add comments to explain the code better

src/diffusers/pipelines/alt_diffusion/pipeline_alt_diffusion.py

@@ -198,7 +198,9 @@ class AltDiffusionPipeline(DiffusionPipeline, TextualInversionLoaderMixin, LoraL
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 

src/diffusers/pipelines/alt_diffusion/pipeline_alt_diffusion_img2img.py

@@ -236,7 +236,9 @@ class AltDiffusionImg2ImgPipeline(
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 

src/diffusers/pipelines/animatediff/pipeline_animatediff.py

@@ -129,7 +129,9 @@ class AnimateDiffPipeline(DiffusionPipeline, TextualInversionLoaderMixin, LoraLo
             motion_adapter=motion_adapter,
             scheduler=scheduler,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
 
     # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.encode_prompt with num_images_per_prompt -> num_videos_per_prompt

src/diffusers/pipelines/audioldm/pipeline_audioldm.py

@@ -94,7 +94,9 @@ class AudioLDMPipeline(DiffusionPipeline):
             scheduler=scheduler,
             vocoder=vocoder,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
 
     # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.enable_vae_slicing
     def enable_vae_slicing(self):

src/diffusers/pipelines/audioldm2/pipeline_audioldm2.py

@@ -171,7 +171,9 @@ class AudioLDM2Pipeline(DiffusionPipeline):
             scheduler=scheduler,
             vocoder=vocoder,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
 
     # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.enable_vae_slicing
     def enable_vae_slicing(self):

src/diffusers/pipelines/controlnet/pipeline_controlnet.py

@@ -174,7 +174,9 @@ class StableDiffusionControlNetPipeline(
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor, do_convert_rgb=True)
         self.control_image_processor = VaeImageProcessor(
             vae_scale_factor=self.vae_scale_factor, do_convert_rgb=True, do_normalize=False

src/diffusers/pipelines/controlnet/pipeline_controlnet_img2img.py

@@ -208,7 +208,9 @@ class StableDiffusionControlNetImg2ImgPipeline(
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor, do_convert_rgb=True)
         self.control_image_processor = VaeImageProcessor(
             vae_scale_factor=self.vae_scale_factor, do_convert_rgb=True, do_normalize=False

src/diffusers/pipelines/controlnet/pipeline_controlnet_inpaint.py

@@ -330,7 +330,9 @@ class StableDiffusionControlNetInpaintPipeline(
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.mask_processor = VaeImageProcessor(
             vae_scale_factor=self.vae_scale_factor, do_normalize=False, do_binarize=True, do_convert_grayscale=True

src/diffusers/pipelines/controlnet/pipeline_controlnet_inpaint_sd_xl.py

@@ -199,7 +199,9 @@ class StableDiffusionXLControlNetInpaintPipeline(
         )
         self.register_to_config(force_zeros_for_empty_prompt=force_zeros_for_empty_prompt)
         self.register_to_config(requires_aesthetics_score=requires_aesthetics_score)
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.mask_processor = VaeImageProcessor(
             vae_scale_factor=self.vae_scale_factor, do_normalize=False, do_binarize=True, do_convert_grayscale=True

src/diffusers/pipelines/controlnet/pipeline_controlnet_sd_xl.py

@@ -172,7 +172,9 @@ class StableDiffusionXLControlNetPipeline(
             controlnet=controlnet,
             scheduler=scheduler,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        out_channels = getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)
+        self.vae_scale_factor = 2 ** (len(out_channels) - 1)
+
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor, do_convert_rgb=True)
         self.control_image_processor = VaeImageProcessor(
             vae_scale_factor=self.vae_scale_factor, do_convert_rgb=True, do_normalize=False

src/diffusers/pipelines/controlnet/pipeline_controlnet_sd_xl_img2img.py

@@ -225,7 +225,9 @@ class StableDiffusionXLControlNetImg2ImgPipeline(
             controlnet=controlnet,
             scheduler=scheduler,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor, do_convert_rgb=True)
         self.control_image_processor = VaeImageProcessor(
             vae_scale_factor=self.vae_scale_factor, do_convert_rgb=True, do_normalize=False

src/diffusers/pipelines/controlnet/pipeline_flax_controlnet.py

@@ -175,7 +175,9 @@ class FlaxStableDiffusionControlNetPipeline(FlaxDiffusionPipeline):
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
 
     def prepare_text_inputs(self, prompt: Union[str, List[str]]):
         if not isinstance(prompt, (str, list)):

src/diffusers/pipelines/latent_consistency_models/pipeline_latent_consistency_img2img.py

