InvokeAI/invokeai/app/invocations/latent.py

# Copyright (c) 2023 Kyle Schouviller (https://github.com/kyle0654)

import math
from contextlib import ExitStack
from functools import singledispatchmethod
from typing import Any, Iterator, List, Literal, Optional, Tuple, Union

import einops
import numpy as np
import numpy.typing as npt
import torch
import torchvision.transforms as T
from diffusers import AutoencoderKL, AutoencoderTiny
from diffusers.configuration_utils import ConfigMixin
from diffusers.image_processor import VaeImageProcessor
from diffusers.models.adapter import T2IAdapter
from diffusers.models.attention_processor import (
    AttnProcessor2_0,
    LoRAAttnProcessor2_0,
    LoRAXFormersAttnProcessor,
    XFormersAttnProcessor,
)
from diffusers.models.unets.unet_2d_condition import UNet2DConditionModel
from diffusers.schedulers import DPMSolverSDEScheduler
from diffusers.schedulers import SchedulerMixin as Scheduler
from PIL import Image
from pydantic import field_validator
from torchvision.transforms.functional import resize as tv_resize

from invokeai.app.invocations.constants import LATENT_SCALE_FACTOR, SCHEDULER_NAME_VALUES
from invokeai.app.invocations.fields import (
    ConditioningField,
    DenoiseMaskField,
    FieldDescriptions,
    ImageField,
    Input,
    InputField,
    LatentsField,
    OutputField,
    UIType,
    WithBoard,
    WithMetadata,
)
from invokeai.app.invocations.ip_adapter import IPAdapterField
from invokeai.app.invocations.primitives import (
    DenoiseMaskOutput,
    ImageOutput,
    LatentsOutput,
)
from invokeai.app.invocations.t2i_adapter import T2IAdapterField
from invokeai.app.services.shared.invocation_context import InvocationContext
from invokeai.app.util.controlnet_utils import prepare_control_image
from invokeai.backend.embeddings.lora import LoRAModelRaw
from invokeai.backend.embeddings.model_patcher import ModelPatcher
from invokeai.backend.ip_adapter.ip_adapter import IPAdapter, IPAdapterPlus
from invokeai.backend.model_manager import BaseModelType, LoadedModel
from invokeai.backend.stable_diffusion import PipelineIntermediateState, set_seamless
from invokeai.backend.stable_diffusion.diffusion.conditioning_data import ConditioningData, IPAdapterConditioningInfo
from invokeai.backend.util.silence_warnings import SilenceWarnings

from ...backend.stable_diffusion.diffusers_pipeline import (
    ControlNetData,
    IPAdapterData,
    StableDiffusionGeneratorPipeline,
    T2IAdapterData,
    image_resized_to_grid_as_tensor,
)
from ...backend.stable_diffusion.diffusion.shared_invokeai_diffusion import PostprocessingSettings
from ...backend.stable_diffusion.schedulers import SCHEDULER_MAP
from ...backend.util.devices import choose_precision, choose_torch_device
from .baseinvocation import (
    BaseInvocation,
    BaseInvocationOutput,
    invocation,
    invocation_output,
)
from .controlnet_image_processors import ControlField
from .model import ModelInfo, UNetField, VaeField

if choose_torch_device() == torch.device("mps"):
    from torch import mps

DEFAULT_PRECISION = choose_precision(choose_torch_device())


@invocation_output("scheduler_output")
class SchedulerOutput(BaseInvocationOutput):
    scheduler: SCHEDULER_NAME_VALUES = OutputField(description=FieldDescriptions.scheduler, ui_type=UIType.Scheduler)


@invocation(
    "scheduler",
    title="Scheduler",
    tags=["scheduler"],
    category="latents",
    version="1.0.0",
)
class SchedulerInvocation(BaseInvocation):
    """Selects a scheduler."""

    scheduler: SCHEDULER_NAME_VALUES = InputField(
        default="euler",
        description=FieldDescriptions.scheduler,
        ui_type=UIType.Scheduler,
    )

    def invoke(self, context: InvocationContext) -> SchedulerOutput:
        return SchedulerOutput(scheduler=self.scheduler)


@invocation(
    "create_denoise_mask",
    title="Create Denoise Mask",
    tags=["mask", "denoise"],
    category="latents",
    version="1.0.1",
)
class CreateDenoiseMaskInvocation(BaseInvocation):
    """Creates mask for denoising model run."""

    vae: VaeField = InputField(description=FieldDescriptions.vae, input=Input.Connection, ui_order=0)
    image: Optional[ImageField] = InputField(default=None, description="Image which will be masked", ui_order=1)
    mask: ImageField = InputField(description="The mask to use when pasting", ui_order=2)
    tiled: bool = InputField(default=False, description=FieldDescriptions.tiled, ui_order=3)
    fp32: bool = InputField(
        default=DEFAULT_PRECISION == "float32",
        description=FieldDescriptions.fp32,
        ui_order=4,
    )

    def prep_mask_tensor(self, mask_image: Image) -> torch.Tensor:
        if mask_image.mode != "L":
            mask_image = mask_image.convert("L")
        mask_tensor: torch.Tensor = image_resized_to_grid_as_tensor(mask_image, normalize=False)
        if mask_tensor.dim() == 3:
            mask_tensor = mask_tensor.unsqueeze(0)
        # if shape is not None:
        #    mask_tensor = tv_resize(mask_tensor, shape, T.InterpolationMode.BILINEAR)
        return mask_tensor

