ip_adapter/ip_adapter.py

import os
from typing import List

import torch
from diffusers import StableDiffusionPipeline
from diffusers.pipelines.controlnet import MultiControlNetModel
from transformers import CLIPVisionModelWithProjection, CLIPImageProcessor
from PIL import Image

from .attention_processor import IPAttnProcessor2_0 as IPAttnProcessor, AttnProcessor2_0 as AttnProcessor, CNAttnProcessor2_0 as CNAttnProcessor
from .resampler import Resampler

class ImageProjModel(torch.nn.Module):
    """Projection Model"""
    def __init__(self, cross_attention_dim=1024, clip_embeddings_dim=1024, clip_extra_context_tokens=4):
        super().__init__()
        
        self.cross_attention_dim = cross_attention_dim
        self.clip_extra_context_tokens = clip_extra_context_tokens
        self.proj = torch.nn.Linear(clip_embeddings_dim, self.clip_extra_context_tokens * cross_attention_dim)
        self.norm = torch.nn.LayerNorm(cross_attention_dim)
        
    def forward(self, image_embeds):
        embeds = image_embeds
        clip_extra_context_tokens = self.proj(embeds).reshape(-1, self.clip_extra_context_tokens, self.cross_attention_dim)
        clip_extra_context_tokens = self.norm(clip_extra_context_tokens)
        return clip_extra_context_tokens

class IPAdapter:
    
    def __init__(self, sd_pipe, image_encoder_path, ip_ckpt, device, num_tokens=4):
        
        self.device = device
        self.image_encoder_path = image_encoder_path
        self.ip_ckpt = ip_ckpt
        self.num_tokens = num_tokens
        
        self.pipe = sd_pipe.to(self.device)
        self.set_ip_adapter()
        
        # load image encoder
        self.image_encoder = CLIPVisionModelWithProjection.from_pretrained(self.image_encoder_path).to(self.device, dtype=torch.bfloat16)
        self.clip_image_processor = CLIPImageProcessor()
        # image proj model
        self.image_proj_model = self.init_proj()
        self.load_ip_adapter()
        
    def init_proj(self):
        image_proj_model = ImageProjModel(
            cross_attention_dim=self.pipe.unet.config.cross_attention_dim,
            clip_embeddings_dim=self.image_encoder.config.projection_dim,
            clip_extra_context_tokens=self.num_tokens,
        ).to(self.device, dtype=torch.bfloat16)
        return image_proj_model
        
    def set_ip_adapter(self):
        unet = self.pipe.unet
        attn_procs = {}
        for name in unet.attn_processors.keys():
            cross_attention_dim = None if name.endswith("attn1.processor") else unet.config.cross_attention_dim
            if name.startswith("mid_block"):
                hidden_size = unet.config.block_out_channels[-1]
            elif name.startswith("up_blocks"):
                block_id = int(name[len("up_blocks.")])
                hidden_size = list(reversed(unet.config.block_out_channels))[block_id]
            elif name.startswith("down_blocks"):
                block_id = int(name[len("down_blocks.")])
                hidden_size = unet.config.block_out_channels[block_id]
            if cross_attention_dim is None:
                attn_procs[name] = AttnProcessor()
            else:
                attn_procs[name] = IPAttnProcessor(hidden_size=hidden_size, cross_attention_dim=cross_attention_dim,
                scale=1.0,num_tokens= self.num_tokens).to(self.device, dtype=torch.bfloat16)
        unet.set_attn_processor(attn_procs)
        if hasattr(self.pipe, "controlnet"):
            if isinstance(self.pipe.controlnet, MultiControlNetModel):
                for controlnet in self.pipe.controlnet.nets:
                    controlnet.set_attn_processor(CNAttnProcessor(num_tokens=self.num_tokens))
            else:
                self.pipe.controlnet.set_attn_processor(CNAttnProcessor(num_tokens=self.num_tokens))