@@ -154,7 +154,9 @@ class LatentConsistencyModelImg2ImgPipeline(
                 " information, please have a look at https://github.com/huggingface/diffusers/pull/254 ."
             )
 
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
 
     # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.enable_vae_slicing

src/diffusers/pipelines/latent_consistency_models/pipeline_latent_consistency_text2img.py

@@ -141,7 +141,9 @@ class LatentConsistencyModelPipeline(
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 

src/diffusers/pipelines/musicldm/pipeline_musicldm.py

@@ -111,7 +111,9 @@ class MusicLDMPipeline(DiffusionPipeline):
             scheduler=scheduler,
             vocoder=vocoder,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
 
     # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.enable_vae_slicing
     def enable_vae_slicing(self):

src/diffusers/pipelines/paint_by_example/pipeline_paint_by_example.py

@@ -205,7 +205,9 @@ class PaintByExamplePipeline(DiffusionPipeline):
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 

src/diffusers/pipelines/pixart_alpha/pipeline_pixart_alpha.py

@@ -154,7 +154,9 @@ class PixArtAlphaPipeline(DiffusionPipeline):
             tokenizer=tokenizer, text_encoder=text_encoder, vae=vae, transformer=transformer, scheduler=scheduler
         )
 
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
 
     # Adapted from https://github.com/PixArt-alpha/PixArt-alpha/blob/master/diffusion/model/utils.py

src/diffusers/pipelines/semantic_stable_diffusion/pipeline_semantic_stable_diffusion.py

@@ -87,7 +87,9 @@ class SemanticStableDiffusionPipeline(DiffusionPipeline):
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 

src/diffusers/pipelines/stable_diffusion/pipeline_cycle_diffusion.py

@@ -224,7 +224,9 @@ class CycleDiffusionPipeline(DiffusionPipeline, TextualInversionLoaderMixin, Lor
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py

@@ -194,7 +194,9 @@ class StableDiffusionPipeline(DiffusionPipeline, TextualInversionLoaderMixin, Lo
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_attend_and_excite.py

@@ -239,7 +239,9 @@ class StableDiffusionAttendAndExcitePipeline(DiffusionPipeline, TextualInversion
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_depth2img.py

@@ -141,7 +141,9 @@ class StableDiffusionDepth2ImgPipeline(DiffusionPipeline, TextualInversionLoader
             depth_estimator=depth_estimator,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
 
     # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline._encode_prompt

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_diffedit.py

@@ -367,7 +367,9 @@ class StableDiffusionDiffEditPipeline(DiffusionPipeline, TextualInversionLoaderM
             feature_extractor=feature_extractor,
             inverse_scheduler=inverse_scheduler,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_gligen.py

@@ -168,7 +168,9 @@ class StableDiffusionGLIGENPipeline(DiffusionPipeline):
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor, do_convert_rgb=True)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_gligen_text_image.py

@@ -226,7 +226,9 @@ class StableDiffusionGLIGENTextImagePipeline(DiffusionPipeline):
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor, do_convert_rgb=True)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_image_variation.py

@@ -126,7 +126,9 @@ class StableDiffusionImageVariationPipeline(DiffusionPipeline):
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_img2img.py

@@ -231,7 +231,9 @@ class StableDiffusionImg2ImgPipeline(
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_inpaint.py

@@ -299,7 +299,9 @@ class StableDiffusionInpaintPipeline(
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.mask_processor = VaeImageProcessor(
             vae_scale_factor=self.vae_scale_factor, do_normalize=False, do_binarize=True, do_convert_grayscale=True

