Use pipelines without vae (#10441)

* Use pipelines without vae

* getattr

* vqvae

---------
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>

src/diffusers/pipelines/flux/pipeline_flux_img2img.py

@@ -211,9 +211,7 @@ class FluxImg2ImgPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFile
             transformer=transformer,
             scheduler=scheduler,
         )
-        self.vae_scale_factor = (
-            2 ** (len(self.vae.config.block_out_channels) - 1) if hasattr(self, "vae") and self.vae is not None else 8
-        )
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         # Flux latents are turned into 2x2 patches and packed. This means the latent width and height has to be divisible
         # by the patch size. So the vae scale factor is multiplied by the patch size to account for this
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor * 2)

src/diffusers/pipelines/flux/pipeline_flux_inpaint.py

@@ -208,15 +208,14 @@ class FluxInpaintPipeline(DiffusionPipeline, FluxLoraLoaderMixin):
             transformer=transformer,
             scheduler=scheduler,
         )
-        self.vae_scale_factor = (
-            2 ** (len(self.vae.config.block_out_channels) - 1) if hasattr(self, "vae") and self.vae is not None else 8
-        )
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         # Flux latents are turned into 2x2 patches and packed. This means the latent width and height has to be divisible
         # by the patch size. So the vae scale factor is multiplied by the patch size to account for this
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor * 2)
+        latent_channels = self.vae.config.latent_channels if getattr(self, "vae", None) else 16
         self.mask_processor = VaeImageProcessor(
             vae_scale_factor=self.vae_scale_factor * 2,
-            vae_latent_channels=self.vae.config.latent_channels,
+            vae_latent_channels=latent_channels,
             do_normalize=False,
             do_binarize=True,
             do_convert_grayscale=True,

src/diffusers/pipelines/hunyuan_video/pipeline_hunyuan_video.py

@@ -184,12 +184,8 @@ class HunyuanVideoPipeline(DiffusionPipeline, HunyuanVideoLoraLoaderMixin):
             tokenizer_2=tokenizer_2,
         )
 
-        self.vae_scale_factor_temporal = (
-            self.vae.temporal_compression_ratio if hasattr(self, "vae") and self.vae is not None else 4
-        )
-        self.vae_scale_factor_spatial = (
-            self.vae.spatial_compression_ratio if hasattr(self, "vae") and self.vae is not None else 8
-        )
+        self.vae_scale_factor_temporal = self.vae.temporal_compression_ratio if getattr(self, "vae", None) else 4
+        self.vae_scale_factor_spatial = self.vae.spatial_compression_ratio if getattr(self, "vae", None) else 8
         self.video_processor = VideoProcessor(vae_scale_factor=self.vae_scale_factor_spatial)
 
     def _get_llama_prompt_embeds(

src/diffusers/pipelines/hunyuandit/pipeline_hunyuandit.py

@@ -240,9 +240,7 @@ class HunyuanDiTPipeline(DiffusionPipeline):
                 " checker. If you do not want to use the safety checker, you can pass `'safety_checker=None'` instead."
             )
 
-        self.vae_scale_factor = (
-            2 ** (len(self.vae.config.block_out_channels) - 1) if hasattr(self, "vae") and self.vae is not None else 8
-        )
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
         self.default_sample_size = (

src/diffusers/pipelines/i2vgen_xl/pipeline_i2vgen_xl.py

@@ -133,7 +133,7 @@ class I2VGenXLPipeline(
             unet=unet,
             scheduler=scheduler,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         # `do_resize=False` as we do custom resizing.
         self.video_processor = VideoProcessor(vae_scale_factor=self.vae_scale_factor, do_resize=False)
 

src/diffusers/pipelines/kolors/pipeline_kolors.py

@@ -188,9 +188,7 @@ class KolorsPipeline(DiffusionPipeline, StableDiffusionMixin, StableDiffusionXLL
             feature_extractor=feature_extractor,
         )
         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) if hasattr(self, "vae") and self.vae is not None else 8
-        )
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
 
         self.default_sample_size = self.unet.config.sample_size

src/diffusers/pipelines/kolors/pipeline_kolors_img2img.py

@@ -207,9 +207,7 @@ class KolorsImg2ImgPipeline(DiffusionPipeline, StableDiffusionMixin, StableDiffu
             feature_extractor=feature_extractor,
         )
         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) if hasattr(self, "vae") and self.vae is not None else 8
-        )
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
 
         self.default_sample_size = self.unet.config.sample_size

src/diffusers/pipelines/latent_consistency_models/pipeline_latent_consistency_img2img.py

