sana pipeline

app.py

@@ -9,10 +9,6 @@ import torch
 
 device = "cuda" if torch.cuda.is_available() else "cpu"
 
-if torch.cuda.is_available():
-    torch_dtype = torch.float16
-else:
-    torch_dtype = torch.float32
 
 MODEL_OPTIONS = {
     "SiD-Flow-SD3-medium": "YGu1998/SiD-Flow-SD3-medium",
@@ -33,16 +29,19 @@ MODEL_OPTIONS = {
 
 def load_model(model_choice):
     model_repo_id = MODEL_OPTIONS[model_choice]
+    time_scale = 1000.0
     if "Sana" in model_choice:
-        pipe = SiDSanaPipeline.from_pretrained(model_repo_id, torch_dtype=torch_dtype)
+        pipe = SiDSanaPipeline.from_pretrained(model_repo_id, torch_dtype=torch.float16)
+        if "Sprint" in model_choice:
+            time_scale = 1.0
     elif "SD3" in model_choice:
-        pipe = SiDSD3Pipeline.from_pretrained(model_repo_id, torch_dtype=torch_dtype)
+        pipe = SiDSD3Pipeline.from_pretrained(model_repo_id, torch_dtype=torch.float16)
     elif "Flux" in model_choice:
-        pipe = SiDFluxPipeline.from_pretrained(model_repo_id, torch_dtype=torch_dtype)
+        pipe = SiDFluxPipeline.from_pretrained(model_repo_id, torch_dtype=torch.float16)
     else:
         raise ValueError(f"Unknown model type for: {model_choice}")
     pipe = pipe.to(device)
-    return pipe
+    return pipe, time_scale
 
 
 MAX_SEED = np.iinfo(np.int32).max
@@ -65,7 +64,7 @@ def infer(
 
     generator = torch.Generator().manual_seed(seed)
 
-    pipe = load_model(model_choice)
+    pipe, time_scale = load_model(model_choice)
 
     image = pipe(
         prompt=prompt,
@@ -74,6 +73,7 @@ def infer(
         width=width,
         height=height,
         generator=generator,
+        time_scale=time_scale,
     ).images[0]
 
     return image, seed

sid/pipeline_sid_sana.py

@@ -700,141 +700,27 @@ class SiDSanaPipeline(DiffusionPipeline, SanaLoraLoaderMixin):
         return self._interrupt
 