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_inpaint_legacy.py

@@ -213,7 +213,9 @@ class StableDiffusionInpaintPipelineLegacy(
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_instruct_pix2pix.py

@@ -133,7 +133,9 @@ class StableDiffusionInstructPix2PixPipeline(DiffusionPipeline, TextualInversion
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_k_diffusion.py

@@ -117,7 +117,9 @@ class StableDiffusionKDiffusionPipeline(DiffusionPipeline, TextualInversionLoade
             feature_extractor=feature_extractor,
         )
         self.register_to_config(requires_safety_checker=requires_safety_checker)
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
 
         model = ModelWrapper(unet, scheduler.alphas_cumprod)

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_latent_upscale.py

@@ -99,7 +99,9 @@ class StableDiffusionLatentUpscalePipeline(DiffusionPipeline, FromSingleFileMixi
             unet=unet,
             scheduler=scheduler,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor, resample="bicubic")
 
     def _encode_prompt(self, prompt, device, do_classifier_free_guidance, negative_prompt):

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_ldm3d.py

@@ -158,7 +158,9 @@ class StableDiffusionLDM3DPipeline(
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessorLDM3D(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_model_editing.py

@@ -114,7 +114,9 @@ class StableDiffusionModelEditingPipeline(DiffusionPipeline, TextualInversionLoa
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_panorama.py

@@ -128,7 +128,9 @@ class StableDiffusionPanoramaPipeline(DiffusionPipeline, TextualInversionLoaderM
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_paradigms.py

@@ -139,7 +139,9 @@ class StableDiffusionParadigmsPipeline(
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_pix2pix_zero.py

@@ -365,7 +365,9 @@ class StableDiffusionPix2PixZeroPipeline(DiffusionPipeline):
             caption_generator=caption_generator,
             inverse_scheduler=inverse_scheduler,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_sag.py

@@ -151,7 +151,9 @@ class StableDiffusionSAGPipeline(DiffusionPipeline, TextualInversionLoaderMixin)
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_upscale.py

@@ -141,7 +141,9 @@ class StableDiffusionUpscalePipeline(
             watermarker=watermarker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor, resample="bicubic")
         self.register_to_config(max_noise_level=max_noise_level)
 

src/diffusers/pipelines/stable_diffusion/pipeline_stable_unclip.py

@@ -147,7 +147,9 @@ class StableUnCLIPPipeline(DiffusionPipeline, TextualInversionLoaderMixin, LoraL
             vae=vae,
         )
 
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
 
     # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.enable_vae_slicing

src/diffusers/pipelines/stable_diffusion/pipeline_stable_unclip_img2img.py

@@ -148,7 +148,9 @@ class StableUnCLIPImg2ImgPipeline(DiffusionPipeline, TextualInversionLoaderMixin
             vae=vae,
         )
 
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
 
     # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.enable_vae_slicing

src/diffusers/pipelines/stable_diffusion_safe/pipeline_stable_diffusion_safe.py

@@ -142,7 +142,9 @@ class StableDiffusionPipelineSafe(DiffusionPipeline):
             feature_extractor=feature_extractor,
         )
         self._safety_text_concept = safety_concept
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 
     @property

src/diffusers/pipelines/stable_diffusion_xl/pipeline_flax_stable_diffusion_xl.py

@@ -65,7 +65,9 @@ class FlaxStableDiffusionXLPipeline(FlaxDiffusionPipeline):
             unet=unet,
             scheduler=scheduler,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
 
     def prepare_inputs(self, prompt: Union[str, List[str]]):
         if not isinstance(prompt, (str, list)):

src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl.py

@@ -177,7 +177,9 @@ class StableDiffusionXLPipeline(
             scheduler=scheduler,
         )
         self.register_to_config(force_zeros_for_empty_prompt=force_zeros_for_empty_prompt)
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
 
         self.default_sample_size = self.unet.config.sample_size

src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_img2img.py

@@ -192,7 +192,9 @@ class StableDiffusionXLImg2ImgPipeline(
         )
         self.register_to_config(force_zeros_for_empty_prompt=force_zeros_for_empty_prompt)
         self.register_to_config(requires_aesthetics_score=requires_aesthetics_score)
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
 
