# Copyright (c) 2024 Jaerin Lee # Permission is hereby granted, free of charge, to any person obtaining a copy # of this software and associated documentation files (the "Software"), to deal # in the Software without restriction, including without limitation the rights # to use, copy, modify, merge, publish, distribute, sublicense, and/or sell # copies of the Software, and to permit persons to whom the Software is # furnished to do so, subject to the following conditions: # The above copyright notice and this permission notice shall be included in all # copies or substantial portions of the Software. # THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR # IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, # FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE # AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER # LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, # OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE # SOFTWARE. from transformers import Blip2Processor, Blip2ForConditionalGeneration from diffusers import DiffusionPipeline, LCMScheduler, DDIMScheduler, AutoencoderTiny import torch import torch.nn as nn import torch.nn.functional as F import torchvision.transforms as T from einops import rearrange from typing import Tuple, List, Literal, Optional, Union from tqdm import tqdm from PIL import Image from util import gaussian_lowpass, blend, get_panorama_views, shift_to_mask_bbox_center class StableMultiDiffusionPipeline(nn.Module): def __init__( self, device: torch.device, dtype: torch.dtype = torch.float16, sd_version: Literal['1.5', '2.0', '2.1', 'xl'] = '1.5', hf_key: Optional[str] = None, lora_key: Optional[str] = None, load_from_local: bool = False, # Turn on if you have already downloaed LoRA & Hugging Face hub is down. default_mask_std: float = 1.0, # 8.0 default_mask_strength: float = 1.0, default_prompt_strength: float = 1.0, # 8.0 default_bootstrap_steps: int = 1, default_boostrap_mix_steps: float = 1.0, default_bootstrap_leak_sensitivity: float = 0.2, default_preprocess_mask_cover_alpha: float = 0.3, t_index_list: List[int] = [0, 4, 12, 25, 37], # [0, 5, 16, 18, 20, 37], # [0, 12, 25, 37], # Magic number. mask_type: Literal['discrete', 'semi-continuous', 'continuous'] = 'discrete', ) -> None: r"""Stabilized MultiDiffusion for fast sampling. Accelrated region-based text-to-image synthesis with Latent Consistency Model while preserving mask fidelity and quality. Args: device (torch.device): Specify CUDA device. dtype (torch.dtype): Default precision used in the sampling process. By default, it is FP16. sd_version (Literal['1.5', '2.0', '2.1', 'xl']): StableDiffusion version. Currently, only 1.5 is supported. hf_key (Optional[str]): Custom StableDiffusion checkpoint for stylized generation. lora_key (Optional[str]): Custom LCM LoRA for acceleration. load_from_local (bool): Turn on if you have already downloaed LoRA & Hugging Face hub is down. default_mask_std (float): Preprocess mask with Gaussian blur with specified standard deviation. default_mask_strength (float): Preprocess mask by multiplying it globally with the specified variable. Caution: extremely sensitive. Recommended range: 0.98-1. default_prompt_strength (float): Preprocess foreground prompts globally by linearly interpolating its embedding with the background prompt embeddint with specified mix ratio. Useful control handle for foreground blending. Recommended range: 0.5-1. default_bootstrap_steps (int): Bootstrapping stage steps to encourage region separation. Recommended range: 1-3. default_boostrap_mix_steps (float): Bootstrapping background is a linear interpolation between background latent and the white image latent. This handle controls the mix ratio. Available range: 0-(number of bootstrapping inference steps). For example, 2.3 means that for the first two steps, white image is used as a bootstrapping background and in the third step, mixture of white (0.3) and registered background (0.7) is used as a bootstrapping background. default_bootstrap_leak_sensitivity (float): Postprocessing at each inference step by masking away the remaining bootstrap backgrounds t Recommended range: 0-1. default_preprocess_mask_cover_alpha (float): Optional preprocessing where each mask covered by other masks is reduced in its alpha value by this specified factor. t_index_list (List[int]): The default scheduling for LCM scheduler. mask_type (Literal['discrete', 'semi-continuous', 'continuous']): defines the mask quantization modes. Details in the codes of `self.process_mask`. Basically, this (subtly) controls the smoothness of foreground-background blending. More continuous means more blending, but smaller generated patch depending on the mask standard deviation. """ super().