[scheduler] fix a bug in add_noise (huggingface#7386)

yiyixuxu · sayakpaul · yiyixuxu · sayakpaul · commit 26b694bc6a8d · 2024-03-20T07:06:21.000+05:30
* fix

* fix

* add a tests

* fix

---------

Co-authored-by: Sayak Paul &lt;spsayakpaul@gmail.com&gt;
Co-authored-by: yiyixuxu &lt;yixu310@gmail,com&gt;
diff --git a/src/diffusers/pipelines/deprecated/stable_diffusion_variants/pipeline_stable_diffusion_inpaint_legacy.py b/src/diffusers/pipelines/deprecated/stable_diffusion_variants/pipeline_stable_diffusion_inpaint_legacy.py
@@ -528,15 +528,12 @@ def check_inputs(
                     f" {negative_prompt_embeds.shape}."
                 )
 
-    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.StableDiffusionImg2ImgPipeline.get_timesteps
     def get_timesteps(self, num_inference_steps, strength, device):
         # get the original timestep using init_timestep
         init_timestep = min(int(num_inference_steps * strength), num_inference_steps)
 
         t_start = max(num_inference_steps - init_timestep, 0)
         timesteps = self.scheduler.timesteps[t_start * self.scheduler.order :]
-        if hasattr(self.scheduler, "set_begin_index"):
-            self.scheduler.set_begin_index(t_start * self.scheduler.order)
 
         return timesteps, num_inference_steps - t_start
 
diff --git a/src/diffusers/pipelines/stable_diffusion_diffedit/pipeline_stable_diffusion_diffedit.py b/src/diffusers/pipelines/stable_diffusion_diffedit/pipeline_stable_diffusion_diffedit.py
@@ -716,15 +716,12 @@ def check_source_inputs(
                     f" `source_negative_prompt_embeds` {source_negative_prompt_embeds.shape}."
                 )
 
-    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.StableDiffusionImg2ImgPipeline.get_timesteps
     def get_timesteps(self, num_inference_steps, strength, device):
         # get the original timestep using init_timestep
         init_timestep = min(int(num_inference_steps * strength), num_inference_steps)
 
         t_start = max(num_inference_steps - init_timestep, 0)
         timesteps = self.scheduler.timesteps[t_start * self.scheduler.order :]
-        if hasattr(self.scheduler, "set_begin_index"):
-            self.scheduler.set_begin_index(t_start * self.scheduler.order)
 
         return timesteps, num_inference_steps - t_start
 
diff --git a/src/diffusers/schedulers/scheduling_consistency_models.py b/src/diffusers/schedulers/scheduling_consistency_models.py
@@ -434,7 +434,11 @@ def add_noise(
         # self.begin_index is None when scheduler is used for training, or pipeline does not implement set_begin_index
         if self.begin_index is None:
             step_indices = [self.index_for_timestep(t, schedule_timesteps) for t in timesteps]
+        elif self.step_index is not None:
+            # add_noise is called after first denoising step (for inpainting)
+            step_indices = [self.step_index] * timesteps.shape[0]
         else:
+            # add noise is called bevore first denoising step to create inital latent(img2img)
             step_indices = [self.begin_index] * timesteps.shape[0]
 
         sigma = sigmas[step_indices].flatten()
diff --git a/src/diffusers/schedulers/scheduling_deis_multistep.py b/src/diffusers/schedulers/scheduling_deis_multistep.py
@@ -768,10 +768,14 @@ def add_noise(
             schedule_timesteps = self.timesteps.to(original_samples.device)
             timesteps = timesteps.to(original_samples.device)
 
