diffusers 0.32.2__py3-none-any.whl → 0.33.0__py3-none-any.whl

diffusers/pipelines/flux/pipeline_flux_fill.py

@@ -82,7 +82,7 @@ def calculate_shift(
     base_seq_len: int = 256,
     max_seq_len: int = 4096,
     base_shift: float = 0.5,
-    max_shift: float = 1.16,
+    max_shift: float = 1.15,
 ):
     m = (max_shift - base_shift) / (max_seq_len - base_seq_len)
     b = base_shift - m * base_seq_len
@@ -221,15 +221,16 @@ class FluxFillPipeline(
             transformer=transformer,
             scheduler=scheduler,
         )
-        self.vae_scale_factor = (
-            2 ** (len(self.vae.config.block_out_channels) - 1) if hasattr(self, "vae") and self.vae is not None else 8
-        )
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         # Flux latents are turned into 2x2 patches and packed. This means the latent width and height has to be divisible
         # by the patch size. So the vae scale factor is multiplied by the patch size to account for this
-        self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor * 2)
+        self.latent_channels = self.vae.config.latent_channels if getattr(self, "vae", None) else 16
+        self.image_processor = VaeImageProcessor(
+            vae_scale_factor=self.vae_scale_factor * 2, vae_latent_channels=self.latent_channels
+        )
         self.mask_processor = VaeImageProcessor(
             vae_scale_factor=self.vae_scale_factor * 2,
-            vae_latent_channels=self.vae.config.latent_channels,
+            vae_latent_channels=self.latent_channels,
             do_normalize=False,
             do_binarize=True,
             do_convert_grayscale=True,
@@ -494,10 +495,38 @@ class FluxFillPipeline(
 
         return prompt_embeds, pooled_prompt_embeds, text_ids
 
+    # Copied from diffusers.pipelines.stable_diffusion_3.pipeline_stable_diffusion_3_inpaint.StableDiffusion3InpaintPipeline._encode_vae_image
+    def _encode_vae_image(self, image: torch.Tensor, generator: torch.Generator):
+        if isinstance(generator, list):
+            image_latents = [
+                retrieve_latents(self.vae.encode(image[i : i + 1]), generator=generator[i])
+                for i in range(image.shape[0])
+            ]
+            image_latents = torch.cat(image_latents, dim=0)
+        else:
+            image_latents = retrieve_latents(self.vae.encode(image), generator=generator)
+
+        image_latents = (image_latents - self.vae.config.shift_factor) * self.vae.config.scaling_factor
+
+        return image_latents
+
+    # Copied from diffusers.pipelines.stable_diffusion_3.pipeline_stable_diffusion_3_img2img.StableDiffusion3Img2ImgPipeline.get_timesteps
+    def get_timesteps(self, num_inference_steps, strength, device):
+        # get the original timestep using init_timestep
+        init_timestep = min(num_inference_steps * strength, num_inference_steps)
+
+        t_start = int(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
+
     def check_inputs(
         self,
         prompt,
         prompt_2,
+        strength,
         height,
         width,
         prompt_embeds=None,
@@ -508,6 +537,9 @@ class FluxFillPipeline(
         mask_image=None,
         masked_image_latents=None,
     ):
+        if strength < 0 or strength > 1:
+            raise ValueError(f"The value of strength should in [0.0, 1.0] but is {strength}")
+
         if height % (self.vae_scale_factor * 2) != 0 or width % (self.vae_scale_factor * 2) != 0:
             logger.warning(
                 f"`height` and `width` have to be divisible by {self.vae_scale_factor * 2} but are {height} and {width}. Dimensions will be resized accordingly"
@@ -625,9 +657,11 @@ class FluxFillPipeline(
         """
         self.vae.disable_tiling()
 
-    # Copied from diffusers.pipelines.flux.pipeline_flux.FluxPipeline.prepare_latents
+    # Copied from diffusers.pipelines.flux.pipeline_flux_img2img.FluxImg2ImgPipeline.prepare_latents
     def prepare_latents(
         self,
+        image,
+        timestep,
         batch_size,
         num_channels_latents,
         height,
@@ -637,28 +671,41 @@ class FluxFillPipeline(
         generator,
         latents=None,
     ):
+        if isinstance(generator, list) and len(generator) != batch_size:
+            raise ValueError(
+                f"You have passed a list of generators of length {len(generator)}, but requested an effective batch"
+                f" size of {batch_size}. Make sure the batch size matches the length of the generators."
+            )
+
         # VAE applies 8x compression on images but we must also account for packing which requires
         # latent height and width to be divisible by 2.
         height = 2 * (int(height) // (self.vae_scale_factor * 2))
         width = 2 * (int(width) // (self.vae_scale_factor * 2))
-
         shape = (batch_size, num_channels_latents, height, width)
+        latent_image_ids = self._prepare_latent_image_ids(batch_size, height // 2, width // 2, device, dtype)
 
