diffusers 0.30.3__py3-none-any.whl → 0.32.0__py3-none-any.whl

diffusers/pipelines/pag/pipeline_pag_sd.py

@@ -60,9 +60,21 @@ EXAMPLE_DOC_STRING = """
 
 # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.rescale_noise_cfg
 def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0):
-    """
-    Rescale `noise_cfg` according to `guidance_rescale`. Based on findings of [Common Diffusion Noise Schedules and
-    Sample Steps are Flawed](https://arxiv.org/pdf/2305.08891.pdf). See Section 3.4
+    r"""
+    Rescales `noise_cfg` tensor based on `guidance_rescale` to improve image quality and fix overexposure. Based on
+    Section 3.4 from [Common Diffusion Noise Schedules and Sample Steps are
+    Flawed](https://arxiv.org/pdf/2305.08891.pdf).
+
+    Args:
+        noise_cfg (`torch.Tensor`):
+            The predicted noise tensor for the guided diffusion process.
+        noise_pred_text (`torch.Tensor`):
+            The predicted noise tensor for the text-guided diffusion process.
+        guidance_rescale (`float`, *optional*, defaults to 0.0):
+            A rescale factor applied to the noise predictions.
+
+    Returns:
+        noise_cfg (`torch.Tensor`): The rescaled noise prediction tensor.
     """
     std_text = noise_pred_text.std(dim=list(range(1, noise_pred_text.ndim)), keepdim=True)
     std_cfg = noise_cfg.std(dim=list(range(1, noise_cfg.ndim)), keepdim=True)
@@ -82,7 +94,7 @@ def retrieve_timesteps(
     sigmas: Optional[List[float]] = None,
     **kwargs,
 ):
-    """
+    r"""
     Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles
     custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`.
 
@@ -993,8 +1005,8 @@ class StableDiffusionPAGPipeline(
 
                 # perform guidance
                 if self.do_perturbed_attention_guidance:
-                    noise_pred = self._apply_perturbed_attention_guidance(
-                        noise_pred, self.do_classifier_free_guidance, self.guidance_scale, t
+                    noise_pred, noise_pred_text = self._apply_perturbed_attention_guidance(
+                        noise_pred, self.do_classifier_free_guidance, self.guidance_scale, t, True
                     )
 
                 elif self.do_classifier_free_guidance:

diffusers/pipelines/pag/pipeline_pag_sd_3.py

@@ -82,7 +82,7 @@ def retrieve_timesteps(
     sigmas: Optional[List[float]] = None,
     **kwargs,
 ):
-    """
+    r"""
     Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles
     custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`.
 
@@ -212,6 +212,9 @@ class StableDiffusion3PAGPipeline(DiffusionPipeline, SD3LoraLoaderMixin, FromSin
             if hasattr(self, "transformer") and self.transformer is not None
             else 128
         )
+        self.patch_size = (
+            self.transformer.config.patch_size if hasattr(self, "transformer") and self.transformer is not None else 2
+        )
 
         self.set_pag_applied_layers(
             pag_applied_layers, pag_attn_processors=(PAGCFGJointAttnProcessor2_0(), PAGJointAttnProcessor2_0())
@@ -542,8 +545,14 @@ class StableDiffusion3PAGPipeline(DiffusionPipeline, SD3LoraLoaderMixin, FromSin
         callback_on_step_end_tensor_inputs=None,
         max_sequence_length=None,
     ):
-        if height % 8 != 0 or width % 8 != 0:
-            raise ValueError(f"`height` and `width` have to be divisible by 8 but are {height} and {width}.")
+        if (
+            height % (self.vae_scale_factor * self.patch_size) != 0
+            or width % (self.vae_scale_factor * self.patch_size) != 0
+        ):
+            raise ValueError(
+                f"`height` and `width` have to be divisible by {self.vae_scale_factor * self.patch_size} but are {height} and {width}."
+                f"You can use height {height - height % (self.vae_scale_factor * self.patch_size)} and width {width - width % (self.vae_scale_factor * self.patch_size)}."
+            )
 
         if callback_on_step_end_tensor_inputs is not None and not all(
             k in self._callback_tensor_inputs for k in callback_on_step_end_tensor_inputs
@@ -684,7 +693,7 @@ class StableDiffusion3PAGPipeline(DiffusionPipeline, SD3LoraLoaderMixin, FromSin
         height: Optional[int] = None,
         width: Optional[int] = None,
         num_inference_steps: int = 28,
-        timesteps: List[int] = None,
+        sigmas: Optional[List[float]] = None,
         guidance_scale: float = 7.0,
         negative_prompt: Optional[Union[str, List[str]]] = None,
         negative_prompt_2: Optional[Union[str, List[str]]] = None,
@@ -726,10 +735,10 @@ class StableDiffusion3PAGPipeline(DiffusionPipeline, SD3LoraLoaderMixin, FromSin
             num_inference_steps (`int`, *optional*, defaults to 50):
                 The number of denoising steps. More denoising steps usually lead to a higher quality image at the
                 expense of slower inference.
-            timesteps (`List[int]`, *optional*):
-                Custom timesteps to use for the denoising process with schedulers which support a `timesteps` argument
-                in their `set_timesteps` method. If not defined, the default behavior when `num_inference_steps` is
-                passed will be used. Must be in descending order.
+            sigmas (`List[float]`, *optional*):
+                Custom sigmas to use for the denoising process with schedulers which support a `sigmas` argument in
+                their `set_timesteps` method. If not defined, the default behavior when `num_inference_steps` is passed
+                will be used.
             guidance_scale (`float`, *optional*, defaults to 7.0):
                 Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
                 `guidance_scale` is defined as `w` of equation 2. of [Imagen
@@ -881,7 +890,7 @@ class StableDiffusion3PAGPipeline(DiffusionPipeline, SD3LoraLoaderMixin, FromSin
             pooled_prompt_embeds = torch.cat([negative_pooled_prompt_embeds, pooled_prompt_embeds], dim=0)
 
         # 4. Prepare timesteps
-        timesteps, num_inference_steps = retrieve_timesteps(self.scheduler, num_inference_steps, device, timesteps)
+        timesteps, num_inference_steps = retrieve_timesteps(self.scheduler, num_inference_steps, device, sigmas=sigmas)
         num_warmup_steps = max(len(timesteps) - num_inference_steps * self.scheduler.order, 0)
         self._num_timesteps = len(timesteps)