__init__() self.device = device self.dtype = dtype self.sd_version = sd_version self.default_mask_std = default_mask_std self.default_mask_strength = default_mask_strength self.default_prompt_strength = default_prompt_strength self.default_t_list = t_index_list self.default_bootstrap_steps = default_bootstrap_steps self.default_boostrap_mix_steps = default_boostrap_mix_steps self.default_bootstrap_leak_sensitivity = default_bootstrap_leak_sensitivity self.default_preprocess_mask_cover_alpha = default_preprocess_mask_cover_alpha self.mask_type = mask_type print(f'[INFO] Loading Stable Diffusion...') variant = None lora_weight_name = None if self.sd_version == '1.5': if hf_key is not None: print(f'[INFO] Using Hugging Face custom model key: {hf_key}') model_key = hf_key else: model_key = 'runwayml/stable-diffusion-v1-5' variant = 'fp16' lora_key = 'latent-consistency/lcm-lora-sdv1-5' lora_weight_name = 'pytorch_lora_weights.safetensors' # elif self.sd_version == 'xl': # model_key = 'stabilityai/stable-diffusion-xl-base-1.0' # lora_key = 'latent-consistency/lcm-lora-sdxl' # variant = 'fp16' # lora_weight_name = 'pytorch_lora_weights.safetensors' else: raise ValueError(f'Stable Diffusion version {self.sd_version} not supported.') # Create model self.i2t_processor = Blip2Processor.from_pretrained('Salesforce/blip2-opt-2.7b') self.i2t_model = Blip2ForConditionalGeneration.from_pretrained('Salesforce/blip2-opt-2.7b') self.pipe = DiffusionPipeline.from_pretrained(model_key, variant=variant, torch_dtype=dtype).to(self.device) if lora_key is None: print(f'[INFO] LCM LoRA is not available for SD version {sd_version}. Using DDIM Scheduler instead...') self.pipe.scheduler = DDIMScheduler.from_config(self.pipe.scheduler.config) self.scheduler = self.pipe.scheduler self.default_num_inference_steps = 50 self.default_guidance_scale = 7.5 else: self.pipe.scheduler = LCMScheduler.from_config(self.pipe.scheduler.config) self.scheduler = self.pipe.scheduler self.pipe.load_lora_weights(lora_key, weight_name=lora_weight_name, adapter_name='lcm') self.default_num_inference_steps = 4 self.default_guidance_scale = 1.0 self.prepare_lcm_schedule(t_index_list, 50) self.vae = self.pipe.vae self.tokenizer = self.pipe.tokenizer self.text_encoder = self.pipe.text_encoder self.unet = self.pipe.unet self.vae_scale_factor = self.pipe.vae_scale_factor # Prepare white background for bootstrapping. self.get_white_background(768, 768) print(f'[INFO] Model is loaded!') def prepare_lcm_schedule( self, t_index_list: Optional[List[int]] = None, num_inference_steps: Optional[int] = None, ) -> None: r"""Set up different inference schedule for the diffusion model. You do not have to run this explicitly if you want to use the default setting, but if you want other time schedules, run this function between the module initialization and the main call. Note: - Recommended t_index_lists for LCMs: - [0, 12, 25, 37]: Default schedule for 4 steps. Best for panorama. Not recommended if you want to use bootstrapping. Because bootstrapping stage affects the initial structuring of the generated image & in this four step LCM, this is done with only at the first step, the structure may be distorted. - [0, 4, 12, 25, 37]: Recommended if you would use 1-step boot- strapping. Default initialization in this implementation. - [0, 5, 16, 18, 20, 37]: Recommended if you would use 2-step bootstrapping. - Due to the characteristic of SD1.5 LCM LoRA, setting `num_inference_steps` larger than 20 may results in overly blurry and unrealistic images. Beware! Args: t_index_list (Optional[List[int]]): The specified scheduling step regarding the maximum timestep as `num_inference_steps`, which is by default, 50. That means that `t_index_list=[0, 12, 25, 37]` is a relative time indices basd on the full scale of 50. If None, reinitialize the module with the default value. num_inference_steps (Optional[int]): The maximum timestep of the sampler. Defines relative scale of the `t_index_list`. Rarely used in practice. If None, reinitialize the module with the default value. """ if t_index_list is None: t_index_list = self.default_t_list if num_inference_steps is None: num_inference_steps = self.default_num_inference_steps self.scheduler.set_timesteps(num_inference_steps) self.timesteps = torch.as_tensor([ self.scheduler.timesteps[t] for t in t_index_list ], dtype=torch.long) shape = (len(t_index_list), 1, 1, 1) c_skip_list = [] c_out_list = [] for timestep in self.timesteps: c_skip, c_out = self.scheduler.get_scalings_for_boundary_condition_discrete(timestep) c_skip_list.append(c_skip) c_out_list.append(c_out) self.c_skip = torch.stack(c_skip_list).view(*shape).to(dtype=self.dtype, device=self.device) self.c_out = torch.stack(c_out_list).view(*shape).to(dtype=self.dtype, device=self.device) alpha_prod_t_sqrt_list = [] beta_prod_t_sqrt_list = [] for timestep in self.timesteps: alpha_prod_t_sqrt = self.scheduler.alphas_cumprod[timestep].sqrt() beta_prod_t_sqrt = (1 - self.scheduler.alphas_cumprod[timestep]).sqrt() alpha_prod_t_sqrt_list.append(alpha_prod_t_sqrt) beta_prod_t_sqrt_list.append(beta_prod_t_sqrt) alpha_prod_t_sqrt = (torch.stack(alpha_prod_t_sqrt_list).view(*shape) .to(dtype=self.dtype, device=self.device)) beta_prod_t_sqrt = (torch.stack(beta_prod_t_sqrt_list).view(*shape) .to(dtype=self.dtype, device=self.device)) self.alpha_prod_t_sqrt = alpha_prod_t_sqrt self.beta_prod_t_sqrt = beta_prod_t_sqrt noise_lvs = (1 - self.scheduler.alphas_cumprod[self.timesteps].to(self.device)) ** 0.5 self.noise_lvs = noise_lvs[None, :, None, None, None] self.next_noise_lvs = torch.cat([noise_lvs[1:], noise_lvs.new_zeros(1)])[None, :, None, None, None] @torch.no_grad() def get_text_embeds(self, prompt: str, negative_prompt: str) -> Tuple[torch.Tensor]: r"""Text embeddings from string text prompts. Args: prompt (str): A text prompt string. negative_prompt: An optional negative text prompt string. Good for high-quality generation. Returns: A tuple of (negative, positive) prompt embeddings of (1, 77, 768). """ kwargs = dict(padding='max_length', max_length=self.tokenizer.model_max_length, return_tensors='pt') # Tokenize text and get embeddings. text_input = self.tokenizer(prompt, truncation=True, **kwargs) text_embeds = self.text_encoder(text_input.input_ids.to(self.device))[0] uncond_input = self.tokenizer(negative_prompt, **kwargs) uncond_embeds = self.text_encoder(uncond_input.input_ids.to(self.device))[0] return uncond_embeds, text_embeds @torch.no_grad() def get_text_prompts(self, image: Image.Image) -> str: r"""A convenient method to extract text prompt from an image. This is called if the user does not provide background prompt but only the background image. We use BLIP-2 to automatically generate prompts. Args: image (Image.Image): A PIL image. Returns: A single string of text prompt. """ question = 'Question: What are in the image? Answer:' inputs = self.i2t_processor(image, question, return_tensors='pt') out = self.i2t_model.generate(**inputs, max_new_tokens=77) prompt = self.i2t_processor.decode(out[0], skip_special_tokens=True).strip() return prompt @torch.no_grad() def encode_imgs( self, imgs: torch.Tensor, generator: Optional[torch.Generator] = None, vae: Optional[nn.Module] = None, ) -> torch.Tensor: r"""A wrapper function for VAE encoder of the latent diffusion model. Args: imgs (torch.Tensor): An image to get StableDiffusion latents. Expected shape: (B, 3, H, W). Expected pixel scale: [0, 1]. generator (Optional[torch.Generator]): Seed for KL-Autoencoder. vae (Optional[nn.Module]): Explicitly specify VAE (used for the demo application with TinyVAE). Returns: An image latent embedding with 1/8 size (depending on the auto- encoder. Shape: (B, 4, H//8, W//8). """ def _retrieve_latents( encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = 'sample', ): if hasattr(encoder_output, 'latent_dist') and sample_mode == 'sample': return encoder_output.latent_dist.sample(generator) elif hasattr(encoder_output, 'latent_dist') and sample_mode == 'argmax': return encoder_output.latent_dist.mode() elif hasattr(encoder_output, 'latents'): return encoder_output.latents else: raise AttributeError('Could not access latents of provided encoder_output') vae = self.vae if vae is None else vae imgs = 2 * imgs - 1 latents = vae.config.scaling_factor * _retrieve_latents(vae.encode(imgs), generator=generator) return latents @torch.no_grad() def decode_latents(self, latents: torch.Tensor, vae: Optional[nn.Module] = None) -> torch.Tensor: r"""A wrapper function for VAE decoder of the latent diffusion model. Args: latents (torch.Tensor): An image latent to get associated images. Expected shape: (B, 4, H//8, W//8). vae (Optional[nn.Module]): Explicitly specify VAE (used for the demo application with TinyVAE). Returns: An image latent embedding with 1/8 size (depending on the auto- encoder. Shape: (B, 3, H, W). """ vae = self.vae if vae is None else vae latents = 1 / vae.config.scaling_factor * latents imgs = vae.decode(latents).sample imgs = (imgs / 2 + 0.5).clip_(0, 1) return imgs @torch.no_grad() def get_white_background(self, height: int, width: int) -> torch.Tensor: r"""White background image latent for bootstrapping or in case of absent background. Additionally stores the maximally-sized white latent for fast retrieval in the future. By default, we initially call this with 768x768 sized white image, so the function is rarely visited twice. Args: height (int): The height of the white *image*, not its latent. width (int): The width of the white *image*, not its latent. Returns: A white image latent of size (1, 4, height//8, width//8). A cropped version of the stored white latent is returned if the requested size is smaller than what we already have created. """ if not hasattr(self, 'white') or self.white.shape[-2] < height or self.white.shape[-1] < width: white = torch.ones(1, 3, height, width, dtype=self.dtype, device=self.device) self.white = self.encode_imgs(white) return self.white return self.white[..., :(height // self.vae_scale_factor), :(width // self.vae_scale_factor)] @torch.no_grad() def process_mask( self, masks: Union[torch.Tensor, Image.Image, List[Image.Image]], strength: Optional[Union[torch.Tensor, float]] = None, std: Optional[Union[torch.Tensor, float]] = None, height: int = 512, width: int = 512, use_boolean_mask: bool = True, timesteps: Optional[torch.Tensor] = None, preprocess_mask_cover_alpha: Optional[float] = None, ) -> Tuple[torch.Tensor]: r"""Fast preprocess of masks for region-based generation with fine- grained controls. Mask preprocessing is done in four steps: 1. Resizing: Resize the masks into the specified width and height by nearest neighbor interpolation. 2. (Optional) Ordering: Masks with higher indices are considered to cover the masks with smaller indices. Covered masks are decayed in its alpha value by the specified factor of `preprocess_mask_cover_alpha`. 3. Blurring: Gaussian blur is applied to the mask with the specified standard deviation (isotropic). This results in gradual increase of masked region as the timesteps evolve, naturally blending fore- ground and the predesignated background. Not strictly required if you want to produce images from scratch withoout background. 