         if latents is not None:
-            latent_image_ids = self._prepare_latent_image_ids(batch_size, height // 2, width // 2, device, dtype)
             return latents.to(device=device, dtype=dtype), latent_image_ids
 
-        if isinstance(generator, list) and len(generator) != batch_size:
+        image = image.to(device=device, dtype=dtype)
+        if image.shape[1] != self.latent_channels:
+            image_latents = self._encode_vae_image(image=image, generator=generator)
+        else:
+            image_latents = image
+        if batch_size > image_latents.shape[0] and batch_size % image_latents.shape[0] == 0:
+            # expand init_latents for batch_size
+            additional_image_per_prompt = batch_size // image_latents.shape[0]
+            image_latents = torch.cat([image_latents] * additional_image_per_prompt, dim=0)
+        elif batch_size > image_latents.shape[0] and batch_size % image_latents.shape[0] != 0:
             raise ValueError(
-                f"You have passed a list of generators of length {len(generator)}, but requested an effective batch"
-                f" size of {batch_size}. Make sure the batch size matches the length of the generators."
+                f"Cannot duplicate `image` of batch size {image_latents.shape[0]} to {batch_size} text prompts."
             )
+        else:
+            image_latents = torch.cat([image_latents], dim=0)
 
-        latents = randn_tensor(shape, generator=generator, device=device, dtype=dtype)
+        noise = randn_tensor(shape, generator=generator, device=device, dtype=dtype)
+        latents = self.scheduler.scale_noise(image_latents, timestep, noise)
         latents = self._pack_latents(latents, batch_size, num_channels_latents, height, width)
-
-        latent_image_ids = self._prepare_latent_image_ids(batch_size, height // 2, width // 2, device, dtype)
-
         return latents, latent_image_ids
 
     @property
@@ -688,6 +735,7 @@ class FluxFillPipeline(
         masked_image_latents: Optional[torch.FloatTensor] = None,
         height: Optional[int] = None,
         width: Optional[int] = None,
+        strength: float = 1.0,
         num_inference_steps: int = 50,
         sigmas: Optional[List[float]] = None,
         guidance_scale: float = 30.0,
@@ -732,6 +780,12 @@ class FluxFillPipeline(
                 The height in pixels of the generated image. This is set to 1024 by default for the best results.
             width (`int`, *optional*, defaults to self.unet.config.sample_size * self.vae_scale_factor):
                 The width in pixels of the generated image. This is set to 1024 by default for the best results.
+            strength (`float`, *optional*, defaults to 1.0):
+                Indicates extent to transform the reference `image`. Must be between 0 and 1. `image` is used as a
+                starting point and more noise is added the higher the `strength`. The number of denoising steps depends
+                on the amount of noise initially added. When `strength` is 1, added noise is maximum and the denoising
+                process runs for the full number of iterations specified in `num_inference_steps`. A value of 1
+                essentially ignores `image`.
             num_inference_steps (`int`, *optional*, defaults to 50):
                 The number of denoising steps. More denoising steps usually lead to a higher quality image at the
                 expense of slower inference.
@@ -739,7 +793,7 @@ class FluxFillPipeline(
                 Custom sigmas to use for the denoising process with schedulers which support a `sigmas` argument in
                 their `set_timesteps` method. If not defined, the default behavior when `num_inference_steps` is passed
                 will be used.
-            guidance_scale (`float`, *optional*, defaults to 7.0):
+            guidance_scale (`float`, *optional*, defaults to 30.0):
                 Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
                 `guidance_scale` is defined as `w` of equation 2. of [Imagen
                 Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
@@ -795,6 +849,7 @@ class FluxFillPipeline(
         self.check_inputs(
             prompt,
             prompt_2,
+            strength,
             height,
             width,
             prompt_embeds=prompt_embeds,
@@ -810,6 +865,9 @@ class FluxFillPipeline(
         self._joint_attention_kwargs = joint_attention_kwargs
         self._interrupt = False
 
+        init_image = self.image_processor.preprocess(image, height=height, width=width)
+        init_image = init_image.to(dtype=torch.float32)
+
         # 2. Define call parameters
         if prompt is not None and isinstance(prompt, str):
             batch_size = 1
@@ -839,9 +897,37 @@ class FluxFillPipeline(
             lora_scale=lora_scale,
         )
 
-        # 4. Prepare latent variables
+        # 4. Prepare timesteps
+        sigmas = np.linspace(1.0, 1 / num_inference_steps, num_inference_steps) if sigmas is None else sigmas
+        image_seq_len = (int(height) // self.vae_scale_factor // 2) * (int(width) // self.vae_scale_factor // 2)
+        mu = calculate_shift(
+            image_seq_len,
+            self.scheduler.config.get("base_image_seq_len", 256),
+            self.scheduler.config.get("max_image_seq_len", 4096),
+            self.scheduler.config.get("base_shift", 0.5),
+            self.scheduler.config.get("max_shift", 1.15),
+        )
+        timesteps, num_inference_steps = retrieve_timesteps(
+            self.scheduler,
+            num_inference_steps,
+            device,
+            sigmas=sigmas,
+            mu=mu,
+        )
+        timesteps, num_inference_steps = self.get_timesteps(num_inference_steps, strength, device)
+
+        if num_inference_steps < 1:
+            raise ValueError(
+                f"After adjusting the num_inference_steps by strength parameter: {strength}, the number of pipeline"
+                f"steps is {num_inference_steps} which is < 1 and not appropriate for this pipeline."
+            )
+        latent_timestep = timesteps[:1].repeat(batch_size * num_images_per_prompt)
+
+        # 5. Prepare latent variables
         num_channels_latents = self.vae.config.latent_channels
         latents, latent_image_ids = self.prepare_latents(
+            init_image,
+            latent_timestep,
             batch_size * num_images_per_prompt,
             num_channels_latents,
             height,
@@ -852,17 +938,16 @@ class FluxFillPipeline(
             latents,
         )
 
-        # 5. Prepare mask and masked image latents
+        # 6. Prepare mask and masked image latents
         if masked_image_latents is not None:
             masked_image_latents = masked_image_latents.to(latents.device)
         else:
-            image = self.image_processor.preprocess(image, height=height, width=width)
             mask_image = self.mask_processor.preprocess(mask_image, height=height, width=width)
 
-            masked_image = image * (1 - mask_image)
+            masked_image = init_image * (1 - mask_image)
             masked_image = masked_image.to(device=device, dtype=prompt_embeds.dtype)
 
-            height, width = image.shape[-2:]
+            height, width = init_image.shape[-2:]
             mask, masked_image_latents = self.prepare_mask_latents(
                 mask_image,
                 masked_image,
@@ -877,23 +962,6 @@ class FluxFillPipeline(
             )
             masked_image_latents = torch.cat((masked_image_latents, mask), dim=-1)
 
-        # 6. Prepare timesteps
-        sigmas = np.linspace(1.0, 1 / num_inference_steps, num_inference_steps) if sigmas is None else sigmas
-        image_seq_len = latents.shape[1]
-        mu = calculate_shift(
-            image_seq_len,
-            self.scheduler.config.base_image_seq_len,
-            self.scheduler.config.max_image_seq_len,
-            self.scheduler.config.base_shift,
-            self.scheduler.config.max_shift,
-        )
-        timesteps, num_inference_steps = retrieve_timesteps(
-            self.scheduler,
-            num_inference_steps,
-            device,
-            sigmas=sigmas,
-            mu=mu,
-        )
         num_warmup_steps = max(len(timesteps) - num_inference_steps * self.scheduler.order, 0)
         self._num_timesteps = len(timesteps)
 

diffusers/pipelines/flux/pipeline_flux_img2img.py

@@ -17,10 +17,17 @@ from typing import Any, Callable, Dict, List, Optional, Union
 
 import numpy as np
 import torch
-from transformers import CLIPTextModel, CLIPTokenizer, T5EncoderModel, T5TokenizerFast
+from transformers import (
+    CLIPImageProcessor,
+    CLIPTextModel,
+    CLIPTokenizer,
+    CLIPVisionModelWithProjection,
+    T5EncoderModel,
+    T5TokenizerFast,
+)
 
