mirror of https://github.com/vladmandic/automatic
613 lines
32 KiB
Python
613 lines
32 KiB
Python
# Copyright (c) 2025 Bytedance Ltd. and/or its affiliates.
|
|
# Copyright (c) 2024 Black Forest Labs, The HuggingFace Team and The InstantX Team. All rights reserved.
|
|
#
|
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
# you may not use this file except in compliance with the License.
|
|
# You may obtain a copy of the License at
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# Unless required by applicable law or agreed to in writing, software
|
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
# See the License for the specific language governing permissions and
|
|
# limitations under the License.
|
|
|
|
import inspect
|
|
from typing import Any, Callable, Dict, List, Optional, Union
|
|
|
|
import numpy as np
|
|
import torch
|
|
from diffusers import FluxControlNetPipeline
|
|
from diffusers.models.controlnet_flux import FluxControlNetModel, FluxMultiControlNetModel
|
|
from diffusers.image_processor import PipelineImageInput
|
|
from diffusers.pipelines.flux.pipeline_output import FluxPipelineOutput
|
|
from diffusers.utils import is_torch_xla_available, logging
|
|
|
|
|
|
if is_torch_xla_available():
|
|
import torch_xla.core.xla_model as xm
|
|
|
|
XLA_AVAILABLE = True
|
|
else:
|
|
XLA_AVAILABLE = False
|
|
|
|
logger = logging.get_logger(__name__) # pylint: disable=invalid-name
|
|
|
|
|
|
# Copied from diffusers.pipelines.flux.pipeline_flux.calculate_shift
|
|
def calculate_shift(
|
|
image_seq_len,
|
|
base_seq_len: int = 256,
|
|
max_seq_len: int = 4096,
|
|
base_shift: float = 0.5,
|
|
max_shift: float = 1.16,
|
|
):
|
|
m = (max_shift - base_shift) / (max_seq_len - base_seq_len)
|
|
b = base_shift - m * base_seq_len
|
|
mu = image_seq_len * m + b
|
|
return mu
|
|
|
|
|
|
# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps
|
|
def retrieve_timesteps(
|
|
scheduler,
|
|
num_inference_steps: Optional[int] = None,
|
|
device: Optional[Union[str, torch.device]] = None,
|
|
timesteps: Optional[List[int]] = None,
|
|
sigmas: Optional[List[float]] = None,
|
|
**kwargs,
|
|
):
|
|
r"""
|
|
Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles
|
|
custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`.
|
|
|
|
Args:
|
|
scheduler (`SchedulerMixin`):
|
|
The scheduler to get timesteps from.
|
|
num_inference_steps (`int`):
|
|
The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps`
|
|
must be `None`.
|
|
device (`str` or `torch.device`, *optional*):
|
|
The device to which the timesteps should be moved to. If `None`, the timesteps are not moved.
|
|
timesteps (`List[int]`, *optional*):
|
|
Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed,
|
|
`num_inference_steps` and `sigmas` must be `None`.
|
|
sigmas (`List[float]`, *optional*):
|
|
Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed,
|
|
`num_inference_steps` and `timesteps` must be `None`.
|
|
|
|
Returns:
|
|
`Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the
|
|
second element is the number of inference steps.
|
|
"""
|
|
if timesteps is not None and sigmas is not None:
|
|
raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values")
|
|
if timesteps is not None:
|
|
accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys())
|
|
if not accepts_timesteps:
|
|
raise ValueError(
|
|
f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom"
|
|
f" timestep schedules. Please check whether you are using the correct scheduler."
|
|
)
|
|
scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs)
|
|
timesteps = scheduler.timesteps
|
|
num_inference_steps = len(timesteps)
|
|
elif sigmas is not None:
|
|
accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys())
|
|
if not accept_sigmas:
|
|
raise ValueError(
|
|
f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom"
|
|
f" sigmas schedules. Please check whether you are using the correct scheduler."