-        # begin_index is None when the scheduler is used for training
+        # begin_index is None when the scheduler is used for training or pipeline does not implement set_begin_index
         if self.begin_index is None:
             step_indices = [self.index_for_timestep(t, schedule_timesteps) for t in timesteps]
+        elif self.step_index is not None:
+            # add_noise is called after first denoising step (for inpainting)
+            step_indices = [self.step_index] * timesteps.shape[0]
         else:
+            # add noise is called bevore first denoising step to create inital latent(img2img)
             step_indices = [self.begin_index] * timesteps.shape[0]
 
         sigma = sigmas[step_indices].flatten()
diff --git a/src/diffusers/schedulers/scheduling_dpmsolver_multistep.py b/src/diffusers/schedulers/scheduling_dpmsolver_multistep.py
@@ -1011,10 +1011,14 @@ def add_noise(
             schedule_timesteps = self.timesteps.to(original_samples.device)
             timesteps = timesteps.to(original_samples.device)
 
-        # begin_index is None when the scheduler is used for training
+        # begin_index is None when the scheduler is used for training or pipeline does not implement set_begin_index
         if self.begin_index is None:
             step_indices = [self.index_for_timestep(t, schedule_timesteps) for t in timesteps]
+        elif self.step_index is not None:
+            # add_noise is called after first denoising step (for inpainting)
+            step_indices = [self.step_index] * timesteps.shape[0]
         else:
+            # add noise is called bevore first denoising step to create inital latent(img2img)
             step_indices = [self.begin_index] * timesteps.shape[0]
 
         sigma = sigmas[step_indices].flatten()
diff --git a/src/diffusers/schedulers/scheduling_dpmsolver_sde.py b/src/diffusers/schedulers/scheduling_dpmsolver_sde.py
@@ -543,7 +543,11 @@ def add_noise(
         # self.begin_index is None when scheduler is used for training, or pipeline does not implement set_begin_index
         if self.begin_index is None:
             step_indices = [self.index_for_timestep(t, schedule_timesteps) for t in timesteps]
+        elif self.step_index is not None:
+            # add_noise is called after first denoising step (for inpainting)
+            step_indices = [self.step_index] * timesteps.shape[0]
         else:
+            # add noise is called bevore first denoising step to create inital latent(img2img)
             step_indices = [self.begin_index] * timesteps.shape[0]
 
         sigma = sigmas[step_indices].flatten()
diff --git a/src/diffusers/schedulers/scheduling_dpmsolver_singlestep.py b/src/diffusers/schedulers/scheduling_dpmsolver_singlestep.py
@@ -961,10 +961,14 @@ def add_noise(
             schedule_timesteps = self.timesteps.to(original_samples.device)
             timesteps = timesteps.to(original_samples.device)
 
-        # begin_index is None when the scheduler is used for training
+        # begin_index is None when the scheduler is used for training or pipeline does not implement set_begin_index
         if self.begin_index is None:
             step_indices = [self.index_for_timestep(t, schedule_timesteps) for t in timesteps]
+        elif self.step_index is not None:
+            # add_noise is called after first denoising step (for inpainting)
+            step_indices = [self.step_index] * timesteps.shape[0]
         else:
+            # add noise is called bevore first denoising step to create inital latent(img2img)
             step_indices = [self.begin_index] * timesteps.shape[0]
 
         sigma = sigmas[step_indices].flatten()
diff --git a/src/diffusers/schedulers/scheduling_edm_dpmsolver_multistep.py b/src/diffusers/schedulers/scheduling_edm_dpmsolver_multistep.py
@@ -669,7 +669,11 @@ def add_noise(
         # self.begin_index is None when scheduler is used for training, or pipeline does not implement set_begin_index
         if self.begin_index is None:
             step_indices = [self.index_for_timestep(t, schedule_timesteps) for t in timesteps]
+        elif self.step_index is not None:
+            # add_noise is called after first denoising step (for inpainting)
+            step_indices = [self.step_index] * timesteps.shape[0]
         else:
+            # add noise is called bevore first denoising step to create inital latent(img2img)
             step_indices = [self.begin_index] * timesteps.shape[0]
 
         sigma = sigmas[step_indices].flatten()
diff --git a/src/diffusers/schedulers/scheduling_edm_euler.py b/src/diffusers/schedulers/scheduling_edm_euler.py
@@ -367,7 +367,11 @@ def add_noise(
         # self.begin_index is None when scheduler is used for training, or pipeline does not implement set_begin_index
         if self.begin_index is None:
             step_indices = [self.index_for_timestep(t, schedule_timesteps) for t in timesteps]
+        elif self.step_index is not None:
+            # add_noise is called after first denoising step (for inpainting)
+            step_indices = [self.step_index] * timesteps.shape[0]
         else:
+            # add noise is called bevore first denoising step to create inital latent(img2img)
             step_indices = [self.begin_index] * timesteps.shape[0]
 
