search

mihirp1998 / VADER

Video Diffusion Alignment via Reward Gradients. We improve a variety of video diffusion models such as VideoCrafter, OpenSora, ModelScope and StableVideoDiffusion by finetuning them using various reward models such as HPS, PickScore, VideoMAE, VJEPA, YOLO, Aesthetics etc.
https://vader-vid.github.io/
210 stars 14 forks source link

Is Reward model about Temporal Consistency V-JEPA suitable for opensora? #10

Open csyufei opened 2 months ago

csyufei commented 2 months ago

Hi, excellent works! However, I want to know whether the reward model Temporal Consistency via V-JEPA is suitable for T2V models such as opensora, videocrafter, modelscope. I saw there is no reward model about temporal consistency in vader-opensora and vader-video crafter in the corresponding code. Thanks.

QinOwen commented 2 months ago

Hi. I believe it would work for these T2V models.