|
|
)
|
|
scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs)
|
|
timesteps = scheduler.timesteps
|
|
num_inference_steps = len(timesteps)
|
|
else:
|
|
scheduler.set_timesteps(num_inference_steps, device=device, **kwargs)
|
|
timesteps = scheduler.timesteps
|
|
return timesteps, num_inference_steps
|
|
|
|
|
|
class FluxInfuseNetPipeline(FluxControlNetPipeline):
|
|
@torch.no_grad()
|
|
def __call__(
|
|
self,
|
|
prompt: Union[str, List[str]] | None = None,
|
|
prompt_2: Optional[Union[str, List[str]]] = None,
|
|
height: Optional[int] = None,
|
|
width: Optional[int] = None,
|
|
num_inference_steps: int = 28,
|
|
timesteps: List[int] | None = None,
|
|
guidance_scale: float = 3.5,
|
|
id_image: PipelineImageInput = None,
|
|
controlnet_guidance_scale: float = 1.0,
|
|
control_guidance_start: Union[float, List[float]] = 0.0,
|
|
control_guidance_end: Union[float, List[float]] = 1.0,
|
|
control_image: PipelineImageInput = None,
|
|
control_mode: Optional[Union[int, List[int]]] = None,
|
|
controlnet_conditioning_scale: Union[float, List[float]] = 1.0,
|
|
num_images_per_prompt: Optional[int] = 1,
|
|
generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
|
|
latents: Optional[torch.FloatTensor] = None,
|
|
prompt_embeds: Optional[torch.FloatTensor] = None,
|
|
pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
|
|
output_type: Optional[str] = "pil",
|
|
return_dict: bool = True,
|
|
joint_attention_kwargs: Optional[Dict[str, Any]] = None,
|
|
callback_on_step_end: Optional[Callable[[int, int, Dict], None]] = None,
|
|
callback_on_step_end_tensor_inputs: List[str] = ["latents"],
|
|
max_sequence_length: int = 512,
|
|
|
|
# ID-specific parameters
|
|
controlnet_prompt_embeds: Optional[torch.FloatTensor] = None,
|
|
|
|
# True CFG parameters
|
|
true_guidance_scale: float = 1.0,
|
|
negative_prompt: Optional[Union[str, List[str]]] = None,
|
|
negative_prompt_2: Optional[Union[str, List[str]]] = None,
|
|
negative_prompt_embeds: Optional[torch.FloatTensor] = None,
|
|
negative_pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
|
|
):
|
|
r"""
|
|
Function invoked when calling the pipeline for generation.
|
|
|
|
Args:
|
|
prompt (`str` or `List[str]`, *optional*):
|
|
The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`.
|
|
instead.
|
|
prompt_2 (`str` or `List[str]`, *optional*):
|
|
The prompt or prompts to be sent to `tokenizer_2` and `text_encoder_2`. If not defined, `prompt` is
|
|
will be used instead
|
|
height (`int`, *optional*, defaults to self.unet.config.sample_size * self.vae_scale_factor):
|
|
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.
|
|
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.
|
|
timesteps (`List[int]`, *optional*):
|
|
Custom timesteps to use for the denoising process with schedulers which support a `timesteps` argument
|
|
in their `set_timesteps` method. If not defined, the default behavior when `num_inference_steps` is
|
|
passed will be used. Must be in descending order.
|
|
guidance_scale (`float`, *optional*, defaults to 7.0):
|
|
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 >
|
|
1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
|
|
usually at the expense of lower image quality.
|
|
controlnet_guidance_scale (`float`, *optional*, defaults to 7.0):
|
|
Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
|
|
`controlnet_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 >
|
|
1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
|
|
usually at the expense of lower image quality.
|
|
control_guidance_start (`float` or `List[float]`, *optional*, defaults to 0.0):
|
|
The percentage of total steps at which the ControlNet starts applying.
|
|
control_guidance_end (`float` or `List[float]`, *optional*, defaults to 1.0):
|
|
The percentage of total steps at which the ControlNet stops applying.