         sigma = sigmas[step_indices].flatten()
diff --git a/src/diffusers/schedulers/scheduling_euler_ancestral_discrete.py b/src/diffusers/schedulers/scheduling_euler_ancestral_discrete.py
@@ -467,7 +467,11 @@ def add_noise(
         # self.begin_index is None when scheduler is used for training, or pipeline does not implement set_begin_index
         if self.begin_index is None:
             step_indices = [self.index_for_timestep(t, schedule_timesteps) for t in timesteps]
+        elif self.step_index is not None:
+            # add_noise is called after first denoising step (for inpainting)
+            step_indices = [self.step_index] * timesteps.shape[0]
         else:
+            # add noise is called bevore first denoising step to create inital latent(img2img)
             step_indices = [self.begin_index] * timesteps.shape[0]
 
         sigma = sigmas[step_indices].flatten()
diff --git a/src/diffusers/schedulers/scheduling_euler_discrete.py b/src/diffusers/schedulers/scheduling_euler_discrete.py
@@ -562,7 +562,11 @@ def add_noise(
         # self.begin_index is None when scheduler is used for training, or pipeline does not implement set_begin_index
         if self.begin_index is None:
             step_indices = [self.index_for_timestep(t, schedule_timesteps) for t in timesteps]
+        elif self.step_index is not None:
+            # add_noise is called after first denoising step (for inpainting)
+            step_indices = [self.step_index] * timesteps.shape[0]
         else:
+            # add noise is called bevore first denoising step to create inital latent(img2img)
             step_indices = [self.begin_index] * timesteps.shape[0]
 
         sigma = sigmas[step_indices].flatten()
diff --git a/src/diffusers/schedulers/scheduling_heun_discrete.py b/src/diffusers/schedulers/scheduling_heun_discrete.py
@@ -468,7 +468,11 @@ def add_noise(
         # self.begin_index is None when scheduler is used for training, or pipeline does not implement set_begin_index
         if self.begin_index is None:
             step_indices = [self.index_for_timestep(t, schedule_timesteps) for t in timesteps]
+        elif self.step_index is not None:
+            # add_noise is called after first denoising step (for inpainting)
+            step_indices = [self.step_index] * timesteps.shape[0]
         else:
+            # add noise is called bevore first denoising step to create inital latent(img2img)
             step_indices = [self.begin_index] * timesteps.shape[0]
 
         sigma = sigmas[step_indices].flatten()
diff --git a/src/diffusers/schedulers/scheduling_k_dpm_2_ancestral_discrete.py b/src/diffusers/schedulers/scheduling_k_dpm_2_ancestral_discrete.py
@@ -494,7 +494,11 @@ def add_noise(
         # self.begin_index is None when scheduler is used for training, or pipeline does not implement set_begin_index
         if self.begin_index is None:
             step_indices = [self.index_for_timestep(t, schedule_timesteps) for t in timesteps]
+        elif self.step_index is not None:
+            # add_noise is called after first denoising step (for inpainting)
+            step_indices = [self.step_index] * timesteps.shape[0]
         else:
+            # add noise is called bevore first denoising step to create inital latent(img2img)
             step_indices = [self.begin_index] * timesteps.shape[0]
 
         sigma = sigmas[step_indices].flatten()
diff --git a/src/diffusers/schedulers/scheduling_k_dpm_2_discrete.py b/src/diffusers/schedulers/scheduling_k_dpm_2_discrete.py
@@ -469,7 +469,11 @@ def add_noise(
         # self.begin_index is None when scheduler is used for training, or pipeline does not implement set_begin_index
         if self.begin_index is None:
             step_indices = [self.index_for_timestep(t, schedule_timesteps) for t in timesteps]
+        elif self.step_index is not None:
+            # add_noise is called after first denoising step (for inpainting)
+            step_indices = [self.step_index] * timesteps.shape[0]
         else:
+            # add noise is called bevore first denoising step to create inital latent(img2img)
             step_indices = [self.begin_index] * timesteps.shape[0]
 