    def update_state_dict(self, state_dict):
        image_proj_dict = {}
        ip_adapter_dict = {}

        for k in state_dict.keys():
            if k.startswith("image_proj_model"):
                image_proj_dict[k.replace("image_proj_model.", "")] = state_dict[k]
            if k.startswith("adapter_modules"):
                ip_adapter_dict[k.replace("adapter_modules.", "")] = state_dict[k]

        dict = {'image_proj': image_proj_dict,
                'ip_adapter' : ip_adapter_dict
                } 
        return dict
    
    def load_ip_adapter(self):
        state_dict = torch.load(self.ip_ckpt, map_location="cpu")
        if "image_proj_model.proj.weight" in state_dict.keys():
            state_dict = self.update_state_dict(state_dict)
        self.image_proj_model.load_state_dict(state_dict["image_proj"])
        ip_layers = torch.nn.ModuleList(self.pipe.unet.attn_processors.values())
        ip_layers.load_state_dict(state_dict["ip_adapter"])
        
    @torch.inference_mode()
    def get_image_embeds(self, pil_image):
        if isinstance(pil_image, Image.Image):
            pil_image = [pil_image]
        clip_image = self.clip_image_processor(images=pil_image, return_tensors="pt").pixel_values
        print('clip_image_processor shape:',clip_image.shape)
        clip_image_embeds = self.image_encoder(clip_image.to(self.device, dtype=torch.bfloat16)).image_embeds
        print('image_encoder shape:',clip_image_embeds.shape)
        image_prompt_embeds = self.image_proj_model(clip_image_embeds)
        print('image_proj_model shape:',image_prompt_embeds.shape)
        uncond_image_prompt_embeds = self.image_proj_model(torch.zeros_like(clip_image_embeds))
        return image_prompt_embeds, uncond_image_prompt_embeds
    
    def set_scale(self, scale):
        for attn_processor in self.pipe.unet.attn_processors.values():
            if isinstance(attn_processor, IPAttnProcessor):
                attn_processor.scale = scale
                
  
    def generate(
        self,
        pil_image,
        prompt=None,
        negative_prompt=None,
        scale=1.0,
        num_samples=4,
        seed=-1,
        guidance_scale=7.5,
        num_inference_steps=30,
        **kwargs,
    ):
        self.set_scale(scale)
        
        if isinstance(pil_image, List):
            num_prompts = len(pil_image)
        else:
            num_prompts = 1
        
        # if isinstance(pil_image, Image.Image):
        #     num_prompts = 1
        # else:
        #     num_prompts = len(pil_image)
        #     print("num promp", num_prompts)
        
        if prompt is None:
            prompt = "best quality, high quality"
        if negative_prompt is None:
            negative_prompt = "monochrome, lowres, bad anatomy, worst quality, low quality"
            
        if not isinstance(prompt, List):
            prompt = [prompt] * num_prompts
        if not isinstance(negative_prompt, List):
            negative_prompt = [negative_prompt] * num_prompts
        
        image_prompt_embeds, uncond_image_prompt_embeds = self.get_image_embeds(pil_image)
        bs_embed, seq_len, _ = image_prompt_embeds.shape
        image_prompt_embeds = image_prompt_embeds.repeat(1, num_samples, 1)
        image_prompt_embeds = image_prompt_embeds.view(bs_embed * num_samples, seq_len, -1)
        uncond_image_prompt_embeds = uncond_image_prompt_embeds.repeat(1, num_samples, 1)
        uncond_image_prompt_embeds = uncond_image_prompt_embeds.view(bs_embed * num_samples, seq_len, -1)

        with torch.inference_mode():
            prompt_embeds = self.pipe._encode_prompt(
                prompt, device=self.device, num_images_per_prompt=num_samples, do_classifier_free_guidance=True, negative_prompt=negative_prompt)
            negative_prompt_embeds_, prompt_embeds_ = prompt_embeds.chunk(2)

