From baf23e2da3f0816d1ebe870ccd66249fa3e5ceaa Mon Sep 17 00:00:00 2001 From: Daniel Gu Date: Wed, 17 Dec 2025 11:14:45 +0100 Subject: [PATCH] Explicitly specify temporal and spatial VAE scale factors when converting --- scripts/convert_ltx2_to_diffusers.py | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/scripts/convert_ltx2_to_diffusers.py b/scripts/convert_ltx2_to_diffusers.py index dfec0262de..85fa169af3 100644 --- a/scripts/convert_ltx2_to_diffusers.py +++ b/scripts/convert_ltx2_to_diffusers.py @@ -241,6 +241,8 @@ def get_ltx2_video_vae_config(version: str) -> Tuple[Dict[str, Any], Dict[str, A "resnet_norm_eps": 1e-6, "encoder_causal": True, "decoder_causal": True, + "spatial_compression_ratio": 32, + "temporal_compression_ratio": 8, }, } rename_dict = LTX_2_0_VIDEO_VAE_RENAME_DICT @@ -274,6 +276,8 @@ def get_ltx2_video_vae_config(version: str) -> Tuple[Dict[str, Any], Dict[str, A "resnet_norm_eps": 1e-6, "encoder_causal": True, "decoder_causal": True, + "spatial_compression_ratio": 32, + "temporal_compression_ratio": 8, }, } rename_dict = LTX_2_0_VIDEO_VAE_RENAME_DICT