 for Instagram Stories and TikTok, or 1:1 for square social posts. Then click "Generate Video."
The generation process takes 1-2 minutes, and you'll see a progress indicator showing the current status. Sora 2 videos may take slightly longer than Veo 3, but both deliver high-quality results. Once complete, the video player displays your generated clip with controls to play, pause, and scrub through frames. You can download the video file (MP4 format) or save it directly to your asset library for future use.
Video prompts should be descriptive and specific about both content and motion. For example: "A smartphone rotating slowly on a white surface, spotlighting its camera array, smooth 360-degree rotation, professional product videography, studio lighting." The more detail about movement, camera angles, and scene composition, the better the AI can interpret your vision. Unlike static images, videos benefit from explicit instructions about what should move and how.
Video Presets for Brand Consistency
Video presets work similarly to image presets but focus on video-specific characteristics like motion style, camera work, pacing, and scene transitions. A well-designed video preset ensures all your video content shares a cohesive visual language. For example, a "Product Demo" preset might specify: "Clean professional product videography: white backgrounds, slow deliberate camera movements, studio lighting, focus on product details, smooth transitions, commercial advertising quality."
Create different presets for different video purposes. A "Social Media Clips" preset might emphasize quick cuts, vibrant colors, and dynamic motion for attention-grabbing content. An "Explainer Videos" preset might focus on clear visuals, smooth transitions, and educational presentation. A "Testimonial Style" preset could define a warm, authentic aesthetic with natural lighting and documentary-style camera work.
When creating video presets, include specifications about duration preferences, aspect ratios you typically use, motion characteristics (fast-paced vs slow and deliberate), lighting style (dramatic vs flat), and overall mood. The more comprehensive your preset, the more consistent your video output across different prompts.
Aspect Ratios and Platform Optimization
Choosing the right aspect ratio is crucial for where you'll publish your videos. 16:9 (landscape) is standard for YouTube, website embeds, presentations, and horizontal displays. It provides a cinematic feel and works well for product demonstrations and tutorials. 9:16 (portrait) is optimized for Instagram Stories, TikTok, YouTube Shorts, and mobile-first platforms where vertical video dominates. This aspect ratio maximizes screen real estate on smartphones. 1:1 (square) works across most platforms and is particularly effective for Instagram feed posts and Facebook, offering a balanced composition that works on both mobile and desktop.
Consider your distribution channels when selecting aspect ratio. If you're creating video for a specific platform, match that platform's preferred format. If you need the same video for multiple platforms, generate it in each aspect ratio—the AI will adapt the composition for each format rather than just cropping, ensuring optimal presentation everywhere.
Using Videos in Campaigns
Videos are powerful campaign deliverables that significantly boost engagement and conversion rates. When you include video as a campaign deliverable, the Campaign Orchestrator generates videos that align with the campaign's topic and complement any copy that was created earlier. For example, if a campaign created a blog post about "10 Productivity Hacks for Remote Workers," the orchestrator might generate videos demonstrating each hack visually.
You can specify video quantity in campaign deliverables. If your campaign description mentions "create 3 videos," the orchestrator will generate three separate videos, each exploring different aspects of the campaign topic. This is useful for content series or when you need multiple assets for different platforms or ad variations.
Campaign videos are stored with metadata linking them to their source campaign, making it easy to track which marketing initiative generated which assets. This organizational structure helps with performance analysis—if a campaign-generated video performs exceptionally well, you can trace it back to understand what prompt and preset combination created it, then replicate that success in future campaigns.
Sora 2 vs Veo 3
Both models are capable of creating excellent videos, but they have different strengths. Sora 2 produces more realistic, photographic-style videos with exceptional motion quality. It's the best choice for product demonstrations, realistic scenes, and content that needs to look like traditional video footage. Sora 2 excels at complex scenes with multiple subjects and accurate physics simulation.
Veo 3 offers strong creative and stylized video generation. Some users prefer its color grading and artistic interpretation for certain content types. It's particularly effective for animated or illustrative content. Veo 3 also provides more granular control over certain generation parameters like negative prompts (things you don't want in the video) and random seeds (for reproducible results).
For most users, starting with Sora 2 is recommended because it's the default, requires no special configuration, and delivers excellent results across a wide range of use cases. Experiment with Veo 3 if you have specific stylistic needs or if Sora 2's output doesn't match your vision. You can specify the model in your preset configuration or when using the Voice Agent ("create a video using Veo 3").
Technical Specifications
Generated videos have these specifications:
- Duration: 5-10 seconds (configurable)
- Resolution: Up to 1080p (1920x1080) for 16:9, 1080x1920 for 9:16, 1080x1080 for 1:1
- Format: MP4 with H.264 video codec
- Frame Rate: 24 or 30 fps
- File Size: Typically 2-10MB depending on duration and complexity
- Audio: Silent (no audio track included)
Videos are stored in cloud storage and accessible via secure URLs. Downloads provide the full-resolution MP4 file compatible with all major video editing software and social media platforms.
Best Practices
Start with Shorter Durations: 5-second videos generate faster and consume the same credits as 10-second videos. Test your prompts with shorter clips before creating longer content.
Describe Motion Explicitly: Don't just describe what's in the scene—describe how it moves. "Camera slowly pushes in toward the subject" or "Product rotates 360 degrees clockwise" gives the AI clear direction.
Use Cinematic Language: Terms like "dolly shot," "pan left," "rack focus," "slow motion," and "time lapse" help the AI understand the cinematography you want.
Keep Scenes Simple: While the models can handle complex scenes, simpler compositions often produce better results. Focus on one primary subject or action rather than trying to include everything in one video.
Test Before Batch Generation: Generate one video to validate your prompt and preset before creating multiple videos or including video in a large campaign. This saves credits and time.