LightX2V
A lightweight any-modality-to-video (X2V) generation acceleration model supporting tasks such as Image-to-Video and Text-to-Video. It is designed to reduce parameter count and inference VRAM consumption, making it suitable for edge deployment.
Timeline (last 90 days)
LightX2V tested the base model on an RTX 5090, with approximately 5.586 seconds at 1024×1024 and approximately 30.333 seconds at 2048×2048 (40 steps, CFG disabled).
The LightX2V team released a new version of Minimax-h3-Turbo, FL2V Turbo 4-step v1.2 (768p), on Hugging Face—an accelerated LoRA designed specifically for video generation models.
The LightX2V team released a new version of Minimax-h3-Turbo on Hugging Face — FL2V Turbo 4-step v1.2 (768p) — an accelerated LoRA that claims to complete video generation in just 4 inference steps.
The author recommends the Base + LightX2V 4-step turbo combination for quick drafts in order to preview prompt effects.
All Facts (3)
LightX2V tested the base model on an RTX 5090, with approximately 5.586 seconds at 1024×1024 and approximately 30.333 seconds at 2048×2048 (40 steps, CFG disabled).
50%UnverifiedThe author recommends the Base + LightX2V 4-step turbo combination for quick drafts in order to preview prompt effects.
50%UnverifiedThe LightX2V team released a new version of Minimax-h3-Turbo on Hugging Face — FL2V Turbo 4-step v1.2 (768p) — an accelerated LoRA that claims to complete video generation in just 4 inference steps.
50%