 from ...image_processor import PipelineImageInput, VaeImageProcessor
-from ...loaders import FluxLoraLoaderMixin, FromSingleFileMixin, TextualInversionLoaderMixin
+from ...loaders import FluxIPAdapterMixin, FluxLoraLoaderMixin, FromSingleFileMixin, TextualInversionLoaderMixin
 from ...models.autoencoders import AutoencoderKL
 from ...models.transformers import FluxTransformer2DModel
 from ...schedulers import FlowMatchEulerDiscreteScheduler
@@ -77,7 +84,7 @@ def calculate_shift(
     base_seq_len: int = 256,
     max_seq_len: int = 4096,
     base_shift: float = 0.5,
-    max_shift: float = 1.16,
+    max_shift: float = 1.15,
 ):
     m = (max_shift - base_shift) / (max_seq_len - base_seq_len)
     b = base_shift - m * base_seq_len
@@ -159,7 +166,7 @@ def retrieve_timesteps(
     return timesteps, num_inference_steps
 
 
-class FluxImg2ImgPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFileMixin):
+class FluxImg2ImgPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFileMixin, FluxIPAdapterMixin):
     r"""
     The Flux pipeline for image inpainting.
 
@@ -186,8 +193,8 @@ class FluxImg2ImgPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFile
             [T5TokenizerFast](https://huggingface.co/docs/transformers/en/model_doc/t5#transformers.T5TokenizerFast).
     """
 
-    model_cpu_offload_seq = "text_encoder->text_encoder_2->transformer->vae"
-    _optional_components = []
+    model_cpu_offload_seq = "text_encoder->text_encoder_2->image_encoder->transformer->vae"
+    _optional_components = ["image_encoder", "feature_extractor"]
     _callback_tensor_inputs = ["latents", "prompt_embeds"]
 
     def __init__(
@@ -199,6 +206,8 @@ class FluxImg2ImgPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFile
         text_encoder_2: T5EncoderModel,
         tokenizer_2: T5TokenizerFast,
         transformer: FluxTransformer2DModel,
+        image_encoder: CLIPVisionModelWithProjection = None,
+        feature_extractor: CLIPImageProcessor = None,
     ):
         super().__init__()
 
@@ -210,13 +219,16 @@ class FluxImg2ImgPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFile
             tokenizer_2=tokenizer_2,
             transformer=transformer,
             scheduler=scheduler,
+            image_encoder=image_encoder,
+            feature_extractor=feature_extractor,
         )
-        self.vae_scale_factor = (
-            2 ** (len(self.vae.config.block_out_channels) - 1) if hasattr(self, "vae") and self.vae is not None else 8
-        )
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) if getattr(self, "vae", None) else 8
         # Flux latents are turned into 2x2 patches and packed. This means the latent width and height has to be divisible
         # by the patch size. So the vae scale factor is multiplied by the patch size to account for this
-        self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor * 2)
+        self.latent_channels = self.vae.config.latent_channels if getattr(self, "vae", None) else 16
+        self.image_processor = VaeImageProcessor(
+            vae_scale_factor=self.vae_scale_factor * 2, vae_latent_channels=self.latent_channels
+        )
         self.tokenizer_max_length = (
             self.tokenizer.model_max_length if hasattr(self, "tokenizer") and self.tokenizer is not None else 77
         )
@@ -397,6 +409,55 @@ class FluxImg2ImgPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFile
 
