Fix incorrect tile_latent_min_width calculations (#11305)

src/diffusers/models/autoencoders/autoencoder_kl_hunyuan_video.py

@@ -829,7 +829,7 @@ class AutoencoderKLHunyuanVideo(ModelMixin, ConfigMixin):
     def _decode(self, z: torch.Tensor, return_dict: bool = True) -> Union[DecoderOutput, torch.Tensor]:
         batch_size, num_channels, num_frames, height, width = z.shape
         tile_latent_min_height = self.tile_sample_min_height // self.spatial_compression_ratio
-        tile_latent_min_width = self.tile_sample_stride_width // self.spatial_compression_ratio
+        tile_latent_min_width = self.tile_sample_min_width // self.spatial_compression_ratio
         tile_latent_min_num_frames = self.tile_sample_min_num_frames // self.temporal_compression_ratio
 
         if self.use_framewise_decoding and num_frames > tile_latent_min_num_frames:

src/diffusers/models/autoencoders/autoencoder_kl_ltx.py

@@ -1285,7 +1285,7 @@ class AutoencoderKLLTXVideo(ModelMixin, ConfigMixin, FromOriginalModelMixin):
     ) -> Union[DecoderOutput, torch.Tensor]:
         batch_size, num_channels, num_frames, height, width = z.shape
         tile_latent_min_height = self.tile_sample_min_height // self.spatial_compression_ratio
-        tile_latent_min_width = self.tile_sample_stride_width // self.spatial_compression_ratio
+        tile_latent_min_width = self.tile_sample_min_width // self.spatial_compression_ratio
         tile_latent_min_num_frames = self.tile_sample_min_num_frames // self.temporal_compression_ratio
 
         if self.use_framewise_decoding and num_frames > tile_latent_min_num_frames:

src/diffusers/models/autoencoders/autoencoder_kl_magvit.py

@@ -887,7 +887,7 @@ class AutoencoderKLMagvit(ModelMixin, ConfigMixin):
     def _decode(self, z: torch.Tensor, return_dict: bool = True) -> Union[DecoderOutput, torch.Tensor]:
         batch_size, num_channels, num_frames, height, width = z.shape
         tile_latent_min_height = self.tile_sample_min_height // self.spatial_compression_ratio
-        tile_latent_min_width = self.tile_sample_stride_width // self.spatial_compression_ratio
+        tile_latent_min_width = self.tile_sample_min_width // self.spatial_compression_ratio
 
         if self.use_tiling and (z.shape[-1] > tile_latent_min_height or z.shape[-2] > tile_latent_min_width):
             return self.tiled_decode(z, return_dict=return_dict)