     @torch.no_grad()
-    @replace_example_docstring(EXAMPLE_DOC_STRING)
     def __call__(
         self,
         prompt: Union[str, List[str]] = None,
-        negative_prompt: str = "",
-        num_inference_steps: int = 20,
-        timesteps: List[int] = None,
-        sigmas: List[float] = None,
-        guidance_scale: float = 4.5,
+        height: Optional[int] = None,
+        width: Optional[int] = None,
+        num_inference_steps: int = 28,
+        guidance_scale: float = 1.0,
         num_images_per_prompt: Optional[int] = 1,
-        height: int = 1024,
-        width: int = 1024,
-        eta: float = 0.0,
         generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
-        latents: Optional[torch.Tensor] = None,
-        prompt_embeds: Optional[torch.Tensor] = None,
-        prompt_attention_mask: Optional[torch.Tensor] = None,
-        negative_prompt_embeds: Optional[torch.Tensor] = None,
-        negative_prompt_attention_mask: Optional[torch.Tensor] = None,
+        latents: Optional[torch.FloatTensor] = None,
+        prompt_embeds: Optional[torch.FloatTensor] = None,
+        pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
-        clean_caption: bool = False,
-        use_resolution_binning: bool = True,
-        attention_kwargs: Optional[Dict[str, Any]] = None,
         callback_on_step_end: Optional[Callable[[int, int, Dict], None]] = None,
         callback_on_step_end_tensor_inputs: List[str] = ["latents"],
-        max_sequence_length: int = 300,
-        complex_human_instruction: List[str] = [
-            "Given a user prompt, generate an 'Enhanced prompt' that provides detailed visual descriptions suitable for image generation. Evaluate the level of detail in the user prompt:",
-            "- If the prompt is simple, focus on adding specifics about colors, shapes, sizes, textures, and spatial relationships to create vivid and concrete scenes.",
-            "- If the prompt is already detailed, refine and enhance the existing details slightly without overcomplicating.",
-            "Here are examples of how to transform or refine prompts:",
-            "- User Prompt: A cat sleeping -> Enhanced: A small, fluffy white cat curled up in a round shape, sleeping peacefully on a warm sunny windowsill, surrounded by pots of blooming red flowers.",
-            "- User Prompt: A busy city street -> Enhanced: A bustling city street scene at dusk, featuring glowing street lamps, a diverse crowd of people in colorful clothing, and a double-decker bus passing by towering glass skyscrapers.",
-            "Please generate only the enhanced description for the prompt below and avoid including any additional commentary or evaluations:",
-            "User Prompt: ",
-        ],
-    ) -> Union[SiDPipelineOutput, Tuple]:
-        """
-        Function invoked when calling the pipeline for generation.
-
-        Args:
-            prompt (`str` or `List[str]`, *optional*):
-                The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`.
-                instead.
-            negative_prompt (`str` or `List[str]`, *optional*):
-                The prompt or prompts not to guide the image generation. If not defined, one has to pass
-                `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
-                less than `1`).
-            num_inference_steps (`int`, *optional*, defaults to 20):
-                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 4.5):
-                Guidance scale as defined in [Classifier-Free Diffusion
-                Guidance](https://huggingface.co/papers/2207.12598). `guidance_scale` is defined as `w` of equation 2.
-                of [Imagen Paper](https://huggingface.co/papers/2205.11487). Guidance scale is enabled by setting
-                `guidance_scale > 1`. Higher guidance scale encourages to generate images that are closely linked to
-                the text `prompt`, usually at the expense of lower image quality.
-            num_images_per_prompt (`int`, *optional*, defaults to 1):
-                The number of images to generate per prompt.
-            height (`int`, *optional*, defaults to self.unet.config.sample_size):
-                The height in pixels of the generated image.
-            width (`int`, *optional*, defaults to self.unet.config.sample_size):
-                The width in pixels of the generated image.
-            eta (`float`, *optional*, defaults to 0.0):
-                Corresponds to parameter eta (η) in the DDIM paper: https://huggingface.co/papers/2010.02502. Only
-                applies to [`schedulers.DDIMScheduler`], will be ignored for others.
-            generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
-                One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
-                to make generation deterministic.
-            latents (`torch.Tensor`, *optional*):