         add_watermarker = add_watermarker if add_watermarker is not None else is_invisible_watermark_available()

src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_inpaint.py

@@ -340,7 +340,9 @@ class StableDiffusionXLInpaintPipeline(
         )
         self.register_to_config(force_zeros_for_empty_prompt=force_zeros_for_empty_prompt)
         self.register_to_config(requires_aesthetics_score=requires_aesthetics_score)
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.mask_processor = VaeImageProcessor(
             vae_scale_factor=self.vae_scale_factor, do_normalize=False, do_binarize=True, do_convert_grayscale=True

src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_instruct_pix2pix.py

@@ -179,7 +179,9 @@ class StableDiffusionXLInstructPix2PixPipeline(
             scheduler=scheduler,
         )
         self.register_to_config(force_zeros_for_empty_prompt=force_zeros_for_empty_prompt)
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.default_sample_size = self.unet.config.sample_size
 

src/diffusers/pipelines/t2i_adapter/pipeline_stable_diffusion_adapter.py

@@ -199,7 +199,9 @@ class StableDiffusionAdapterPipeline(DiffusionPipeline):
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 

src/diffusers/pipelines/t2i_adapter/pipeline_stable_diffusion_xl_adapter.py

@@ -188,7 +188,9 @@ class StableDiffusionXLAdapterPipeline(
             scheduler=scheduler,
         )
         self.register_to_config(force_zeros_for_empty_prompt=force_zeros_for_empty_prompt)
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.default_sample_size = self.unet.config.sample_size
 

src/diffusers/pipelines/text_to_video_synthesis/pipeline_text_to_video_synth.py

@@ -116,7 +116,9 @@ class TextToVideoSDPipeline(DiffusionPipeline, TextualInversionLoaderMixin, Lora
             unet=unet,
             scheduler=scheduler,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
 
     # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.enable_vae_slicing
     def enable_vae_slicing(self):

src/diffusers/pipelines/text_to_video_synthesis/pipeline_text_to_video_synth_img2img.py

@@ -178,7 +178,9 @@ class VideoToVideoSDPipeline(DiffusionPipeline, TextualInversionLoaderMixin, Lor
             unet=unet,
             scheduler=scheduler,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
 
     # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.enable_vae_slicing
     def enable_vae_slicing(self):

src/diffusers/pipelines/unidiffuser/pipeline_unidiffuser.py

@@ -117,7 +117,9 @@ class UniDiffuserPipeline(DiffusionPipeline):
             scheduler=scheduler,
         )
 
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
 
         self.num_channels_latents = vae.config.latent_channels

src/diffusers/pipelines/versatile_diffusion/pipeline_versatile_diffusion.py

@@ -76,7 +76,9 @@ class VersatileDiffusionPipeline(DiffusionPipeline):
             vae=vae,
             scheduler=scheduler,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
 
     @torch.no_grad()
     def image_variation(

src/diffusers/pipelines/versatile_diffusion/pipeline_versatile_diffusion_dual_guided.py

@@ -94,7 +94,9 @@ class VersatileDiffusionDualGuidedPipeline(DiffusionPipeline):
             vae=vae,
             scheduler=scheduler,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
 
         if self.text_unet is not None and (

src/diffusers/pipelines/versatile_diffusion/pipeline_versatile_diffusion_image_variation.py

@@ -77,7 +77,9 @@ class VersatileDiffusionImageVariationPipeline(DiffusionPipeline):
             vae=vae,
             scheduler=scheduler,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
 
     def _encode_prompt(self, prompt, device, num_images_per_prompt, do_classifier_free_guidance, negative_prompt):

src/diffusers/pipelines/versatile_diffusion/pipeline_versatile_diffusion_text_to_image.py

@@ -82,7 +82,9 @@ class VersatileDiffusionTextToImagePipeline(DiffusionPipeline):
             vae=vae,
             scheduler=scheduler,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (
+            len(getattr(self.vae.config, "block_out_channels", self.vae.config.decoder_block_out_channels)) - 1
+        )
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
 
         if self.text_unet is not None: