TAAFT
Free mode
100% free
Freemium
Free Trial
Deals

VOID

By Netflix
VOID, short for Video Object and Interaction Deletion, is a Netflix research model for realistic video editing through interaction-aware object removal. The repository says it is fine-tuned on top of CogVideoX for video inpainting with mask conditioning, and is specifically designed to delete both an object and the scene interactions it induces, including physical consequences like a guitar falling after the person holding it is removed. It uses two sequential transformer checkpoints, with Pass 1 providing base inpainting and Pass 2 refining temporal consistency. The repo also notes a mask-generation stage that uses Gemini plus SAM2, and recommends a 40GB+ GPU for the included notebook workflow.
New Multimodal Gen 3
Released: April 3, 2026

Overview

VOID is Netflix’s open-source video object and interaction deletion model for interaction-aware video inpainting. It removes not only the target object from a video, but also the physical and visual effects that object causes in the scene, such as shadows, reflections, or objects that should move after the removal. It is built on top of CogVideoX and uses a two-pass transformer pipeline for higher temporal consistency.

About Netflix

Netflix is one of the world's leading entertainment services with over 247 million paid memberships in over 190 countries. It offers TV series, films, and games across a wide variety of genres and languages, including original content production. The company operates streaming, advertising, and gaming businesses.

Industry: Entertainment
Company Size: 14000
Location: Los Gatos, California, US
Website: netflix.com
View Company Profile

Tools using VOID

No tools found for this model yet.

Last updated: April 3, 2026
0 AIs selected
Clear selection
#
Name
Task