4. Quantization: Split the real-numbered masks of value between [0, 1] into predefined noise levels for each quantized scheduling step of the diffusion sampler. For example, if the diffusion model sampler has noise level of [0.9977, 0.9912, 0.9735, 0.8499, 0.5840], which is the default noise level of this module with schedule [0, 4, 12, 25, 37], the masks are split into binary masks whose values are greater than these levels. This results in tradual increase of mask region as the timesteps increase. Details are described in our paper at https://arxiv.org/pdf/2403.09055.pdf. On the Three Modes of `mask_type`: `self.mask_type` is predefined at the initialization stage of this pipeline. Three possible modes are available: 'discrete', 'semi- continuous', and 'continuous'. These define the mask quantization modes we use. Basically, this (subtly) controls the smoothness of foreground-background blending. Continuous modes produces nonbinary masks to further blend foreground and background latents by linear- ly interpolating between them. Semi-continuous masks only applies continuous mask at the last step of the LCM sampler. Due to the large step size of the LCM scheduler, we find that our continuous blending helps generating seamless inpainting and editing results. Args: masks (Union[torch.Tensor, Image.Image, List[Image.Image]]): Masks. strength (Optional[Union[torch.Tensor, float]]): Mask strength that overrides the default value. A globally multiplied factor to the mask at the initial stage of processing. Can be applied seperately for each mask. std (Optional[Union[torch.Tensor, float]]): Mask blurring Gaussian kernel's standard deviation. Overrides the default value. Can be applied seperately for each mask. height (int): The height of the expected generation. Mask is resized to (height//8, width//8) with nearest neighbor inter- polation. width (int): The width of the expected generation. Mask is resized to (height//8, width//8) with nearest neighbor interpolation. use_boolean_mask (bool): Specify this to treat the mask image as a boolean tensor. The retion with dark part darker than 0.5 of the maximal pixel value (that is, 127.5) is considered as the designated mask. timesteps (Optional[torch.Tensor]): Defines the scheduler noise levels that acts as bins of mask quantization. preprocess_mask_cover_alpha (Optional[float]): Optional pre- processing where each mask covered by other masks is reduced in its alpha value by this specified factor. Overrides the default value. Returns: A tuple of tensors. - masks: Preprocessed (ordered, blurred, and quantized) binary/non- binary masks (see the explanation on `mask_type` above) for region-based image synthesis. - masks_blurred: Gaussian blurred masks. Used for optionally specified foreground-background blending after image generation. - std: Mask blur standard deviation. Used for optionally specified foreground-background blending after image generation. """ if isinstance(masks, Image.Image): masks = [masks] if isinstance(masks, (tuple, list)): # Assumes white background for Image.Image; # inverted boolean masks with shape (1, 1, H, W) for torch.Tensor. if use_boolean_mask: proc = lambda m: T.ToTensor()(m)[None, -1:] < 0.5 else: proc = lambda m: 1.0 - T.ToTensor()(m)[None, -1:] masks = torch.cat([proc(mask) for mask in masks], dim=0).float().clip_(0, 1) masks = F.interpolate(masks.float(), size=(height, width), mode='bilinear', align_corners=False) masks = masks.to(self.device) # Background mask alpha is decayed by the specified factor where foreground masks covers it. if preprocess_mask_cover_alpha is None: preprocess_mask_cover_alpha = self.default_preprocess_mask_cover_alpha if preprocess_mask_cover_alpha > 0: masks = torch.stack([ torch.where( masks[i + 1:].sum(dim=0) > 0, mask * preprocess_mask_cover_alpha, mask, ) if i < len(masks) - 1 else mask for i, mask in enumerate(masks) ], dim=0) # Scheduler noise levels for mask quantization. if timesteps is None: noise_lvs = self.noise_lvs next_noise_lvs = self.next_noise_lvs else: noise_lvs_ = (1 - self.scheduler.alphas_cumprod[timesteps].to(self.device)) ** 0.5 noise_lvs = noise_lvs_[None, :, None, None, None] next_noise_lvs = torch.cat([noise_lvs_[1:], noise_lvs_.new_zeros(1)])[None, :, None, None, None] # Mask preprocessing parameters are fetched from the default settings. if std is None: std = self.default_mask_std if isinstance(std, (int, float)): std = [std] * len(masks) if isinstance(std, (list, tuple)): std = torch.as_tensor(std, dtype=torch.float, device=self.device) if strength is None: strength = self.default_mask_strength if isinstance(strength, (int, float)): strength = [strength] * len(masks) if isinstance(strength, (list, tuple)): strength = torch.as_tensor(strength, dtype=torch.float, device=self.device) if (std > 0).any(): std = torch.where(std > 0, std, 1e-5) masks = gaussian_lowpass(masks, std) masks_blurred = masks # NOTE: This `strength` aligns with `denoising strength`. However, with LCM, using strength < 0.96 # gives unpleasant results. masks = masks * strength[:, None, None, None] masks = masks.unsqueeze(1).repeat(1, noise_lvs.shape[1], 1, 1, 1) # Mask is quantized according to the current noise levels specified by the scheduler. if self.mask_type == 'discrete': # Discrete mode. masks = masks > noise_lvs elif self.mask_type == 'semi-continuous': # Semi-continuous mode (continuous at the last step only). masks = torch.cat(( masks[:, :-1] > noise_lvs[:, :-1], ( (masks[:, -1:] - next_noise_lvs[:, -1:]) / (noise_lvs[:, -1:] - next_noise_lvs[:, -1:]) ).clip_(0, 1), ), dim=1) elif self.mask_type == 'continuous': # Continuous mode: Have the exact same `1` coverage with discrete mode, but the mask gradually # decreases continuously after the discrete mode boundary to become `0` at the # next lower threshold. masks = ((masks - next_noise_lvs) / (noise_lvs - next_noise_lvs)).clip_(0, 1) # NOTE: Post processing mask strength does not align with conventional 'denoising_strength'. However, # fine-grained mask alpha channel tuning is available with this form. # masks = masks * strength[None, :, None, None, None] h = height // self.vae_scale_factor w = width // self.vae_scale_factor masks = rearrange(masks.float(), 'p t () h w -> (p t) () h w') masks = F.interpolate(masks, size=(h, w), mode='nearest') masks = rearrange(masks.to(self.dtype), '(p t) () h w -> p t () h w', p=len(std)) return masks, masks_blurred, std def scheduler_step( self, noise_pred: torch.Tensor, idx: int, latent: torch.Tensor, ) -> torch.Tensor: r"""Denoise-only step for reverse diffusion scheduler. Designed to match the interface of the original `pipe.scheduler.step`, which is a combination of this method and the following `scheduler_add_noise`. Args: noise_pred (torch.Tensor): Noise prediction results from the U-Net. idx (int): Instead of timesteps (in [0, 1000]-scale) use indices for the timesteps tensor (ranged in [0, len(timesteps)-1]). latent (torch.Tensor): Noisy latent. Returns: A denoised tensor with the same size as latent. """ F_theta = (latent - self.beta_prod_t_sqrt[idx] * noise_pred) / self.alpha_prod_t_sqrt[idx] return self.c_out[idx] * F_theta + self.c_skip[idx] * latent def scheduler_add_noise( self, latent: torch.Tensor, noise: Optional[torch.Tensor], idx: int, ) -> torch.Tensor: r"""Separated noise-add step for the reverse diffusion scheduler. Designed to match the interface of the original `pipe.scheduler.add_noise`. Args: latent (torch.Tensor): Denoised latent. noise (torch.Tensor): Added noise. Can be None. If None, a random noise is newly sampled for addition. idx (int): Instead of timesteps (in [0, 1000]-scale) use indices for the timesteps tensor (ranged in [0, len(timesteps)-1]). Returns: A noisy tensor with the same size as latent. """ if idx >= len(self.alpha_prod_t_sqrt) or idx < 0: # The last step does not require noise addition. return latent noise = torch.randn_like(latent) if noise is None else noise return self.alpha_prod_t_sqrt[idx] * latent + self.beta_prod_t_sqrt[idx] * noise @torch.no_grad() def sample( self, prompts: Union[str, List[str]], negative_prompts: Union[str, List[str]] = '', height: int = 512, width: int = 512, num_inference_steps: Optional[int] = None, guidance_scale: Optional[float] = None, batch_size: int = 1, ) -> Image.Image: r"""StableDiffusionPipeline for single-prompt single-tile generation. Minimal Example: >>> device = torch.device('cuda:0') >>> smd = StableMultiDiffusionPipeline(device) >>> image = smd.sample('A photo of the dolomites') >>> image.save('my_creation.png') Args: prompts (Union[str, List[str]]): A text prompt. negative_prompts (Union[str, List[str]]): A negative text prompt. height (int): Height of a generated image. width (int): Width of a generated image. num_inference_steps (Optional[int]): Number of inference steps. Default inference scheduling is used if none is specified. guidance_scale (Optional[float]): Classifier guidance scale. Default value is used if none is specified. batch_size (int): Number of images to generate. Returns: A PIL.Image image. """ if num_inference_steps is None: num_inference_steps = self.default_num_inference_steps if guidance_scale is None: guidance_scale = self.default_guidance_scale self.scheduler.set_timesteps(num_inference_steps) if isinstance(prompts, str): prompts = [prompts] if isinstance(negative_prompts, str): negative_prompts = [negative_prompts] # Calculate text embeddings. uncond_embeds, text_embeds = self.get_text_embeds(prompts, negative_prompts) # [2, 77, 768] text_embeds = torch.cat([uncond_embeds.mean(dim=0, keepdim=True), text_embeds.mean(dim=0, keepdim=True)]) h = height // self.vae_scale_factor w = width // self.vae_scale_factor latent = torch.randn((batch_size, self.unet.config.in_channels, h, w), dtype=self.dtype, device=self.device) with torch.autocast('cuda'): for i, t in enumerate(tqdm(self.scheduler.timesteps)): # Expand the latents if we are doing classifier-free guidance. latent_model_input = torch.cat([latent] * 2) # Perform one step of the reverse diffusion. noise_pred = self.unet(latent_model_input, t, encoder_hidden_states=text_embeds)['sample'] noise_pred_uncond, noise_pred_cond = noise_pred.chunk(2) noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_cond - noise_pred_uncond) latent = self.scheduler.step(noise_pred, t, latent)['prev_sample'] # Return PIL Image. latent = latent.to(dtype=self.dtype) imgs = [T.ToPILImage()(self.decode_latents(l[None])[0]) for l in latent] return imgs @torch.no_grad() def sample_panorama( self, prompts: Union[str, List[str]], negative_prompts: Union[str, List[str]] = '', height: int = 512, width: int = 2048, num_inference_steps: Optional[int] = None, guidance_scale: Optional[float] = None, tile_size: Optional[int] = None, ) -> Image.Image: r"""Large size image generation from a single set of prompts. Minimal Example: >>> device = torch.device('cuda:0') >>> smd = StableMultiDiffusionPipeline(device) >>> image = smd.sample_panorama( >>> 'A photo of Alps', height=512, width=3072) >>> image.save('my_panorama_creation.png') Args: prompts (Union[str, List[str]]): A text prompt. negative_prompts (Union[str, List[str]]): A negative text prompt. height (int): Height of a generated image. It is tiled if larger than `tile_size`. width (int): Width of a generated image. It is tiled if larger than `tile_size`. num_inference_steps (Optional[int]): Number of inference steps. Default inference scheduling is used if none is specified. guidance_scale (Optional[float]): Classifier guidance scale. Default value is used if none is specified. tile_size (Optional[int]): Tile size of the panorama generation. Works best with the default training size of the Stable- Diffusion model, i.e., 512 or 768 for SD1.5 and 1024 for SDXL. Returns: A PIL.Image image of a panorama (large-size) image. """ if num_inference_steps is None: num_inference_steps = self.default_num_inference_steps self.scheduler.set_timesteps(num_inference_steps) timesteps = self.timesteps use_custom_timesteps = False else: self.scheduler.set_timesteps(num_inference_steps) timesteps = self.scheduler.timesteps use_custom_timesteps = True if guidance_scale is None: guidance_scale = self.default_guidance_scale if isinstance(prompts, str): prompts = [prompts] if isinstance(negative_prompts, str): negative_prompts = [negative_prompts] # Calculate text embeddings. uncond_embeds, text_embeds = self.get_text_embeds(prompts, negative_prompts) # [2, 77, 768] text_embeds = torch.cat([uncond_embeds.mean(dim=0, keepdim=True), text_embeds.mean(dim=0, keepdim=True)]) # Define panorama grid and get views h = height // self.vae_scale_factor w = width // self.vae_scale_factor latent = torch.randn((1, self.unet.config.in_channels, h, w), dtype=self.dtype, device=self.device) if tile_size is None: tile_size = min(min(height, width), 512) views, masks = get_panorama_views(h, w, tile_size // self.vae_scale_factor) masks = masks.to(dtype=self.dtype, device=self.device) value = torch.zeros_like(latent) with torch.autocast('cuda'): for i, t in enumerate(tqdm(timesteps)): value.zero_() for j, (h_start, h_end, w_start, w_end) in enumerate(views): # TODO we can support batches, and pass multiple views at once to the unet latent_view = latent[:, :, h_start:h_end, w_start:w_end] # Expand the latents if we are doing classifier-free guidance. latent_model_input = torch.cat([latent_view] * 2) # Perform one step of the reverse diffusion. noise_pred = self.unet(latent_model_input, t, encoder_hidden_states=text_embeds)['sample'] noise_pred_uncond, noise_pred_cond = noise_pred.chunk(2) noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_cond - noise_pred_uncond) # Compute the denoising step. latents_view_denoised = self.scheduler_step(noise_pred, i, latent_view) # (1, 4, h, w) mask = masks[..., j:j + 1, h_start:h_end, w_start:w_end] # (1, 1, h, w) value[..., h_start:h_end, w_start:w_end] += mask * latents_view_denoised # (1, 1, h, w) # Update denoised latent. latent = value.clone() if i < len(timesteps) - 1: latent = self.scheduler_add_noise(latent, None, i + 1) # Return PIL Image. imgs = self.decode_latents(latent) img = T.ToPILImage()(imgs[0].cpu()) return img @torch.no_grad() def __call__( self, prompts: Optional[Union[str, List[str]]] = None, negative_prompts: Union[str, List[str]] = '', suffix: Optional[str] = None, #', background is ', background: Optional[Union[torch.Tensor, Image.Image]] = None, background_prompt: Optional[str] = None, background_negative_prompt: str = '', height: int = 512, width: int = 512, num_inference_steps: Optional[int] = None, guidance_scale: Optional[float] = None, prompt_strengths: Optional[Union[torch.Tensor, float, List[float]]] = None, masks: Optional[Union[Image.Image, List[Image.Image]]] = None, mask_strengths: Optional[Union[torch.Tensor, float, List[float]]] = None, mask_stds: Optional[Union[torch.Tensor, float, List[float]]] = None, use_boolean_mask: bool = True, do_blend: bool = True, tile_size: int = 768, bootstrap_steps: Optional[int] = None, boostrap_mix_steps: Optional[float] = None, bootstrap_leak_sensitivity: Optional[float] = None, preprocess_mask_cover_alpha: Optional[float] = None, ) -> Image.Image: r"""Arbitrary-size image generation from multiple pairs of (regional) text prompt-mask pairs. This is a main routine for this pipeline. Example: >>> device = torch.device('cuda:0') >>> smd = StableMultiDiffusionPipeline(device) >>> prompts = {... specify prompts} >>> masks = {... specify mask tensors} >>> height, width = masks.shape[-2:] >>> image = smd( >>> prompts, masks=masks.float(), height=height, width=width) >>> image.save('my_beautiful_creation.png') Args: prompts (Union[str, List[str]]): A text prompt. negative_prompts (Union[str, List[str]]): A negative text prompt. suffix (Optional[str]): One option for blending foreground prompts with background prompts by simply appending background prompt to the end of each foreground prompt with this `middle word` in between. For example, if you set this as `, background is`, then the foreground prompt will be changed into `(fg), background is (bg)` before conditional generation. background (Optional[Union[torch.Tensor, Image.Image]]): a background image, if the user wants to draw in front of the specified image. Background prompt will automatically generated with a BLIP-2 model. background_prompt (Optional[str]): The background prompt is used for preprocessing foreground prompt embeddings to blend foreground and background. background_negative_prompt (Optional[str]): The negative background prompt. height (int): Height of a generated image. It is tiled if larger than `tile_size`. width (int): Width of a generated image. It is tiled if larger than `tile_size`. num_inference_steps (Optional[int]): Number of inference steps. Default inference scheduling is used if none is specified. guidance_scale (Optional[float]): Classifier guidance scale. Default value is used if none is specified. prompt_strength (float): Overrides default value. Preprocess foreground prompts globally by linearly interpolating its embedding with the background prompt embeddint with specified mix ratio. Useful control handle for foreground blending. Recommended range: 0.5-1. masks (Optional[Union[Image.Image, List[Image.Image]]]): a list of mask images. Each mask associates with each of the text prompts and each of the negative prompts. If