-                Pre-generated noisy latents, sampled from a Gaussian distribution, to be used as inputs for image
-                generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
-                tensor will ge generated by sampling using the supplied random `generator`.
-            prompt_embeds (`torch.Tensor`, *optional*):
-                Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
-                provided, text embeddings will be generated from `prompt` input argument.
-            prompt_attention_mask (`torch.Tensor`, *optional*): Pre-generated attention mask for text embeddings.
-            negative_prompt_embeds (`torch.Tensor`, *optional*):
-                Pre-generated negative text embeddings. For PixArt-Sigma this negative prompt should be "". If not
-                provided, negative_prompt_embeds will be generated from `negative_prompt` input argument.
-            negative_prompt_attention_mask (`torch.Tensor`, *optional*):
-                Pre-generated attention mask for negative text embeddings.
-            output_type (`str`, *optional*, defaults to `"pil"`):
-                The output format of the generate image. Choose between
-                [PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`.
-            return_dict (`bool`, *optional*, defaults to `True`):
-                Whether or not to return a [`~pipelines.stable_diffusion.IFPipelineOutput`] instead of a plain tuple.
-            attention_kwargs:
-                A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
-                `self.processor` in
-                [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
-            clean_caption (`bool`, *optional*, defaults to `True`):
-                Whether or not to clean the caption before creating embeddings. Requires `beautifulsoup4` and `ftfy` to
-                be installed. If the dependencies are not installed, the embeddings will be created from the raw
-                prompt.
-            use_resolution_binning (`bool` defaults to `True`):
-                If set to `True`, the requested height and width are first mapped to the closest resolutions using
-                `ASPECT_RATIO_1024_BIN`. After the produced latents are decoded into images, they are resized back to
-                the requested resolution. Useful for generating non-square images.
-            callback_on_step_end (`Callable`, *optional*):
-                A function that calls at the end of each denoising steps during the inference. The function is called
-                with the following arguments: `callback_on_step_end(self: DiffusionPipeline, step: int, timestep: int,
-                callback_kwargs: Dict)`. `callback_kwargs` will include a list of all tensors as specified by
-                `callback_on_step_end_tensor_inputs`.
-            callback_on_step_end_tensor_inputs (`List`, *optional*):
-                The list of tensor inputs for the `callback_on_step_end` function. The tensors specified in the list
-                will be passed as `callback_kwargs` argument. You will only be able to include variables listed in the
-                `._callback_tensor_inputs` attribute of your pipeline class.
-            max_sequence_length (`int` defaults to `300`):
-                Maximum sequence length to use with the `prompt`.
-            complex_human_instruction (`List[str]`, *optional*):
-                Instructions for complex human attention:
-                https://github.com/NVlabs/Sana/blob/main/configs/sana_app_config/Sana_1600M_app.yaml#L55.
-
-        Examples:
-
-        Returns:
-            [`~pipelines.sana.pipeline_output.SanaPipelineOutput`] or `tuple`:
-                If `return_dict` is `True`, [`~pipelines.sana.pipeline_output.SanaPipelineOutput`] is returned,
-                otherwise a `tuple` is returned where the first element is a list with the generated images
-        """
-
-        if isinstance(callback_on_step_end, (PipelineCallback, MultiPipelineCallbacks)):
-            callback_on_step_end_tensor_inputs = callback_on_step_end.tensor_inputs
-
-        # 1. Check inputs. Raise error if not correct
+        max_sequence_length: int = 256,
+        noise_type: str = "fresh",  # 'fresh', 'ddim', 'fixed'
+        time_scale: float = 1000.0,
+        use_resolution_binning: bool = True,
+    ):
         if use_resolution_binning:
             if self.transformer.config.sample_size == 128:
                 aspect_ratio_bin = ASPECT_RATIO_4096_BIN
@@ -848,24 +734,24 @@ class SiDSanaPipeline(DiffusionPipeline, SanaLoraLoaderMixin):
                 raise ValueError("Invalid sample size")
             orig_height, orig_width = height, width
             height, width = self.image_processor.classify_height_width_bin(height, width, ratios=aspect_ratio_bin)
+        
+        if isinstance(callback_on_step_end, (PipelineCallback, MultiPipelineCallbacks)):
+            callback_on_step_end_tensor_inputs = callback_on_step_end.tensor_inputs
 
