MV-Forcing: Long Multi-View Video Generation via
4D-Grounded Spatio-Temporal Self-Forcing

Supplementary Material

Please click on the sections below to view our interactive results, comparisons, and ablations:

• Our Results
• Real-World Results
• Short Sequence Comparisons
• Long Sequence Comparisons
• Component Ablations
• View Scaling Ablations
• Temporal Scaling Ablations

Real-World Results

We demonstrate generalization of MV-Forcing to real-world data (Sec. 3.5 in the paper). We replace the synthetic SynCamMaster teacher with a finetuned ReCamMaster and train the view-sequential conditioning path on real-world videos from the Mixkit dataset. Starting from our synthetic-trained model, only 500 finetuning iterations are needed. At inference, Self-Forcing generates the first view from text, and our student autoregressively generates subsequent views. Hover over any video to play.

"A woman in a red sweater, Christmas hat, and blue jeans sitting on the floor holding a present in front of a brick fireplace and decorated Christmas tree."

"A curly-haired boy and a girl with long hair playing a video game in a colorful indoor arcade with wooden ceiling beams and vibrant game displays."

"A woman with dark hair and makeup turning her head in a dramatic black and white close-up portrait."

"A pedestrian crossing signal displaying red and green walking figures on a city street pole with buildings and trees in the blurred background."

"A couple sitting on a blanket on a sandy beach, the woman in a white dress gesturing towards the ocean waves under a sunset."

"A bearded man with curly hair in a white collared shirt talking on the phone against a green screen background."

"A close-up of a burning fire with intense orange and yellow flames flickering and swirling against a dark smoky background."

"A person wearing a protective mask picking up a small object."

"A colorful globe spinning against a black background."

"A woman in a red and white Christmas sweater holding a small ornament in a decorated living room with lights and a Christmas tree."