Bria Video Eraser Mask

Bria Video Eraser Mask

Playground

Bria Video Eraser (Mask) removes unwanted objects from videos using a user-provided mask video. Mark regions frame-by-frame (black/white or alpha), and the model performs AI video inpainting to reconstruct clean, temporally consistent backgrounds for people, logos, text, and props. Ready-to-use REST API with fast response, best performance, no cold starts, and affordable pricing.

Features

Bria Video Eraser (Mask-Based) is a precision video inpainting tool that removes objects using a mask video. Provide the original video plus a matching mask video (white = erase, black = keep), and the model removes the masked regions frame-by-frame while reconstructing the background for clean, production-ready results.

This mode is ideal for creators and post-production teams who need pixel-accurate control over what gets removed—without manual frame-by-frame painting.

Why it stands out

• Mask-based control for precise, frame-consistent object removal.
• Clean inpainting to reconstruct backgrounds after erasing masked regions.
• Works well for VFX cleanup, unwanted object removal, and scene polishing.
• copy_audio toggle to preserve or remove the original audio track.
• Transparent, duration-based pricing based on BRIA official per-second rates.

Capabilities

• Mask-video guided object removal (white regions removed, black regions preserved)
• Background reconstruction (video inpainting) across frames
• Frame-by-frame precision for moving objects and tracked masks
• Optional original audio preservation via copy_audio

Parameters

How to use

1. Upload the video you want to edit (or paste a public URL).
2. Upload a matching mask_video:

• White areas = regions to erase
• Black areas = regions to keep

3. Make sure the mask video matches the input video’s resolution, duration, and timing for best results.
4. Choose copy_audio:

• true to preserve the original audio track
• false to output a silent video

5. Run the model, preview the output, and refine the mask if needed.

Mask tips (best practices)

• Use pure white for removal and pure black for preservation whenever possible.
• Slight edge feathering can reduce hard seams and improve blending.
• If the mask “misses” parts of the object during motion, expand the mask slightly and re-run.
• For complex motion, generate masks with rotoscoping tools or segmentation models, then export as a mask video.

Pricing

Pricing is $0.05 per second of input video duration.

Current endpoint supports videos up to 5 seconds per request.

Notes

• The mask video must be properly aligned; misaligned masks can cause jittery edges or incomplete removals.
• Fast motion blur and heavy occlusion may reduce inpainting quality—use tighter masks and short test clips first.
• If you need quick removal without mask prep, use the prompt-based variant instead.
• If you want to generate a mask video. Please use the SAM-3 Video to build one!

Related Models

• Bria Video Eraser (Prompt) — Prompt-based video object removal for quick, text-driven cleanup without mask preparation.

• bria/remove-background — Fast image background removal for product photos, portraits, and design cutouts.

Authentication

For authentication details, please refer to the Authentication Guide.

API Endpoints

Submit Task & Query Result

set -euo pipefail

export WAVESPEED_API_KEY="your-api-key"

REQUEST_BODY=$(cat <<'JSON'
{
  "video": "https://interactive-examples.mdn.mozilla.net/media/cc0-videos/flower.mp4",
  "mask_video": "https://interactive-examples.mdn.mozilla.net/media/cc0-videos/flower.mp4",
  "copy_audio": true
}
JSON
)

# 1. Submit the prediction.
SUBMIT_RESPONSE=$(curl --silent --show-error --fail-with-body \
  -X POST "https://api.wavespeed.ai/api/v3/bria/video-eraser/mask" \
  -H "Authorization: Bearer ${WAVESPEED_API_KEY}" \
  -H "Content-Type: application/json" \
  -d "${REQUEST_BODY}")

TASK=$(printf '%s' "${SUBMIT_RESPONSE}" | jq 'if type == "object" and has("data") then .data else . end')
PREDICTION_ID=$(printf '%s' "${TASK}" | jq -r '.id // empty')
if [ -z "${PREDICTION_ID}" ]; then
  printf 'Submission response did not contain a prediction id
' >&2
  exit 1
fi
RESULT_URL=$(printf '%s' "${TASK}" | jq -r '.urls.get // empty')
if [ -z "${RESULT_URL}" ]; then RESULT_URL="https://api.wavespeed.ai/api/v3/predictions/${PREDICTION_ID}/result"; fi

# 2. Poll until the prediction finishes.
while true; do
  RESPONSE=$(curl --silent --show-error --fail-with-body \
    "${RESULT_URL}" \
    -H "Authorization: Bearer ${WAVESPEED_API_KEY}")
  RESULT=$(printf '%s' "${RESPONSE}" | jq 'if type == "object" and has("data") then .data else . end')
  STATUS=$(printf '%s' "${RESULT}" | jq -r '.status // empty')

  case "${STATUS}" in
    completed) printf '%s\n' "${RESULT}" | jq '.outputs'; break ;;
    failed|cancelled|timeout) printf '%s\n' "${RESULT}" | jq . >&2; exit 1 ;;
    created|processing) sleep 2 ;;
    *) printf 'Unexpected status: %s
' "${STATUS}" >&2; exit 1 ;;
  esac
done

Parameters

Task Submission Parameters

Request Parameters

Response Parameters

Result Request Parameters

Result Response Parameters