|
|
control_image (`torch.Tensor`, `PIL.Image.Image`, `np.ndarray`, `List[torch.Tensor]`, `List[PIL.Image.Image]`, `List[np.ndarray]`,:
|
|
`List[List[torch.Tensor]]`, `List[List[np.ndarray]]` or `List[List[PIL.Image.Image]]`):
|
|
The ControlNet input condition to provide guidance to the `unet` for generation. If the type is
|
|
specified as `torch.Tensor`, it is passed to ControlNet as is. `PIL.Image.Image` can also be accepted
|
|
as an image. The dimensions of the output image defaults to `image`'s dimensions. If height and/or
|
|
width are passed, `image` is resized accordingly. If multiple ControlNets are specified in `init`,
|
|
images must be passed as a list such that each element of the list can be correctly batched for input
|
|
to a single ControlNet.
|
|
controlnet_conditioning_scale (`float` or `List[float]`, *optional*, defaults to 1.0):
|
|
The outputs of the ControlNet are multiplied by `controlnet_conditioning_scale` before they are added
|
|
to the residual in the original `unet`. If multiple ControlNets are specified in `init`, you can set
|
|
the corresponding scale as a list.
|
|
control_mode (`int` or `List[int]`,, *optional*, defaults to None):
|
|
The control mode when applying ControlNet-Union.
|
|
num_images_per_prompt (`int`, *optional*, defaults to 1):
|
|
The number of images to generate per prompt.
|
|
generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
|
|
One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
|
|
to make generation deterministic.
|
|
latents (`torch.FloatTensor`, *optional*):
|
|
Pre-generated noisy latents, sampled from a Gaussian distribution, to be used as inputs for image
|
|
generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
|
|
tensor will ge generated by sampling using the supplied random `generator`.
|
|
prompt_embeds (`torch.FloatTensor`, *optional*):
|
|
Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
|
|
provided, text embeddings will be generated from `prompt` input argument.
|
|
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.
|
|
output_type (`str`, *optional*, defaults to `"pil"`):
|
|
The output format of the generate image. Choose between
|
|
[PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`.
|
|
return_dict (`bool`, *optional*, defaults to `True`):
|
|
Whether or not to return a [`~pipelines.flux.FluxPipelineOutput`] instead of a plain tuple.
|
|
joint_attention_kwargs (`dict`, *optional*):
|
|
A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
|
|
`self.processor` in
|
|
[diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
|
|
callback_on_step_end (`Callable`, *optional*):
|
|
A function that calls at the end of each denoising steps during the inference. The function is called
|
|
with the following arguments: `callback_on_step_end(self: DiffusionPipeline, step: int, timestep: int,
|
|
callback_kwargs: Dict)`. `callback_kwargs` will include a list of all tensors as specified by
|
|
`callback_on_step_end_tensor_inputs`.
|
|
callback_on_step_end_tensor_inputs (`List`, *optional*):
|
|
The list of tensor inputs for the `callback_on_step_end` function. The tensors specified in the list
|
|
will be passed as `callback_kwargs` argument. You will only be able to include variables listed in the
|
|
`._callback_tensor_inputs` attribute of your pipeline class.
|
|
max_sequence_length (`int` defaults to 512): Maximum sequence length to use with the `prompt`.
|
|
controlnet_prompt_embeds (`torch.FloatTensor`, *optional*):
|
|
Pre-generated embeddings for the InfuseNet. Can be used to easily tweak inputs, *e.g.* image embeddings.
|
|
If not provided, embeddings will be generated from `prompt` or `prompt_embeds` input arguments.
|
|
true_guidance_scale (`float`, *optional*, defaults to 1.0):
|
|
True CFG scale as defined in [Classifier-Free Diffusion Guidance]((https://arxiv.org/abs/2207.12598).
|
|
negative_prompt (`str` or `List[str]`, *optional*):
|
|
The negative prompt or negative prompts to guide the image generation. If not defined, one has to pass
|
|
`negative_prompt_embeds`. instead.