specified as an image, it regards the image as a boolean mask. Also accepts torch.Tensor masks, which can have nonbinary values for fine-grained controls in mixing regional generations. mask_strengths (Optional[Union[torch.Tensor, float, List[float]]]): Overrides the default value. an be assigned for each mask separately. Preprocess mask by multiplying it globally with the specified variable. Caution: extremely sensitive. Recommended range: 0.98-1. mask_stds (Optional[Union[torch.Tensor, float, List[float]]]): Overrides the default value. Can be assigned for each mask separately. Preprocess mask with Gaussian blur with specified standard deviation. Recommended range: 0-64. use_boolean_mask (bool): Turn this off if you want to treat the mask image as nonbinary one. The module will use the last channel of the given image in `masks` as the mask value. do_blend (bool): Blend the generated foreground and the optionally predefined background by smooth boundary obtained from Gaussian blurs of the foreground `masks` with the given `mask_stds`. tile_size (Optional[int]): Tile size of the panorama generation. Works best with the default training size of the Stable- Diffusion model, i.e., 512 or 768 for SD1.5 and 1024 for SDXL. bootstrap_steps (int): Overrides the default value. Bootstrapping stage steps to encourage region separation. Recommended range: 1-3. boostrap_mix_steps (float): Overrides the default value. Bootstrapping background is a linear interpolation between background latent and the white image latent. This handle controls the mix ratio. Available range: 0-(number of bootstrapping inference steps). For example, 2.3 means that for the first two steps, white image is used as a bootstrapping background and in the third step, mixture of white (0.3) and registered background (0.7) is used as a bootstrapping background. bootstrap_leak_sensitivity (float): Overrides the default value. Postprocessing at each inference step by masking away the remaining bootstrap backgrounds t Recommended range: 0-1. preprocess_mask_cover_alpha (float): Overrides the default value. Optional preprocessing where each mask covered by other masks is reduced in its alpha value by this specified factor. Returns: A PIL.Image image of a panorama (large-size) image. """ ### Simplest cases # prompts is None: return background. # masks is None but prompts is not None: return prompts # masks is not None and prompts is not None: Do StableMultiDiffusion. if prompts is None or (isinstance(prompts, (list, tuple, str)) and len(prompts) == 0): if background is None and background_prompt is not None: return sample(background_prompt, background_negative_prompt, height, width, num_inference_steps, guidance_scale) return background elif masks is None or (isinstance(masks, (list, tuple)) and len(masks) == 0): return sample(prompts, negative_prompts, height, width, num_inference_steps, guidance_scale) ### Prepare generation if num_inference_steps is not None: self.prepare_lcm_schedule(list(range(num_inference_steps)), num_inference_steps) if guidance_scale is None: guidance_scale = self.default_guidance_scale ### Prompts & Masks # asserts #m > 0 and #p > 0. # #m == #p == #n > 0: We happily generate according to the prompts & masks. # #m != #p: #p should be 1 and we will broadcast text embeds of p through m masks. # #p != #n: #n should be 1 and we will broadcast negative embeds n through p prompts. if isinstance(masks, Image.Image): masks = [masks] if isinstance(prompts, str): prompts = [prompts] if isinstance(negative_prompts, str): negative_prompts = [negative_prompts] num_masks = len(masks) num_prompts = len(prompts) num_nprompts = len(negative_prompts) assert num_prompts in (num_masks, 1), \ f'The number of prompts {num_prompts} should match the number of masks {num_masks}!' assert num_nprompts in (num_prompts, 1), \ f'The number of negative prompts {num_nprompts} should match the number of prompts {num_prompts}!' fg_masks, masks_g, std = self.process_mask( masks, mask_strengths, mask_stds, height=height, width=width, use_boolean_mask=use_boolean_mask, timesteps=self.timesteps, preprocess_mask_cover_alpha=preprocess_mask_cover_alpha, ) # (p, t, 1, H, W) bg_masks = (1 - fg_masks.sum(dim=0)).clip_(0, 1) # (T, 1, h, w) has_background = bg_masks.sum() > 0 h = (height + self.vae_scale_factor - 1) // self.vae_scale_factor w = (width + self.vae_scale_factor - 1) // self.vae_scale_factor ### Background # background == None && background_prompt == None: Initialize with white background. # background == None && background_prompt != None: Generate background *along with other prompts*. # background != None && background_prompt == None: Retrieve text prompt using BLIP. # background != None && background_prompt != None: Use the given arguments. # not has_background: no effect of prompt_strength (the mix ratio between fg prompt & bg prompt) # has_background && prompt_strength != 1: mix only for this case. bg_latent = None if has_background: if background is None and background_prompt is not None: fg_masks = torch.cat((bg_masks[None], fg_masks), dim=0) if suffix is not None: prompts = [p + suffix + background_prompt for p in prompts] prompts = [background_prompt] + prompts negative_prompts = [background_negative_prompt] + negative_prompts has_background = False # Regard that background does not exist. else: if background is None and background_prompt is None: background = torch.ones(1, 3, height, width, dtype=self.dtype, device=self.device) background_prompt = 'simple white background image' elif background is