Text Prompts

When to Use

• Crowded scenes where default detection picks wrong subject.
• Product videos with specific object isolation needs.
• Character videos requiring accessory preservation.

Prompt Writing Tips

1. Use concrete noun first: person, dog, red car.
2. Add unique descriptors: clothing, color, position.
3. Avoid long narrative sentences.
4. If unstable, simplify prompt to core object description.

Request Example

curl -X POST https://api.removebgvideo.com/v1/jobs \
  -H "X-Api-Key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "video_url": "https://cdn.example.com/input.mp4",
    "model": "pro",
    "text_prompt": "person wearing black hoodie",
    "background": { "type": "transparent" },
    "output_format": "webm",
    "auto_start": true
  }'

Good Prompt Examples

• person wearing red jacket
• golden retriever dog
• blue sports car
• wooden dining table
• anime character with green dress and staff

Quality Guardrails

• Prefer representative source clips when tuning model defaults.
• Keep input compression moderate to preserve edge details.
• Use explicit output format policy for transparent vs non-transparent workflows.
• Run A/B validation with the same clip set before changing defaults.

Implementation Checklist

1. Define payload schema validation in backend before forwarding requests.
2. Store model/output_format/background settings with each job record.
3. Add internal quality review for difficult scenes (hair, glass, motion blur).
4. Create runbook entries for model-specific failure cases.

Prompt Debugging Playbook

• Start with concise noun phrases, then add qualifiers.
• Avoid contradictory descriptors in one prompt.
• Use comma-separated target objects for multi-subject retention.
• Keep prompt stable across retries to isolate model behavior changes.