         return prompt_embeds, pooled_prompt_embeds, text_ids
 
+    # Copied from diffusers.pipelines.flux.pipeline_flux.FluxPipeline.encode_image
+    def encode_image(self, image, device, num_images_per_prompt):
+        dtype = next(self.image_encoder.parameters()).dtype
+
+        if not isinstance(image, torch.Tensor):
+            image = self.feature_extractor(image, return_tensors="pt").pixel_values
+
+        image = image.to(device=device, dtype=dtype)
+        image_embeds = self.image_encoder(image).image_embeds
+        image_embeds = image_embeds.repeat_interleave(num_images_per_prompt, dim=0)
+        return image_embeds
+
+    # Copied from diffusers.pipelines.flux.pipeline_flux.FluxPipeline.prepare_ip_adapter_image_embeds
+    def prepare_ip_adapter_image_embeds(
+        self, ip_adapter_image, ip_adapter_image_embeds, device, num_images_per_prompt
+    ):
+        image_embeds = []
+        if ip_adapter_image_embeds is None:
+            if not isinstance(ip_adapter_image, list):
+                ip_adapter_image = [ip_adapter_image]
+
+            if len(ip_adapter_image) != self.transformer.encoder_hid_proj.num_ip_adapters:
+                raise ValueError(
+                    f"`ip_adapter_image` must have same length as the number of IP Adapters. Got {len(ip_adapter_image)} images and {self.transformer.encoder_hid_proj.num_ip_adapters} IP Adapters."
+                )
+
+            for single_ip_adapter_image in ip_adapter_image:
+                single_image_embeds = self.encode_image(single_ip_adapter_image, device, 1)
+                image_embeds.append(single_image_embeds[None, :])
+        else:
+            if not isinstance(ip_adapter_image_embeds, list):
+                ip_adapter_image_embeds = [ip_adapter_image_embeds]
+
+            if len(ip_adapter_image_embeds) != self.transformer.encoder_hid_proj.num_ip_adapters:
+                raise ValueError(
+                    f"`ip_adapter_image_embeds` must have same length as the number of IP Adapters. Got {len(ip_adapter_image_embeds)} image embeds and {self.transformer.encoder_hid_proj.num_ip_adapters} IP Adapters."
+                )
+
+            for single_image_embeds in ip_adapter_image_embeds:
+                image_embeds.append(single_image_embeds)
+
+        ip_adapter_image_embeds = []
+        for single_image_embeds in image_embeds:
+            single_image_embeds = torch.cat([single_image_embeds] * num_images_per_prompt, dim=0)
+            single_image_embeds = single_image_embeds.to(device=device)
+            ip_adapter_image_embeds.append(single_image_embeds)
+
+        return ip_adapter_image_embeds
+
     # Copied from diffusers.pipelines.stable_diffusion_3.pipeline_stable_diffusion_3_inpaint.StableDiffusion3InpaintPipeline._encode_vae_image
     def _encode_vae_image(self, image: torch.Tensor, generator: torch.Generator):
         if isinstance(generator, list):
@@ -431,8 +492,12 @@ class FluxImg2ImgPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFile
         strength,
         height,
         width,
+        negative_prompt=None,
+        negative_prompt_2=None,
         prompt_embeds=None,
+        negative_prompt_embeds=None,
         pooled_prompt_embeds=None,
+        negative_pooled_prompt_embeds=None,
         callback_on_step_end_tensor_inputs=None,
         max_sequence_length=None,
     ):
@@ -470,10 +535,33 @@ class FluxImg2ImgPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFile
         elif prompt_2 is not None and (not isinstance(prompt_2, str) and not isinstance(prompt_2, list)):
             raise ValueError(f"`prompt_2` has to be of type `str` or `list` but is {type(prompt_2)}")
 
+        if negative_prompt is not None and negative_prompt_embeds is not None:
+            raise ValueError(
+                f"Cannot forward both `negative_prompt`: {negative_prompt} and `negative_prompt_embeds`:"
+                f" {negative_prompt_embeds}. Please make sure to only forward one of the two."
+            )
+        elif negative_prompt_2 is not None and negative_prompt_embeds is not None:
+            raise ValueError(
+                f"Cannot forward both `negative_prompt_2`: {negative_prompt_2} and `negative_prompt_embeds`:"
+                f" {negative_prompt_embeds}. Please make sure to only forward one of the two."
+            )
+
+        if prompt_embeds is not None and negative_prompt_embeds is not None:
+            if prompt_embeds.shape != negative_prompt_embeds.shape:
+                raise ValueError(
+                    "`prompt_embeds` and `negative_prompt_embeds` must have the same shape when passed directly, but"
+                    f" got: `prompt_embeds` {prompt_embeds.shape} != `negative_prompt_embeds`"
+                    f" {negative_prompt_embeds.shape}."
+                )
+
         if prompt_embeds is not None and pooled_prompt_embeds is None:
             raise ValueError(
                 "If `prompt_embeds` are provided, `pooled_prompt_embeds` also have to be passed. Make sure to generate `pooled_prompt_embeds` from the same text encoder that was used to generate `prompt_embeds`."
             )
+        if negative_prompt_embeds is not None and negative_pooled_prompt_embeds is None:
+            raise ValueError(
+                "If `negative_prompt_embeds` are provided, `negative_pooled_prompt_embeds` also have to be passed. Make sure to generate `negative_pooled_prompt_embeds` from the same text encoder that was used to generate `negative_prompt_embeds`."
+            )
 
