Skip to content

vllm_omni.diffusion.models.ltx2.ltx2_conditioning

Task-specific conditioning shared by LTX model versions.

LTXI2VConditioningMixin

First-frame conditioning with optional unified T2V/I2V dispatch.

support_image_input class-attribute instance-attribute 

support_image_input = True

unified_text_image_entry class-attribute instance-attribute 

unified_text_image_entry = False

video_processor instance-attribute 

video_processor = VideoProcessor(
    vae_scale_factor=self.vae_spatial_compression_ratio,
    resample="bilinear",
)

prepare_latents 

prepare_latents(
    image: Tensor | None = None,
    batch_size: int = 1,
    num_channels_latents: int = 128,
    height: int = 512,
    width: int = 768,
    num_frames: int = 121,
    noise_scale: float = 0.0,
    dtype: dtype | None = None,
    device: device | None = None,
    generator: Generator | list[Generator] | None = None,
    latents: Tensor | None = None,
) -> Tensor | tuple[Tensor, Tensor]

LTXPromptContext dataclass

Connector outputs consumed by an LTX denoise phase.

batch_size instance-attribute 

batch_size: int

connector_attention_mask instance-attribute 

connector_attention_mask: Tensor

connector_audio_prompt_embeds instance-attribute 

connector_audio_prompt_embeds: Tensor

connector_prompt_embeds instance-attribute 

connector_prompt_embeds: Tensor

negative_connector_attention_mask instance-attribute 

negative_connector_attention_mask: Tensor | None

negative_connector_audio_prompt_embeds instance-attribute 

negative_connector_audio_prompt_embeds: Tensor | None

negative_connector_prompt_embeds instance-attribute 

negative_connector_prompt_embeds: Tensor | None

positive_connector_attention_mask instance-attribute 

positive_connector_attention_mask: Tensor

positive_connector_audio_prompt_embeds instance-attribute 

positive_connector_audio_prompt_embeds: Tensor

positive_connector_prompt_embeds instance-attribute 

positive_connector_prompt_embeds: Tensor

LTXTextConditioningMixin

Shared Gemma encoding and text connector orchestration.

connector_batches_cfg class-attribute instance-attribute 

connector_batches_cfg = False

encode_prompt 

encode_prompt(
    prompt: str | list[str] | None,
    negative_prompt: str | list[str] | None = None,
    do_classifier_free_guidance: bool = True,
    num_videos_per_prompt: int = 1,
    prompt_embeds: Tensor | None = None,
    negative_prompt_embeds: Tensor | None = None,
    prompt_attention_mask: Tensor | None = None,
    negative_prompt_attention_mask: Tensor | None = None,
    max_sequence_length: int = 1024,
    scale_factor: int = 8,
    device: device | None = None,
    dtype: dtype | None = None,
) -> tuple[Tensor, Tensor, Tensor | None, Tensor | None]