            prompt_embeds = torch.cat([prompt_embeds_, image_prompt_embeds], dim=1)
            negative_prompt_embeds = torch.cat([negative_prompt_embeds_, uncond_image_prompt_embeds], dim=1)
            
        generator = torch.Generator(self.device).manual_seed(seed) if seed is not None else None
        images = self.pipe(
            prompt_embeds=prompt_embeds,
            negative_prompt_embeds=negative_prompt_embeds,
            guidance_scale=guidance_scale,
            num_inference_steps=num_inference_steps,
            generator=generator,
            **kwargs,
        ).images
        
        return images
    
class IPAdapterXL(IPAdapter):
    """SDXL"""
    
    def get_scale(self):
        for attn_processor in self.pipe.unet.attn_processors.values():
            if isinstance(attn_processor, IPAttnProcessor):
                print('IP attn_scale:')
                print(attn_processor.scale)
        for attn_processor in self.pipe.unet.attn_processors.values():
            if isinstance(attn_processor, AttnProcessor):
                print('UNET attn_scale:')
                print(attn_processor.scale)
                
    def generate(
        self,
        pil_image_1,
        pil_image_2=None,
        pil_image_3=None,
        pil_image_4=None,
        pil_image_5=None,
        prompt=None,
        negative_prompt=None,
        text_scale=1.0,
        ip_scale=1.0,
        scale_1=1.0,
        scale_2=1.0,
        scale_3=1.0,
        scale_4=1.0,
        scale_5=1.0,
        num_samples=1,
        seed=-1,
        num_inference_steps=30,
        guidance_scale=7.5,
        **kwargs,
    ):
        #self.get_scale()
        self.set_scale(ip_scale)
        
        if isinstance(pil_image_1, Image.Image):
            num_prompts = 1
        else:
            num_prompts = len(pil_image_1)
        
        if prompt is None:
            prompt = "best quality, high quality"
        if negative_prompt is None:
            negative_prompt = "monochrome, lowres, bad anatomy, worst quality, low quality"
            
        if not isinstance(prompt, List):
            prompt = [prompt] * num_prompts
        if not isinstance(negative_prompt, List):
            negative_prompt = [negative_prompt] * num_prompts
            
        image_prompt_embeds_list = []
        uncond_image_prompt_embeds_list = []
        print('Using primary image.')

        image_prompt_embeds_1, uncond_image_prompt_embeds_1 = self.get_image_embeds(pil_image_1)
        image_prompt_embeds_1 = image_prompt_embeds_1 * scale_1
        image_prompt_embeds_list.append(image_prompt_embeds_1)
        uncond_image_prompt_embeds_list.append(uncond_image_prompt_embeds_1)
        
        if pil_image_2 != None:
            print('Using secondary image.')
            image_prompt_embeds_2, uncond_image_prompt_embeds_2 = self.get_image_embeds(pil_image_2)
            image_prompt_embeds_2 = image_prompt_embeds_2 * scale_2
            image_prompt_embeds_list.append(image_prompt_embeds_2)
            uncond_image_prompt_embeds_list.append(uncond_image_prompt_embeds_2)
        if pil_image_3 != None:
            print('Using tertiary image.')
            image_prompt_embeds_3, uncond_image_prompt_embeds_3 = self.get_image_embeds(pil_image_3)
            image_prompt_embeds_3 = image_prompt_embeds_3 * scale_3
            image_prompt_embeds_list.append(image_prompt_embeds_3)
            uncond_image_prompt_embeds_list.append(uncond_image_prompt_embeds_3)
        if pil_image_4 != None:
            print('Using quaternary image.')
            image_prompt_embeds_4, uncond_image_prompt_embeds_4 = self.get_image_embeds(pil_image_4)
            image_prompt_embeds_4 = image_prompt_embeds_4 * scale_4
            image_prompt_embeds_list.append(image_prompt_embeds_4)
            uncond_image_prompt_embeds_list.append(uncond_image_prompt_embeds_4)
        if pil_image_5 != None:
            print('Using quinary image.')
            image_prompt_embeds_5, uncond_image_prompt_embeds_5 = self.get_image_embeds(pil_image_5)
            image_prompt_embeds_5 = image_prompt_embeds_5 * scale_5
            image_prompt_embeds_list.append(image_prompt_embeds_5)
            uncond_image_prompt_embeds_list.append(uncond_image_prompt_embeds_5)