         if max_sequence_length is not None and max_sequence_length > 512:
             raise ValueError(f"`max_sequence_length` cannot be greater than 512 but is {max_sequence_length}")
@@ -549,7 +637,10 @@ class FluxImg2ImgPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFile
             return latents.to(device=device, dtype=dtype), latent_image_ids
 
         image = image.to(device=device, dtype=dtype)
-        image_latents = self._encode_vae_image(image=image, generator=generator)
+        if image.shape[1] != self.latent_channels:
+            image_latents = self._encode_vae_image(image=image, generator=generator)
+        else:
+            image_latents = image
         if batch_size > image_latents.shape[0] and batch_size % image_latents.shape[0] == 0:
             # expand init_latents for batch_size
             additional_image_per_prompt = batch_size // image_latents.shape[0]
@@ -588,6 +679,9 @@ class FluxImg2ImgPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFile
         self,
         prompt: Union[str, List[str]] = None,
         prompt_2: Optional[Union[str, List[str]]] = None,
+        negative_prompt: Union[str, List[str]] = None,
+        negative_prompt_2: Optional[Union[str, List[str]]] = None,
+        true_cfg_scale: float = 1.0,
         image: PipelineImageInput = None,
         height: Optional[int] = None,
         width: Optional[int] = None,
@@ -600,6 +694,12 @@ class FluxImg2ImgPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFile
         latents: Optional[torch.FloatTensor] = None,
         prompt_embeds: Optional[torch.FloatTensor] = None,
         pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
+        ip_adapter_image: Optional[PipelineImageInput] = None,
+        ip_adapter_image_embeds: Optional[List[torch.Tensor]] = None,
+        negative_ip_adapter_image: Optional[PipelineImageInput] = None,
+        negative_ip_adapter_image_embeds: Optional[List[torch.Tensor]] = None,
+        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        negative_pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
         joint_attention_kwargs: Optional[Dict[str, Any]] = None,
@@ -661,6 +761,17 @@ class FluxImg2ImgPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFile
             pooled_prompt_embeds (`torch.FloatTensor`, *optional*):
                 Pre-generated pooled text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting.
                 If not provided, pooled text embeddings will be generated from `prompt` input argument.
+            ip_adapter_image: (`PipelineImageInput`, *optional*): Optional image input to work with IP Adapters.
+            ip_adapter_image_embeds (`List[torch.Tensor]`, *optional*):
+                Pre-generated image embeddings for IP-Adapter. It should be a list of length same as number of
+                IP-adapters. Each element should be a tensor of shape `(batch_size, num_images, emb_dim)`. If not
+                provided, embeddings are computed from the `ip_adapter_image` input argument.
+            negative_ip_adapter_image:
+                (`PipelineImageInput`, *optional*): Optional image input to work with IP Adapters.
+            negative_ip_adapter_image_embeds (`List[torch.Tensor]`, *optional*):
+                Pre-generated image embeddings for IP-Adapter. It should be a list of length same as number of
+                IP-adapters. Each element should be a tensor of shape `(batch_size, num_images, emb_dim)`. If not
+                provided, embeddings are computed from the `ip_adapter_image` input argument.
             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`.
@@ -699,8 +810,12 @@ class FluxImg2ImgPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFile
             strength,
             height,
             width,
+            negative_prompt=negative_prompt,
+            negative_prompt_2=negative_prompt_2,
             prompt_embeds=prompt_embeds,
+            negative_prompt_embeds=negative_prompt_embeds,
             pooled_prompt_embeds=pooled_prompt_embeds,
+            negative_pooled_prompt_embeds=negative_pooled_prompt_embeds,
             callback_on_step_end_tensor_inputs=callback_on_step_end_tensor_inputs,
             max_sequence_length=max_sequence_length,
         )
@@ -726,6 +841,7 @@ class FluxImg2ImgPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFile
         lora_scale = (
             self.joint_attention_kwargs.get("scale", None) if self.joint_attention_kwargs is not None else None
         )
+        do_true_cfg = true_cfg_scale > 1 and negative_prompt is not None
         (
             prompt_embeds,
             pooled_prompt_embeds,
@@ -740,16 +856,31 @@ class FluxImg2ImgPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFile
             max_sequence_length=max_sequence_length,
             lora_scale=lora_scale,
         )
+        if do_true_cfg:
+            (
+                negative_prompt_embeds,
+                negative_pooled_prompt_embeds,
+                _,
+            ) = self.encode_prompt(
+                prompt=negative_prompt,
+                prompt_2=negative_prompt_2,
+                prompt_embeds=negative_prompt_embeds,
+                pooled_prompt_embeds=negative_pooled_prompt_embeds,
+                device=device,
+                num_images_per_prompt=num_images_per_prompt,
+                max_sequence_length=max_sequence_length,
+                lora_scale=lora_scale,
+            )
 
