Void Video Inpainting

All video to video models

Video to Video

VOID removes objects from videos along with all interactions they induce on the scene

Inference$ Commercial useFrom 2 credits per run

About Void Video Inpainting

VOID removes objects from videos along with all interactions they induce on the scene

utilityediting

Pricing

A run with the default settings costs 2 credits. The exact cost updates in the playground as you change settings, and you're only charged when you press Run. Failed runs are refunded automatically.

Inputs

video_urlfilerequired

URL of the input video containing the object to remove.

quad_mask_video_urlfiledefault: ""

URL of a mask video for the removal target. For best results this should be a VOID-style quadmask video with 4 grayscale values: 0=object to remove, 63=overlap, 127=affected region, 255=background to keep. A simple binary mask (0=remove, 255=keep) also works. If omitted, the app will generate a temporary mask video from `mask_prompt` using the SAM-3 video endpoint and convert it to a quadmask automatically.

mask_prompttextarea

Text description of what should be masked in the input video, such as the object or person to remove. Used to generate a temporary mask video with SAM-3 when `quad_mask_video_url` is not provided.

prompttextarearequired

Text description of the desired background after object removal.

enable_pass2_refinementbooleandefault: false

Run VOID Pass 2 warped-noise refinement after Pass 1. This is slower but can improve temporal consistency on longer clips.

negative_prompttextareadefault: "The video is not of a high quality, it has a low resolution. Watermark present in each frame. The background is solid. Strange body and strange trajectory. Distortion."

Negative prompt to guide generation away from undesired outputs.

num_inference_stepsintegerdefault: 30

Number of denoising steps. Higher values improve quality but increase latency.

guidance_scalenumberdefault: 1

Classifier-free guidance scale.

strengthnumberdefault: 1

Denoising strength. 1.0 means full denoising.

seedinteger

Random seed for reproducibility.

num_framesintegerdefault: 85

Temporal window size for inference. The backend snaps this to the nearest CogVideoX-safe value that works with temporal compression and patching. Valid outputs are 69, 77, 85, ..., 197.

enable_safety_checkerbooleandefault: true

Enable content safety checking on the output.

Related video to video models

Wan 3.0

Wan 3.0 is the latest generation AI video model, delivering enhanced motion smoothness, superior scene fidelity, and greater visual coherence.

stylizedtransform

video-to-video

✦ new

Kling 4.0 Reference to Video

Kling 4.0 generates videos from text prompts and image or video references at 720p or 1080p, with durations from 3 to 30 seconds and per-shot prompts.

reference-to-videomulti-shot

video-to-video

sync-3 Lipsync

sync-3 most powerful lipsync model yet, featuring native visual intelligence for professional-quality video.

stylizedtransform

video-to-video

Kling O3 Edit Video [Pro]

Edit videos using Kling O3 from Kling Team!

video-to-video

Sync Lipsync 2.0

Generate realistic lipsync animations from audio using advanced algorithms for high-quality synchronization with Sync Lipsync 2.0 model

animationlip sync

video-to-video

✦ new

Kling 4.0 Video Editing

Kling 4.0 edits an input video using text instructions and optional image references, producing 3 to 30-second videos at 720p or 1080p.

video-editingimage-reference

video-to-video

Wan 3.0 Prime

Wan 3.0 Prime Reference-to-Video combines reference images, videos, and audio into a unified video with fast generation and strong multimodal coherence. It follows character identity, visual style, movement, and sound cues across references to create controlled, consistent, and production-ready resu

referencevideo

video-to-video

Topaz Upscale Video Precision

Professional video upscaling powered by Topaz Labs. Precision models (Proteus, Artemis, Iris, Dione, Theia, Gaia, Rhea) enhance footage up to 4x while staying faithful to the source. Best for clean, natural upscales of real-world footage.

upscalevideo

video-to-video

FFmpeg API Compose

Compose videos from multiple media sources using FFmpeg API.

ffmpeg

video-to-video

Kling Video

Transfer movements from a reference video to any character image. Cost-effective mode for motion transfer, perfect for portraits and simple animations.

stylizedtransform

video-to-video

H3 Max Recast

Recast the people in a video using reference photos with H3 Max, while preserving the source motion, camera, cuts, and audio.

video-editingcharacter-replacement

video-to-video

Ffmpeg Api Merge Audio-Video

Merge videos with standalone audio files or audio from video files.

ffmpeg

video-to-video