not None and background_prompt is None: background_prompt = self.get_text_prompts(background) if suffix is not None: prompts = [p + suffix + background_prompt for p in prompts] prompts = [background_prompt] + prompts negative_prompts = [background_negative_prompt] + negative_prompts if isinstance(background, Image.Image): background = T.ToTensor()(background).to(dtype=self.dtype, device=self.device)[None] background = F.interpolate(background, size=(height, width), mode='bicubic', align_corners=False) bg_latent = self.encode_imgs(background) # Bootstrapping stage preparation. if bootstrap_steps is None: bootstrap_steps = self.default_bootstrap_steps if boostrap_mix_steps is None: boostrap_mix_steps = self.default_boostrap_mix_steps if bootstrap_leak_sensitivity is None: bootstrap_leak_sensitivity = self.default_bootstrap_leak_sensitivity if bootstrap_steps > 0: height_ = min(height, tile_size) width_ = min(width, tile_size) white = self.get_white_background(height, width) # (1, 4, h, w) ### Prepare text embeddings (optimized for the minimal encoder batch size) uncond_embeds, text_embeds = self.get_text_embeds(prompts, negative_prompts) # [2 * len(prompts), 77, 768] if has_background: # First channel is background prompt text embeds. Background prompt itself is not used for generation. s = prompt_strengths if prompt_strengths is None: s = self.default_prompt_strength if isinstance(s, (int, float)): s = [s] * num_prompts if isinstance(s, (list, tuple)): assert len(s) == num_prompts, \ f'The number of prompt strengths {len(s)} should match the number of prompts {num_prompts}!' s = torch.as_tensor(s, dtype=self.dtype, device=self.device) s = s[:, None, None] be = text_embeds[:1] bu = uncond_embeds[:1] fe = text_embeds[1:] fu = uncond_embeds[1:] if num_prompts > num_nprompts: # # negative prompts = 1; # prompts > 1. assert fu.shape[0] == 1 and fe.shape == num_prompts fu = fu.repeat(num_prompts, 1, 1) text_embeds = torch.lerp(be, fe, s) # (p, 77, 768) uncond_embeds = torch.lerp(bu, fu, s) # (n, 77, 768) elif num_prompts > num_nprompts: # # negative prompts = 1; # prompts > 1. assert uncond_embeds.shape[0] == 1 and text_embeds.shape[0] == num_prompts uncond_embeds = uncond_embeds.repeat(num_prompts, 1, 1) assert uncond_embeds.shape[0] == text_embeds.shape[0] == num_prompts if num_masks > num_prompts: assert masks.shape[0] == num_masks and num_prompts == 1 text_embeds = text_embeds.repeat(num_masks, 1, 1) uncond_embeds = uncond_embeds.repeat(num_masks, 1, 1) text_embeds = torch.cat([uncond_embeds, text_embeds]) ### Run # Latent initialization. if self.timesteps[0] < 999 and has_background: latent = self.scheduler_add_noise(bg_latent, None, 0) else: latent = torch.randn((1, self.unet.config.in_channels, h, w), dtype=self.dtype, device=self.device) # Tiling (if needed). if height > tile_size or width > tile_size: t = (tile_size + self.vae_scale_factor - 1) // self.vae_scale_factor views, tile_masks = get_panorama_views(h, w, t) tile_masks = tile_masks.to(self.device) else: views = [(0, h, 0, w)] tile_masks = latent.new_ones((1, 1, h, w)) value = torch.zeros_like(latent) count_all = torch.zeros_like(latent) with torch.autocast('cuda'): for i, t in enumerate(tqdm(self.timesteps)): fg_mask = fg_masks[:, i] bg_mask = bg_masks[i:i + 1] value.zero_() count_all.zero_() for j, (h_start, h_end, w_start, w_end) in enumerate(views): fg_mask_ = fg_mask[..., h_start:h_end, w_start:w_end] latent_ = latent[..., h_start:h_end, w_start:w_end].repeat(num_masks, 1, 1, 1) # Bootstrap for tight background. if i < bootstrap_steps: mix_ratio = min(1, max(0, boostrap_mix_steps - i)) # Treat the first foreground latent as the background latent if one does not exist. bg_latent_ = bg_latent[..., h_start:h_end, w_start:w_end] if has_background else latent_[:1] white_ = white[..., h_start:h_end, w_start:w_end] bg_latent_ = mix_ratio * white_ + (1.0 - mix_ratio) * bg_latent_ bg_latent_ = self.scheduler_add_noise(bg_latent_, None, i) latent_ = (1.0 - fg_mask_) * bg_latent_ + fg_mask_ * latent_ # Centering. latent_ = shift_to_mask_bbox_center(latent_, fg_mask_, reverse=True) # Perform one step of the reverse diffusion. noise_pred = self.unet(torch.cat([latent_] * 2), t, encoder_hidden_states=text_embeds)['sample'] noise_pred_uncond, noise_pred_cond = noise_pred.chunk(2) noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_cond - noise_pred_uncond) latent_ = self.scheduler_step(noise_pred, i, latent_) if i < bootstrap_steps: # Uncentering. latent_ = shift_to_mask_bbox_center(latent_, fg_mask_) # Remove leakage (optional). leak = (latent_ - bg_latent_).pow(2).mean(dim=1, keepdim=True) leak_sigmoid = torch.sigmoid(leak / bootstrap_leak_sensitivity) * 2 - 1 fg_mask_ = fg_mask_ * leak_sigmoid # Mix the latents. fg_mask_ = fg_mask_ * tile_masks[:, j:j + 1, h_start:h_end, w_start:w_end] value[..., h_start:h_end, w_start:w_end] += (fg_mask_ * latent_).sum(dim=0, keepdim=True) count_all[..., h_start:h_end, w_start:w_end] += fg_mask_.sum(dim=0, keepdim=True) latent = torch.where(count_all > 0, value / count_all, value) bg_mask = (1 - count_all).clip_(0, 1) # (T, 1, h, w) if has_background: latent = (1 - bg_mask) * latent + bg_mask * bg_latent # Noise is added after mixing. if i < len(self.timesteps) - 1: latent = self.scheduler_add_noise(latent, None, i + 1) # Return PIL Image. image = self.decode_latents(latent.to(dtype=self.dtype))[0] if has_background and do_blend: fg_mask = torch.sum(masks_g, dim=0).clip_(0, 1) image = blend(image, background[0], fg_mask) else: image = T.ToPILImage()(image) return image