|
|
negative_prompt_2 (`str` or `List[str]`, *optional*):
|
|
The negative prompt or negative prompts to be sent to `tokenizer_2` and `text_encoder_2`. If not defined,
|
|
`negative_prompt` is will be used instead.
|
|
negative_prompt_embeds (`torch.FloatTensor`, *optional*):
|
|
Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
|
|
weighting. If not provided, negative text embeddings will be generated from `negative_prompt` input
|
|
argument.
|
|
negative_pooled_prompt_embeds (`torch.FloatTensor`, *optional*):
|
|
Pre-generated negative pooled text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
|
|
weighting. If not provided, negative pooled text embeddings will be generated from
|
|
`negative_prompt` input argument.
|
|
|
|
Examples:
|
|
|
|
Returns:
|
|
[`~pipelines.flux.FluxPipelineOutput`] or `tuple`: [`~pipelines.flux.FluxPipelineOutput`] if `return_dict`
|
|
is True, otherwise a `tuple`. When returning a tuple, the first element is a list with the generated
|
|
images.
|
|
"""
|
|
|
|
height = height or self.default_sample_size * self.vae_scale_factor
|
|
width = width or self.default_sample_size * self.vae_scale_factor
|
|
|
|
if not isinstance(control_guidance_start, list) and isinstance(control_guidance_end, list):
|
|
control_guidance_start = len(control_guidance_end) * [control_guidance_start]
|
|
elif not isinstance(control_guidance_end, list) and isinstance(control_guidance_start, list):
|
|
control_guidance_end = len(control_guidance_start) * [control_guidance_end]
|
|
elif not isinstance(control_guidance_start, list) and not isinstance(control_guidance_end, list):
|
|
mult = len(self.controlnet.nets) if isinstance(self.controlnet, FluxMultiControlNetModel) else 1
|
|
control_guidance_start, control_guidance_end = (
|
|
mult * [control_guidance_start],
|
|
mult * [control_guidance_end],
|
|
)
|
|
|
|
# 1. Check inputs. Raise error if not correct
|
|
self.check_inputs(
|
|
prompt,
|
|
prompt_2,
|
|
height,
|
|
width,
|
|
prompt_embeds=prompt_embeds,
|
|
pooled_prompt_embeds=pooled_prompt_embeds,
|
|
callback_on_step_end_tensor_inputs=callback_on_step_end_tensor_inputs,
|
|
max_sequence_length=max_sequence_length,
|
|
)
|
|
|
|
self._guidance_scale = guidance_scale
|
|
self._controlnet_guidance_scale = controlnet_guidance_scale
|
|
self._true_guidance_scale = true_guidance_scale
|
|
self._joint_attention_kwargs = joint_attention_kwargs
|
|
self._interrupt = False
|
|
|
|
# 2. Define call parameters
|
|
if prompt is not None and isinstance(prompt, str):
|
|
batch_size = 1
|
|
elif prompt is not None and isinstance(prompt, list):
|
|
batch_size = len(prompt)
|
|
else:
|
|
batch_size = prompt_embeds.shape[0]
|
|
|
|
device = self._execution_device
|
|
dtype = self.transformer.dtype
|
|
|
|
lora_scale = (
|
|
self.joint_attention_kwargs.get("scale", None) if self.joint_attention_kwargs is not None else None
|
|
)
|
|
(
|
|
prompt_embeds,
|
|
pooled_prompt_embeds,
|
|
text_ids,
|
|
) = self.encode_prompt(
|
|
prompt=prompt,
|
|
prompt_2=prompt_2,
|
|
prompt_embeds=prompt_embeds,
|
|
pooled_prompt_embeds=pooled_prompt_embeds,
|
|
device=device,
|
|
num_images_per_prompt=num_images_per_prompt,
|
|
max_sequence_length=max_sequence_length,
|
|
lora_scale=lora_scale,
|
|
)
|
|
if negative_prompt is not None or (negative_prompt_embeds is not None and negative_pooled_prompt_embeds is not None):
|
|
(
|
|