+        # 1. Check inputs. Raise error if not correct
         self.check_inputs(
             prompt,
             height,
             width,
-            callback_on_step_end_tensor_inputs,
-            negative_prompt,
-            prompt_embeds,
-            negative_prompt_embeds,
-            prompt_attention_mask,
-            negative_prompt_attention_mask,
+            prompt_embeds=prompt_embeds,
+            pooled_prompt_embeds=pooled_prompt_embeds,
+            callback_on_step_end_tensor_inputs=callback_on_step_end_tensor_inputs,
         )
 
         self._guidance_scale = guidance_scale
-        self._attention_kwargs = attention_kwargs
         self._interrupt = False
 
-        # 2. Default height and width to transformer
+        # 2. Define call parameters
         if prompt is not None and isinstance(prompt, str):
             batch_size = 1
         elif prompt is not None and isinstance(prompt, list):
@@ -874,134 +760,89 @@ class SiDSanaPipeline(DiffusionPipeline, SanaLoraLoaderMixin):
             batch_size = prompt_embeds.shape[0]
 
         device = self._execution_device
-        lora_scale = self.attention_kwargs.get("scale", None) if self.attention_kwargs is not None else None
 
-        # 3. Encode input prompt
         (
             prompt_embeds,
-            prompt_attention_mask,
-            negative_prompt_embeds,
-            negative_prompt_attention_mask,
+            pooled_prompt_embeds,
+            _, _,
         ) = self.encode_prompt(
             prompt,
-            self.do_classifier_free_guidance,
-            negative_prompt=negative_prompt,
-            num_images_per_prompt=num_images_per_prompt,
-            device=device,
             prompt_embeds=prompt_embeds,
-            negative_prompt_embeds=negative_prompt_embeds,
-            prompt_attention_mask=prompt_attention_mask,
-            negative_prompt_attention_mask=negative_prompt_attention_mask,
-            clean_caption=clean_caption,
+            pooled_prompt_embeds=pooled_prompt_embeds,
+            device=device,
+            num_images_per_prompt=num_images_per_prompt,
             max_sequence_length=max_sequence_length,
-            complex_human_instruction=complex_human_instruction,
-            lora_scale=lora_scale,
         )
-        if self.do_classifier_free_guidance:
-            prompt_embeds = torch.cat([negative_prompt_embeds, prompt_embeds], dim=0)
-            prompt_attention_mask = torch.cat([negative_prompt_attention_mask, prompt_attention_mask], dim=0)
-
-        # 4. Prepare timesteps
-        timesteps, num_inference_steps = retrieve_timesteps(
-            self.scheduler, num_inference_steps, device, timesteps, sigmas
-        )
-
-        # 5. Prepare latents.
-        latent_channels = self.transformer.config.in_channels
+        # 3. Prepare latents
+        num_channels_latents = self.transformer.config.in_channels
         latents = self.prepare_latents(
             batch_size * num_images_per_prompt,
-            latent_channels,
+            num_channels_latents,
             height,
             width,
-            torch.float32,
+            prompt_embeds.dtype,
             device,
             generator,
             latents,
         )
 