         sigma = sigmas[step_indices].flatten()
diff --git a/src/diffusers/schedulers/scheduling_lms_discrete.py b/src/diffusers/schedulers/scheduling_lms_discrete.py
@@ -461,7 +461,11 @@ def add_noise(
         # self.begin_index is None when scheduler is used for training, or pipeline does not implement set_begin_index
         if self.begin_index is None:
             step_indices = [self.index_for_timestep(t, schedule_timesteps) for t in timesteps]
+        elif self.step_index is not None:
+            # add_noise is called after first denoising step (for inpainting)
+            step_indices = [self.step_index] * timesteps.shape[0]
         else:
+            # add noise is called bevore first denoising step to create inital latent(img2img)
             step_indices = [self.begin_index] * timesteps.shape[0]
 
         sigma = sigmas[step_indices].flatten()
diff --git a/src/diffusers/schedulers/scheduling_unipc_multistep.py b/src/diffusers/schedulers/scheduling_unipc_multistep.py
@@ -862,10 +862,14 @@ def add_noise(
             schedule_timesteps = self.timesteps.to(original_samples.device)
             timesteps = timesteps.to(original_samples.device)
 
-        # begin_index is None when the scheduler is used for training
+        # begin_index is None when the scheduler is used for training or pipeline does not implement set_begin_index
         if self.begin_index is None:
             step_indices = [self.index_for_timestep(t, schedule_timesteps) for t in timesteps]
+        elif self.step_index is not None:
+            # add_noise is called after first denoising step (for inpainting)
+            step_indices = [self.step_index] * timesteps.shape[0]
         else:
+            # add noise is called bevore first denoising step to create inital latent(img2img)
             step_indices = [self.begin_index] * timesteps.shape[0]
 
         sigma = sigmas[step_indices].flatten()
diff --git a/tests/pipelines/stable_diffusion/test_stable_diffusion_inpaint.py b/tests/pipelines/stable_diffusion/test_stable_diffusion_inpaint.py
@@ -29,6 +29,7 @@
     AutoencoderKL,
     DDIMScheduler,
     DPMSolverMultistepScheduler,
+    EulerAncestralDiscreteScheduler,
     LCMScheduler,
     LMSDiscreteScheduler,
     PNDMScheduler,
@@ -557,6 +558,29 @@ def test_stable_diffusion_inpaint_2_images(self):
         image_slice2 = images[1, -3:, -3:, -1]
         assert np.abs(image_slice1.flatten() - image_slice2.flatten()).max() > 1e-2
 
+    def test_stable_diffusion_inpaint_euler(self):
+        device = "cpu"  # ensure determinism for the device-dependent torch.Generator
+        components = self.get_dummy_components(time_cond_proj_dim=256)
+        sd_pipe = StableDiffusionInpaintPipeline(**components)
+        sd_pipe.scheduler = EulerAncestralDiscreteScheduler.from_config(sd_pipe.scheduler.config)
+        sd_pipe = sd_pipe.to(device)
+        sd_pipe.set_progress_bar_config(disable=None)
+
+        inputs = self.get_dummy_inputs(device, output_pil=False)
+        half_dim = inputs["image"].shape[2] // 2
+        inputs["mask_image"][0, 0, :half_dim, :half_dim] = 0
+
+        inputs["num_inference_steps"] = 4
+        image = sd_pipe(**inputs).images
+        image_slice = image[0, -3:, -3:, -1]
+
+        assert image.shape == (1, 64, 64, 3)
+
+        expected_slice = np.array(
+            [[0.6387283, 0.5564158, 0.58631873, 0.5539942, 0.5494673, 0.6461868, 0.5251618, 0.5497595, 0.5508756]]
+        )
+        assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-4
+
 
 @slow
 @require_torch_gpu