negative_prompt_embeds,
|
|
negative_pooled_prompt_embeds,
|
|
negative_text_ids,
|
|
) = 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,
|
|
)
|
|
|
|
if controlnet_prompt_embeds is None:
|
|
controlnet_prompt_embeds = prompt_embeds
|
|
(
|
|
controlnet_prompt_embeds,
|
|
pooled_prompt_embeds,
|
|
controlnet_text_ids,
|
|
) = self.encode_prompt(
|
|
prompt=prompt,
|
|
prompt_2=prompt_2,
|
|
prompt_embeds=controlnet_prompt_embeds,
|
|
pooled_prompt_embeds=pooled_prompt_embeds,
|
|
device=device,
|
|
num_images_per_prompt=num_images_per_prompt,
|
|
max_sequence_length=max_sequence_length,
|
|
lora_scale=lora_scale,
|
|
)
|
|
|
|
# 3. Prepare control image
|
|
num_channels_latents = self.transformer.config.in_channels // 4
|
|
if isinstance(self.controlnet, FluxControlNetModel) or True:
|
|
control_image = self.prepare_image(
|
|
image=control_image,
|
|
width=width,
|
|
height=height,
|
|
batch_size=batch_size * num_images_per_prompt,
|
|
num_images_per_prompt=num_images_per_prompt,
|
|
device=device,
|
|
dtype=self.vae.dtype,
|
|
)
|
|
height, width = control_image.shape[-2:]
|
|
|
|
# xlab controlnet has a input_hint_block and instantx controlnet does not
|
|
controlnet_blocks_repeat = False if self.controlnet.input_hint_block is None else True
|
|
if self.controlnet.input_hint_block is None:
|
|
# vae encode
|
|
control_image = self.vae.encode(control_image).latent_dist.sample()
|
|
control_image = (control_image - self.vae.config.shift_factor) * self.vae.config.scaling_factor
|
|
|
|
# pack
|
|
height_control_image, width_control_image = control_image.shape[2:]
|
|
control_image = self._pack_latents(
|
|
control_image,
|
|
batch_size * num_images_per_prompt,
|
|
num_channels_latents,
|
|
height_control_image,
|
|
width_control_image,
|
|
)
|
|
|
|
# Here we ensure that `control_mode` has the same length as the control_image.
|
|
if control_mode is not None:
|
|
if not isinstance(control_mode, int):
|
|
raise ValueError(" For `FluxControlNet`, `control_mode` should be an `int` or `None`")
|
|
control_mode = torch.tensor(control_mode).to(device, dtype=torch.long)
|
|
control_mode = control_mode.view(-1, 1).expand(control_image.shape[0], 1)
|
|
|
|
elif isinstance(self.controlnet, FluxMultiControlNetModel):
|
|
control_images = []
|
|
# xlab controlnet has a input_hint_block and instantx controlnet does not
|
|
controlnet_blocks_repeat = False if self.controlnet.nets[0].input_hint_block is None else True
|
|
for _i, control_image_ in enumerate(control_image):
|
|
control_image_ = self.prepare_image(
|
|
image=control_image_,
|
|
width=width,
|
|
height=height,
|
|
batch_size=batch_size * num_images_per_prompt,
|
|
num_images_per_prompt=num_images_per_prompt,
|
|
device=device,
|
|
dtype=self.vae.dtype,
|
|
)
|
|
height, width = control_image_.shape[-2:]
|
|
|
|
if self.controlnet.nets[0].input_hint_block is None:
|
|
# vae encode
|
|
control_image_ = self.vae.encode(control_image_).latent_dist.sample()
|
|
control_image_ = (control_image_ - self.vae.config.shift_factor) * self.vae.config.scaling_factor
|
|
|
|
# pack
|
|
height_control_image, width_control_image = control_image_.shape[2:]
|
|
control_image_ = self._pack_latents(
|
|
control_image_,
|
|
batch_size * num_images_per_prompt,
|
|
num_channels_latents,
|
|
height_control_image,
|
|
width_control_image,
|
|
)
|
|
control_images.append(control_image_)