-        # 6. Prepare extra step kwargs. TODO: Logic should ideally just be moved out of the pipeline
-        extra_step_kwargs = self.prepare_extra_step_kwargs(generator, eta)
-
-        # 7. Denoising loop
-        num_warmup_steps = max(len(timesteps) - num_inference_steps * self.scheduler.order, 0)
-        self._num_timesteps = len(timesteps)
-
-        transformer_dtype = self.transformer.dtype
-        with self.progress_bar(total=num_inference_steps) as progress_bar:
-            for i, t in enumerate(timesteps):
-                if self.interrupt:
-                    continue
-
-                latent_model_input = torch.cat([latents] * 2) if self.do_classifier_free_guidance else latents
-
-                # broadcast to batch dimension in a way that's compatible with ONNX/Core ML
-                timestep = t.expand(latent_model_input.shape[0])
-                timestep = timestep * self.transformer.config.timestep_scale
-
-                # predict noise model_output
-                noise_pred = self.transformer(
-                    latent_model_input.to(dtype=transformer_dtype),
-                    encoder_hidden_states=prompt_embeds.to(dtype=transformer_dtype),
-                    encoder_attention_mask=prompt_attention_mask,
-                    timestep=timestep,
-                    return_dict=False,
-                    attention_kwargs=self.attention_kwargs,
-                )[0]
-                noise_pred = noise_pred.float()
-
-                # perform guidance
-                if self.do_classifier_free_guidance:
-                    noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
-                    noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)
-
-                # learned sigma
-                if self.transformer.config.out_channels // 2 == latent_channels:
-                    noise_pred = noise_pred.chunk(2, dim=1)[0]
-
-                # compute previous image: x_t -> x_t-1
-                latents = self.scheduler.step(noise_pred, t, latents, **extra_step_kwargs, return_dict=False)[0]
-
-                if callback_on_step_end is not None:
-                    callback_kwargs = {}
-                    for k in callback_on_step_end_tensor_inputs:
-                        callback_kwargs[k] = locals()[k]
-                    callback_outputs = callback_on_step_end(self, i, t, callback_kwargs)
-
-                    latents = callback_outputs.pop("latents", latents)
-                    prompt_embeds = callback_outputs.pop("prompt_embeds", prompt_embeds)
-                    negative_prompt_embeds = callback_outputs.pop("negative_prompt_embeds", negative_prompt_embeds)
-
-                # call the callback, if provided
-                if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
-                    progress_bar.update()
-
-                if XLA_AVAILABLE:
-                    xm.mark_step()
-
-        if output_type == "latent":
-            image = latents
-        else:
-            latents = latents.to(self.vae.dtype)
-            torch_accelerator_module = getattr(torch, get_device(), torch.cuda)
-            oom_error = (
-                torch.OutOfMemoryError
-                if is_torch_version(">=", "2.5.0")
-                else torch_accelerator_module.OutOfMemoryError
-            )
-            try:
-                image = self.vae.decode(latents / self.vae.config.scaling_factor, return_dict=False)[0]
-            except oom_error as e:
-                warnings.warn(
-                    f"{e}. \n"
-                    f"Try to use VAE tiling for large images. For example: \n"
-                    f"pipe.vae.enable_tiling(tile_sample_min_width=512, tile_sample_min_height=512)"
+        # 4. SiD sampling loop
+        # Initialize D_x
+        D_x = torch.zeros_like(latents).to(latents.device)
+        # Use fixed noise for now (can be extended as needed)
+        initial_latents = latents.clone()
+        for i in range(num_inference_steps):
+            if noise_type == "fresh":
+                noise = (
+                    latents if i == 0 else torch.randn_like(latents).to(latents.device)
+                )
+            elif noise_type == "ddim":
+                noise = (
+                    latents if i == 0 else ((latents - (1.0 - t) * D_x) / t).detach()
                 )
-            if use_resolution_binning:
-                image = self.image_processor.resize_and_crop_tensor(image, orig_width, orig_height)
+            elif noise_type == "fixed":
+                noise = initial_latents  # Use the initial, unmodified latents
+            else:
+                raise ValueError(f"Unknown noise_type: {noise_type}")
 
-        if not output_type == "latent":
-            image = self.image_processor.postprocess(image, output_type=output_type)
+            # Compute t value, normalized to [0, 1]
+            init_timesteps = 999
+            scalar_t = float(init_timesteps) * (
+                1.0 - float(i) / float(num_inference_steps)
+            )
+            t_val = scalar_t / 999.0
+            t = torch.full(
+                (latents.shape[0],), t_val, device=latents.device, dtype=latents.dtype
+            )
+            t_flattern = t.flatten()
+            if t.numel() > 1:
+                t = t.view(-1, 1, 1, 1)
+
+            latents = (1.0 - t) * D_x + t * noise
+            latent_model_input = latents
+
+            flow_pred = self.transformer(
+                hidden_states=latent_model_input,
+                encoder_hidden_states=prompt_embeds,
+                # encoder_attention_mask=prompt_attention_mask,
+                pooled_projections=pooled_prompt_embeds,
+                timestep=time_scale * t_flattern,
+                return_dict=False,
+            )[0]
+            D_x = latents - (
+                t * flow_pred
+                if torch.numel(t) == 1
+                else t.view(-1, 1, 1, 1) * flow_pred
+            )
 
+        # 5. Decode latent to image
+        image = self.vae.decode(
+            (D_x / self.vae.config.scaling_factor),
+            return_dict=False,
+        )[0]
+        if use_resolution_binning:
+            image = self.image_processor.resize_and_crop_tensor(image, orig_height, orig_width)
+        image = self.image_processor.postprocess(image, output_type=output_type)
         # Offload all models
         self.maybe_free_model_hooks()
 

sid/pipeline_sid_sd3.py

@@ -54,6 +54,7 @@ else:
 