         # 4.Prepare timesteps
         sigmas = np.linspace(1.0, 1 / num_inference_steps, num_inference_steps) if sigmas is None else sigmas
         image_seq_len = (int(height) // self.vae_scale_factor // 2) * (int(width) // self.vae_scale_factor // 2)
         mu = calculate_shift(
             image_seq_len,
-            self.scheduler.config.base_image_seq_len,
-            self.scheduler.config.max_image_seq_len,
-            self.scheduler.config.base_shift,
-            self.scheduler.config.max_shift,
+            self.scheduler.config.get("base_image_seq_len", 256),
+            self.scheduler.config.get("max_image_seq_len", 4096),
+            self.scheduler.config.get("base_shift", 0.5),
+            self.scheduler.config.get("max_shift", 1.15),
         )
         timesteps, num_inference_steps = retrieve_timesteps(
             self.scheduler,
@@ -793,12 +924,43 @@ class FluxImg2ImgPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFile
         else:
             guidance = None
 
+        if (ip_adapter_image is not None or ip_adapter_image_embeds is not None) and (
+            negative_ip_adapter_image is None and negative_ip_adapter_image_embeds is None
+        ):
+            negative_ip_adapter_image = np.zeros((width, height, 3), dtype=np.uint8)
+        elif (ip_adapter_image is None and ip_adapter_image_embeds is None) and (
+            negative_ip_adapter_image is not None or negative_ip_adapter_image_embeds is not None
+        ):
+            ip_adapter_image = np.zeros((width, height, 3), dtype=np.uint8)
+
+        if self.joint_attention_kwargs is None:
+            self._joint_attention_kwargs = {}
+
+        image_embeds = None
+        negative_image_embeds = None
+        if ip_adapter_image is not None or ip_adapter_image_embeds is not None:
+            image_embeds = self.prepare_ip_adapter_image_embeds(
+                ip_adapter_image,
+                ip_adapter_image_embeds,
+                device,
+                batch_size * num_images_per_prompt,
+            )
+        if negative_ip_adapter_image is not None or negative_ip_adapter_image_embeds is not None:
+            negative_image_embeds = self.prepare_ip_adapter_image_embeds(
+                negative_ip_adapter_image,
+                negative_ip_adapter_image_embeds,
+                device,
+                batch_size * num_images_per_prompt,
+            )
+
         # 6. Denoising loop
         with self.progress_bar(total=num_inference_steps) as progress_bar:
             for i, t in enumerate(timesteps):
                 if self.interrupt:
                     continue
 
+                if image_embeds is not None:
+                    self._joint_attention_kwargs["ip_adapter_image_embeds"] = image_embeds
                 # broadcast to batch dimension in a way that's compatible with ONNX/Core ML
                 timestep = t.expand(latents.shape[0]).to(latents.dtype)
                 noise_pred = self.transformer(
@@ -813,6 +975,22 @@ class FluxImg2ImgPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFile
                     return_dict=False,
                 )[0]
 
+                if do_true_cfg:
+                    if negative_image_embeds is not None:
+                        self._joint_attention_kwargs["ip_adapter_image_embeds"] = negative_image_embeds
+                    neg_noise_pred = self.transformer(
+                        hidden_states=latents,
+                        timestep=timestep / 1000,
+                        guidance=guidance,
+                        pooled_projections=negative_pooled_prompt_embeds,
+                        encoder_hidden_states=negative_prompt_embeds,
+                        txt_ids=text_ids,
+                        img_ids=latent_image_ids,
+                        joint_attention_kwargs=self.joint_attention_kwargs,
+                        return_dict=False,
+                    )[0]
+                    noise_pred = neg_noise_pred + true_cfg_scale * (noise_pred - neg_noise_pred)
+
                 # compute the previous noisy sample x_t -> x_t-1
                 latents_dtype = latents.dtype
                 latents = self.scheduler.step(noise_pred, t, latents, return_dict=False)[0]