|
|
|
|
control_image = control_images
|
|
|
|
# Here we ensure that `control_mode` has the same length as the control_image.
|
|
if isinstance(control_mode, list) and len(control_mode) != len(control_image):
|
|
raise ValueError("For Multi-ControlNet, `control_mode` must be a list of the same length as the number of controlnets (control images) specified")
|
|
if not isinstance(control_mode, list):
|
|
control_mode = [control_mode] * len(control_image)
|
|
# set control mode
|
|
control_modes = []
|
|
for cmode in control_mode:
|
|
if cmode is None:
|
|
cmode = -1
|
|
control_mode = torch.tensor(cmode).expand(control_images[0].shape[0]).to(device, dtype=torch.long)
|
|
control_modes.append(control_mode)
|
|
control_mode = control_modes
|
|
|
|
# 4. Prepare latent variables
|
|
num_channels_latents = self.transformer.config.in_channels // 4
|
|
latents, latent_image_ids = self.prepare_latents(
|
|
batch_size * num_images_per_prompt,
|
|
num_channels_latents,
|
|
height,
|
|
width,
|
|
prompt_embeds.dtype,
|
|
device,
|
|
generator,
|
|
latents,
|
|
)
|
|
|
|
# 5. Prepare timesteps
|
|
sigmas = np.linspace(1.0, 1 / num_inference_steps, num_inference_steps)
|
|
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,
|
|
timesteps,
|
|
sigmas,
|
|
mu=mu,
|
|
)
|
|
|
|
num_warmup_steps = max(len(timesteps) - num_inference_steps * self.scheduler.order, 0)
|
|
self._num_timesteps = len(timesteps)
|
|
|
|
# 6. Create tensor stating which controlnets to keep
|
|
controlnet_keep = []
|
|
for i in range(len(timesteps)):
|
|
keeps = [
|
|
1.0 - float(i / len(timesteps) < s or (i + 1) / len(timesteps) > e)
|
|
for s, e in zip(control_guidance_start, control_guidance_end)
|
|
]
|
|
controlnet_keep.append(keeps[0] if isinstance(self.controlnet, FluxControlNetModel) else keeps)
|
|
|
|
# 7. Denoising loop
|
|
with self.progress_bar(total=num_inference_steps) as progress_bar:
|
|
for i, t in enumerate(timesteps):
|
|
if self.interrupt:
|
|
continue
|
|
|
|
# broadcast to batch dimension in a way that's compatible with ONNX/Core ML
|
|
timestep = t.expand(latents.shape[0]).to(latents.dtype)
|
|
|
|
if isinstance(self.controlnet, FluxMultiControlNetModel):
|
|
use_guidance = self.controlnet.nets[0].config.guidance_embeds
|
|
else:
|
|
use_guidance = self.controlnet.config.guidance_embeds
|
|
|
|
guidance = torch.tensor([controlnet_guidance_scale], device=device) if use_guidance else None
|
|
guidance = guidance.expand(latents.shape[0]) if guidance is not None else None
|
|
|
|
if isinstance(controlnet_keep[i], list):
|
|
if not isinstance(controlnet_conditioning_scale, list):
|
|
controlnet_conditioning_scale = len(controlnet_keep) * [controlnet_conditioning_scale]
|
|
cond_scale = [c * s for c, s in zip(controlnet_conditioning_scale, controlnet_keep[i])]
|
|
controlnet_conditioning_scale = controlnet_conditioning_scale[0]
|
|
else:
|
|
controlnet_cond_scale = controlnet_conditioning_scale
|
|
if isinstance(controlnet_cond_scale, list):
|
|
controlnet_cond_scale = controlnet_cond_scale[0]
|
|
cond_scale = controlnet_cond_scale * controlnet_keep[i]
|
|
|
|
# controlnet
|
|
controlnet_block_samples, controlnet_single_block_samples = self.controlnet(
|
|
hidden_states=latents,
|
|
controlnet_cond=control_image,