 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
 
+
 # Copied from diffusers.pipelines.flux.pipeline_flux.calculate_shift
 def calculate_shift(
     image_seq_len,
@@ -683,7 +684,8 @@ class SiDSD3Pipeline(
         callback_on_step_end_tensor_inputs: List[str] = ["latents"],
         max_sequence_length: int = 256,
         use_sd3_shift: bool = False,
-        noise_type: str = 'fresh', # 'fresh', 'ddim', 'fixed'
+        noise_type: str = "fresh",  # 'fresh', 'ddim', 'fixed'
+        time_scale: float = 1000.0,
     ):
         height = height or self.default_sample_size * self.vae_scale_factor
         width = width or self.default_sample_size * self.vae_scale_factor
@@ -749,25 +751,33 @@ class SiDSD3Pipeline(
         # Use fixed noise for now (can be extended as needed)
         initial_latents = latents.clone()
         for i in range(num_inference_steps):
-            if noise_type == 'fresh':
-                noise = latents if i == 0 else torch.randn_like(latents).to(latents.device)
-            elif noise_type=='ddim':
-                noise = latents if i == 0 else ((latents - (1.0 - t) * D_x) / t).detach()
-            elif noise_type == 'fixed':
+            if noise_type == "fresh":
+                noise = (
+                    latents if i == 0 else torch.randn_like(latents).to(latents.device)
+                )
+            elif noise_type == "ddim":
+                noise = (
+                    latents if i == 0 else ((latents - (1.0 - t) * D_x) / t).detach()
+                )
+            elif noise_type == "fixed":
                 noise = initial_latents  # Use the initial, unmodified latents
             else:
                 raise ValueError(f"Unknown noise_type: {noise_type}")
-                
+
             # Compute t value, normalized to [0, 1]
             init_timesteps = 999
-            scalar_t = float(init_timesteps) * (1.0 - float(i) / float(num_inference_steps))
+            scalar_t = float(init_timesteps) * (
+                1.0 - float(i) / float(num_inference_steps)
+            )
             t_val = scalar_t / 999.0
             # t_val = 1.0 - float(i) / float(num_inference_steps)
             if use_sd3_shift:
                 shift = 3.0
                 t_val = shift * t_val / (1 + (shift - 1) * t_val)
-                
-            t = torch.full((latents.shape[0],), t_val, device=latents.device, dtype=latents.dtype)
+
+            t = torch.full(
+                (latents.shape[0],), t_val, device=latents.device, dtype=latents.dtype
+            )
             t_flattern = t.flatten()
             if t.numel() > 1:
                 t = t.view(-1, 1, 1, 1)
@@ -778,19 +788,28 @@ class SiDSD3Pipeline(
             flow_pred = self.transformer(
                 hidden_states=latent_model_input,
                 encoder_hidden_states=prompt_embeds,
-                #encoder_attention_mask=prompt_attention_mask,
+                # encoder_attention_mask=prompt_attention_mask,
                 pooled_projections=pooled_prompt_embeds,
-                timestep=1000*t_flattern,
+                timestep=time_scale * t_flattern,
                 return_dict=False,
             )[0]
-            D_x = latents - (t * flow_pred if torch.numel(t) == 1 else t.view(-1, 1, 1, 1) * flow_pred)
-        
+            D_x = latents - (
+                t * flow_pred
+                if torch.numel(t) == 1
+                else t.view(-1, 1, 1, 1) * flow_pred
+            )
+
         # 5. Decode latent to image
-        image = self.vae.decode((D_x / self.vae.config.scaling_factor) + self.vae.config.shift_factor, return_dict=False)[0]
+        image = self.vae.decode(
+            (D_x / self.vae.config.scaling_factor) + self.vae.config.shift_factor,
+            return_dict=False,
+        )[0]
         image = self.image_processor.postprocess(image, output_type=output_type)
+        
+        self.maybe_free_model_hooks()
 
         # 6. Return output
         if not return_dict:
             return (image,)
-        
-        return SiDPipelineOutput(images=image)
+
+        return SiDPipelineOutput(images=image)