@@ -226,7 +226,7 @@ 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(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         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/latent_consistency_models/pipeline_latent_consistency_text2img.py

@@ -209,7 +209,7 @@ class LatentConsistencyModelPipeline(
             feature_extractor=feature_extractor,
             image_encoder=image_encoder,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 

src/diffusers/pipelines/latte/pipeline_latte.py

@@ -180,7 +180,7 @@ class LattePipeline(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(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         self.video_processor = VideoProcessor(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/ledits_pp/pipeline_leditspp_stable_diffusion.py

@@ -389,7 +389,7 @@ class LEditsPPPipelineStableDiffusion(
             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(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
 

src/diffusers/pipelines/ledits_pp/pipeline_leditspp_stable_diffusion_xl.py

@@ -372,7 +372,7 @@ class LEditsPPPipelineStableDiffusionXL(
             feature_extractor=feature_extractor,
         )
         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(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
 
         if not isinstance(scheduler, DDIMScheduler) and not isinstance(scheduler, DPMSolverMultistepScheduler):

src/diffusers/pipelines/marigold/pipeline_marigold_depth.py

@@ -174,7 +174,7 @@ class MarigoldDepthPipeline(DiffusionPipeline):
             default_processing_resolution=default_processing_resolution,
         )
 
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
 
         self.scale_invariant = scale_invariant
         self.shift_invariant = shift_invariant

src/diffusers/pipelines/marigold/pipeline_marigold_normals.py

@@ -161,7 +161,7 @@ class MarigoldNormalsPipeline(DiffusionPipeline):
             default_processing_resolution=default_processing_resolution,
         )
 
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
 
         self.use_full_z_range = use_full_z_range
         self.default_denoising_steps = default_denoising_steps

src/diffusers/pipelines/musicldm/pipeline_musicldm.py

@@ -111,7 +111,7 @@ class MusicLDMPipeline(DiffusionPipeline, StableDiffusionMixin):
             scheduler=scheduler,
             vocoder=vocoder,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
 
     def _encode_prompt(
         self,

src/diffusers/pipelines/pag/pipeline_pag_controlnet_sd.py

@@ -251,7 +251,7 @@ class StableDiffusionControlNetPAGPipeline(
             feature_extractor=feature_extractor,
             image_encoder=image_encoder,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         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/pag/pipeline_pag_controlnet_sd_inpaint.py

@@ -228,7 +228,7 @@ class StableDiffusionControlNetPAGInpaintPipeline(
             feature_extractor=feature_extractor,
             image_encoder=image_encoder,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         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/pag/pipeline_pag_controlnet_sd_xl.py

@@ -280,7 +280,7 @@ class StableDiffusionXLControlNetPAGPipeline(
             feature_extractor=feature_extractor,
             image_encoder=image_encoder,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         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/pag/pipeline_pag_controlnet_sd_xl_img2img.py

@@ -270,7 +270,7 @@ class StableDiffusionXLControlNetPAGImg2ImgPipeline(
             feature_extractor=feature_extractor,
             image_encoder=image_encoder,
         )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         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/pag/pipeline_pag_hunyuandit.py

@@ -245,9 +245,7 @@ class HunyuanDiTPAGPipeline(DiffusionPipeline, PAGMixin):
                 " checker. If you do not want to use the safety checker, you can pass `'safety_checker=None'` instead."
             )
 
-        self.vae_scale_factor = (
-            2 ** (len(self.vae.config.block_out_channels) - 1) if hasattr(self, "vae") and self.vae is not None else 8
-        )
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
         self.register_to_config(requires_safety_checker=requires_safety_checker)
         self.default_sample_size = (