[Diffusion] Add LTX-2.5 Diffusion VAE decoder support (#6189)

Signed-off-by: mglyn <1203789601@qq.com>
Co-authored-by: Hongsheng Liu <liuhongsheng4@huawei.com>
This commit is contained in:
Mu GuanLinandHongsheng Liu authored and GitHub committed 2026-08-27 23:54:40 +08:00
1 parent 262ff14693
commit 5e4daaf197
22 files changed
+1832 -61

No files matched your search

@@ -1,5 +1,5 @@
# SPDX-License-Identifier: Apache-2.0
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
"""Request normalization and validation shared by LTX pipeline variants."""
@@ -86,6 +86,7 @@ def validate_pipeline_request(
vae_spatial_compression_ratio: int,
vae_temporal_compression_ratio: int,
pipeline_name: str,
min_video_latent_spatial_size: tuple[int, int] | None = None,
request_sigmas: list[float] | None = None,
request_phase_sigmas: tuple[list[float] | None, ...] | None = None,
) -> None:
@@ -97,6 +98,18 @@ def validate_pipeline_request(
f"{request_inputs.width}x{request_inputs.height}."
)
if min_video_latent_spatial_size is not None:
min_latent_height, min_latent_width = min_video_latent_spatial_size
latent_height = request_inputs.height // vae_spatial_compression_ratio
latent_width = request_inputs.width // vae_spatial_compression_ratio
if latent_height < min_latent_height or latent_width < min_latent_width:
raise ValueError(
f"{pipeline_name} DiffVAE requires video latent spatial dimensions of at least "
f"{min_latent_width}x{min_latent_height} for neighborhood attention, got "
f"{latent_width}x{latent_height} from request resolution "
f"{request_inputs.width}x{request_inputs.height}."
)
if request_inputs.num_frames < 1 or (request_inputs.num_frames - 1) % vae_temporal_compression_ratio != 0:
raise ValueError(
f"{pipeline_name} `num_frames` must be {vae_temporal_compression_ratio} * k + 1, "