TurboDiffusion sampling parameters.
TurboDiffusion uses RCM (recurrent Consistency Model) scheduler for 1-4 step video generation with no classifier-free guidance.
Classes¶
dataclass
¶
TurboDiffusionI2V_A14B_SamplingParam(data_type: str = 'video', image_path: str | None = None, pil_image: Any | None = None, video_path: str | None = None, mouse_cond: Any | None = None, keyboard_cond: Any | None = None, grid_sizes: Any | None = None, refine_from: str | None = None, t_thresh: float = 0.5, spatial_refine_only: bool = False, num_cond_frames: int = 0, stage1_video: Any | None = None, prompt: str | list[str] | None = None, negative_prompt: str | None = None, prompt_path: str | None = None, output_path: str = 'outputs/', output_video_name: str | None = None, num_videos_per_prompt: int = 1, seed: int = 1024, num_frames: int = 81, num_frames_round_down: bool = False, height: int = 720, width: int = 1280, fps: int = 16, num_inference_steps: int = 4, guidance_scale: float = 1.0, guidance_rescale: float = 0.0, boundary_ratio: float | None = None, sigmas: list[float] | None = None, enable_teacache: bool = False, save_video: bool = True, return_frames: bool = False, return_trajectory_latents: bool = False, return_trajectory_decoded: bool = False)
Bases: SamplingParam
Sampling parameters for TurboDiffusion I2V A14B model.
Uses 4-step RCM sampling with dual-model switching (high/low noise).
dataclass
¶
TurboDiffusionT2V_14B_SamplingParam(data_type: str = 'video', image_path: str | None = None, pil_image: Any | None = None, video_path: str | None = None, mouse_cond: Any | None = None, keyboard_cond: Any | None = None, grid_sizes: Any | None = None, refine_from: str | None = None, t_thresh: float = 0.5, spatial_refine_only: bool = False, num_cond_frames: int = 0, stage1_video: Any | None = None, prompt: str | list[str] | None = None, negative_prompt: str | None = None, prompt_path: str | None = None, output_path: str = 'outputs/', output_video_name: str | None = None, num_videos_per_prompt: int = 1, seed: int = 1024, num_frames: int = 81, num_frames_round_down: bool = False, height: int = 720, width: int = 1280, fps: int = 16, num_inference_steps: int = 4, guidance_scale: float = 1.0, guidance_rescale: float = 0.0, boundary_ratio: float | None = None, sigmas: list[float] | None = None, enable_teacache: bool = False, save_video: bool = True, return_frames: bool = False, return_trajectory_latents: bool = False, return_trajectory_decoded: bool = False)
Bases: SamplingParam
Sampling parameters for TurboDiffusion T2V 14B model.
Uses 4-step RCM sampling with guidance_scale=1.0 (no CFG).
dataclass
¶
TurboDiffusionT2V_1_3B_SamplingParam(data_type: str = 'video', image_path: str | None = None, pil_image: Any | None = None, video_path: str | None = None, mouse_cond: Any | None = None, keyboard_cond: Any | None = None, grid_sizes: Any | None = None, refine_from: str | None = None, t_thresh: float = 0.5, spatial_refine_only: bool = False, num_cond_frames: int = 0, stage1_video: Any | None = None, prompt: str | list[str] | None = None, negative_prompt: str | None = None, prompt_path: str | None = None, output_path: str = 'outputs/', output_video_name: str | None = None, num_videos_per_prompt: int = 1, seed: int = 1024, num_frames: int = 81, num_frames_round_down: bool = False, height: int = 480, width: int = 832, fps: int = 16, num_inference_steps: int = 4, guidance_scale: float = 1.0, guidance_rescale: float = 0.0, boundary_ratio: float | None = None, sigmas: list[float] | None = None, enable_teacache: bool = False, save_video: bool = True, return_frames: bool = False, return_trajectory_latents: bool = False, return_trajectory_decoded: bool = False)
Bases: SamplingParam
Sampling parameters for TurboDiffusion T2V 1.3B model.
Uses 4-step RCM sampling with guidance_scale=1.0 (no CFG).