        image_prompt_embeds = torch.cat(image_prompt_embeds_list).mean(dim=0).unsqueeze(0)
        print('catted embeds list with mean and unsqueeze shape: ',image_prompt_embeds.shape)
        bs_embed, seq_len, _ = image_prompt_embeds.shape
        image_prompt_embeds = image_prompt_embeds.repeat(1, num_samples, 1)
        print('catted embeds repeat: ',image_prompt_embeds.shape)
        image_prompt_embeds = image_prompt_embeds.view(bs_embed * num_samples, seq_len, -1)
        print('viewed embeds: ',image_prompt_embeds.shape)
        uncond_image_prompt_embeds = torch.cat(uncond_image_prompt_embeds_list).mean(dim=0).unsqueeze(0)  
        uncond_image_prompt_embeds = uncond_image_prompt_embeds.repeat(1, num_samples, 1)
        uncond_image_prompt_embeds = uncond_image_prompt_embeds.view(bs_embed * num_samples, seq_len, -1)

        with torch.inference_mode():
            prompt_embeds, negative_prompt_embeds, pooled_prompt_embeds, negative_pooled_prompt_embeds = self.pipe.encode_prompt(
                prompt, num_images_per_prompt=num_samples, do_classifier_free_guidance=True, negative_prompt=negative_prompt)
            prompt_embeds = prompt_embeds * text_scale
            prompt_embeds = torch.cat([prompt_embeds, image_prompt_embeds], dim=1)
            negative_prompt_embeds = torch.cat([negative_prompt_embeds, uncond_image_prompt_embeds], dim=1)
            
        generator = torch.Generator(self.device).manual_seed(seed) if seed is not None else None

        images = self.pipe(
            prompt_embeds=prompt_embeds,
            negative_prompt_embeds=negative_prompt_embeds,
            pooled_prompt_embeds=pooled_prompt_embeds,
            negative_pooled_prompt_embeds=negative_pooled_prompt_embeds,
            num_inference_steps=num_inference_steps,
            guidance_scale=guidance_scale,
            generator=generator,
            **kwargs,
        ).images
        
        return images
    
    
class IPAdapterPlus(IPAdapter):
    """IP-Adapter with fine-grained features"""
    
    def init_proj(self):
        image_proj_model = Resampler(
            dim=self.pipe.unet.config.cross_attention_dim,
            depth=4,
            dim_head=64,
            heads=12,
            num_queries=self.num_tokens,
            embedding_dim=self.image_encoder.config.hidden_size,
            output_dim=self.pipe.unet.config.cross_attention_dim,
            ff_mult=4
        ).to(self.device, dtype=torch.bfloat16)
        return image_proj_model
    
    @torch.inference_mode()
    def get_image_embeds(self, pil_image):
        if isinstance(pil_image, Image.Image):
            pil_image = [pil_image]
        clip_image = self.clip_image_processor(images=pil_image, return_tensors="pt").pixel_values
        clip_image = clip_image.to(self.device, dtype=torch.bfloat16)
        clip_image_embeds = self.image_encoder(clip_image, output_hidden_states=True).hidden_states[-2]
        image_prompt_embeds = self.image_proj_model(clip_image_embeds)
        uncond_clip_image_embeds = self.image_encoder(torch.zeros_like(clip_image), output_hidden_states=True).hidden_states[-2]
        uncond_image_prompt_embeds = self.image_proj_model(uncond_clip_image_embeds)
        return image_prompt_embeds, uncond_image_prompt_embeds