|
|
controlnet_mode=control_mode,
|
|
conditioning_scale=cond_scale[0],
|
|
timestep=timestep / 1000,
|
|
guidance=guidance,
|
|
pooled_projections=pooled_prompt_embeds,
|
|
encoder_hidden_states=controlnet_prompt_embeds,
|
|
txt_ids=controlnet_text_ids,
|
|
img_ids=latent_image_ids,
|
|
joint_attention_kwargs=self.joint_attention_kwargs,
|
|
return_dict=False,
|
|
)
|
|
|
|
guidance = (
|
|
torch.tensor([guidance_scale], device=device) if self.transformer.config.guidance_embeds else None
|
|
)
|
|
guidance = guidance.expand(latents.shape[0]) if guidance is not None else None
|
|
|
|
noise_pred = self.transformer(
|
|
hidden_states=latents,
|
|
timestep=timestep / 1000,
|
|
guidance=guidance,
|
|
pooled_projections=pooled_prompt_embeds,
|
|
encoder_hidden_states=prompt_embeds,
|
|
controlnet_block_samples=controlnet_block_samples,
|
|
controlnet_single_block_samples=controlnet_single_block_samples,
|
|
txt_ids=text_ids,
|
|
img_ids=latent_image_ids,
|
|
joint_attention_kwargs=self.joint_attention_kwargs,
|
|
return_dict=False,
|
|
controlnet_blocks_repeat=controlnet_blocks_repeat,
|
|
)[0]
|
|
|
|
# perform true CFG
|
|
if negative_prompt_embeds is not None and negative_pooled_prompt_embeds is not None and negative_text_ids is not None:
|
|
noise_pred_uncond = self.transformer(
|
|
hidden_states=latents,
|
|
timestep=timestep / 1000,
|
|
guidance=guidance,
|
|
pooled_projections=negative_pooled_prompt_embeds,
|
|
encoder_hidden_states=negative_prompt_embeds,
|
|
controlnet_block_samples=None,
|
|
controlnet_single_block_samples=None,
|
|
txt_ids=negative_text_ids,
|
|
img_ids=latent_image_ids,
|
|
joint_attention_kwargs=self.joint_attention_kwargs,
|
|
return_dict=False,
|
|
controlnet_blocks_repeat=controlnet_blocks_repeat,
|
|
)[0]
|
|
|
|
noise_pred = noise_pred_uncond + true_guidance_scale * (noise_pred - noise_pred_uncond)
|
|
|
|
# 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]
|
|
|
|
if latents.dtype != latents_dtype:
|
|
if torch.backends.mps.is_available():
|
|
# some platforms (eg. apple mps) misbehave due to a pytorch bug: https://github.com/pytorch/pytorch/pull/99272
|
|
latents = latents.to(latents_dtype)
|
|
|
|
if callback_on_step_end is not None:
|
|
callback_kwargs = {}
|
|
for k in callback_on_step_end_tensor_inputs:
|
|
callback_kwargs[k] = locals()[k]
|
|
callback_outputs = callback_on_step_end(self, i, t, callback_kwargs)
|
|
|
|
latents = callback_outputs.pop("latents", latents)
|
|
prompt_embeds = callback_outputs.pop("prompt_embeds", prompt_embeds)
|
|
|
|
# call the callback, if provided
|
|
if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
|
|
progress_bar.update()
|
|
|
|
if XLA_AVAILABLE:
|
|
xm.mark_step()
|
|
|
|
if output_type == "latent":
|
|
image = latents
|
|
|
|
else:
|
|
latents = self._unpack_latents(latents, height, width, self.vae_scale_factor)
|
|
latents = (latents / self.vae.config.scaling_factor) + self.vae.config.shift_factor
|
|
|
|
image = self.vae.decode(latents, return_dict=False)[0]
|
|
image = self.image_processor.postprocess(image, output_type=output_type)
|
|
|
|
# Offload all models
|
|
self.maybe_free_model_hooks()
|
|
|
|
if not return_dict:
|
|
return (image,)
|
|
|
|
return FluxPipelineOutput(images=image)
|