    @torch.no_grad()
    def invoke(self, context: InvocationContext) -> DenoiseMaskOutput:
        if self.image is not None:
            image = context.images.get_pil(self.image.image_name)
            image_tensor = image_resized_to_grid_as_tensor(image.convert("RGB"))
            if image_tensor.dim() == 3:
                image_tensor = image_tensor.unsqueeze(0)
        else:
            image_tensor = None

        mask = self.prep_mask_tensor(
            context.images.get_pil(self.mask.image_name),
        )

        if image_tensor is not None:
            vae_info = context.models.load(**self.vae.vae.model_dump())

            img_mask = tv_resize(mask, image_tensor.shape[-2:], T.InterpolationMode.BILINEAR, antialias=False)
            masked_image = image_tensor * torch.where(img_mask < 0.5, 0.0, 1.0)
            # TODO:
            masked_latents = ImageToLatentsInvocation.vae_encode(vae_info, self.fp32, self.tiled, masked_image.clone())

            masked_latents_name = context.tensors.save(tensor=masked_latents)
        else:
            masked_latents_name = None

        mask_name = context.tensors.save(tensor=mask)

        return DenoiseMaskOutput.build(
            mask_name=mask_name,
            masked_latents_name=masked_latents_name,
        )


def get_scheduler(
    context: InvocationContext,
    scheduler_info: ModelInfo,
    scheduler_name: str,
    seed: int,
) -> Scheduler:
    scheduler_class, scheduler_extra_config = SCHEDULER_MAP.get(scheduler_name, SCHEDULER_MAP["ddim"])
    orig_scheduler_info = context.models.load(**scheduler_info.model_dump())
    with orig_scheduler_info as orig_scheduler:
        scheduler_config = orig_scheduler.config

    if "_backup" in scheduler_config:
        scheduler_config = scheduler_config["_backup"]
    scheduler_config = {
        **scheduler_config,
        **scheduler_extra_config,  # FIXME
        "_backup": scheduler_config,
    }

    # make dpmpp_sde reproducable(seed can be passed only in initializer)
    if scheduler_class is DPMSolverSDEScheduler:
        scheduler_config["noise_sampler_seed"] = seed

    scheduler = scheduler_class.from_config(scheduler_config)

    # hack copied over from generate.py
    if not hasattr(scheduler, "uses_inpainting_model"):
        scheduler.uses_inpainting_model = lambda: False
    assert isinstance(scheduler, Scheduler)
    return scheduler


@invocation(
    "denoise_latents",
    title="Denoise Latents",
    tags=["latents", "denoise", "txt2img", "t2i", "t2l", "img2img", "i2i", "l2l"],
    category="latents",
    version="1.5.2",
)
class DenoiseLatentsInvocation(BaseInvocation):
    """Denoises noisy latents to decodable images"""

    positive_conditioning: ConditioningField = InputField(
        description=FieldDescriptions.positive_cond, input=Input.Connection, ui_order=0
    )
    negative_conditioning: ConditioningField = InputField(
        description=FieldDescriptions.negative_cond, input=Input.Connection, ui_order=1
    )
    noise: Optional[LatentsField] = InputField(
        default=None,
        description=FieldDescriptions.noise,
        input=Input.Connection,
        ui_order=3,
    )
    steps: int = InputField(default=10, gt=0, description=FieldDescriptions.steps)
    cfg_scale: Union[float, List[float]] = InputField(
        default=7.5, ge=1, description=FieldDescriptions.cfg_scale, title="CFG Scale"
    )
    denoising_start: float = InputField(
        default=0.0,
        ge=0,
        le=1,
        description=FieldDescriptions.denoising_start,
    )
    denoising_end: float = InputField(default=1.0, ge=0, le=1, description=FieldDescriptions.denoising_end)
    scheduler: SCHEDULER_NAME_VALUES = InputField(
        default="euler",
        description=FieldDescriptions.scheduler,
        ui_type=UIType.Scheduler,
    )
    unet: UNetField = InputField(
        description=FieldDescriptions.unet,
        input=Input.Connection,
        title="UNet",
        ui_order=2,
    )
    control: Optional[Union[ControlField, list[ControlField]]] = InputField(
        default=None,
        input=Input.Connection,
        ui_order=5,
    )
    ip_adapter: Optional[Union[IPAdapterField, list[IPAdapterField]]] = InputField(
        description=FieldDescriptions.ip_adapter,
        title="IP-Adapter",
        default=None,
        input=Input.Connection,
        ui_order=6,
    )
    t2i_adapter: Optional[Union[T2IAdapterField, list[T2IAdapterField]]] = InputField(
        description=FieldDescriptions.t2i_adapter,
        title="T2I-Adapter",
        default=None,
        input=Input.Connection,
        ui_order=7,
    )
    cfg_rescale_multiplier: float = InputField(
        title="CFG Rescale Multiplier", default=0, ge=0, lt=1, description=FieldDescriptions.cfg_rescale_multiplier
    )
    latents: Optional[LatentsField] = InputField(
        default=None,
        description=FieldDescriptions.latents,
        input=Input.Connection,
        ui_order=4,
    )
    denoise_mask: Optional[DenoiseMaskField] = InputField(
        default=None,
        description=FieldDescriptions.mask,
        input=Input.Connection,
        ui_order=8,
    )

    @field_validator("cfg_scale")
    def ge_one(cls, v: Union[List[float], float]) -> Union[List[float], float]:
        """validate that all cfg_scale values are >= 1"""
        if isinstance(v, list):
            for i in v:
                if i < 1:
                    raise ValueError("cfg_scale must be greater than 1")
        else:
            if v < 1:
                raise ValueError("cfg_scale must be greater than 1")
        return v

    def get_conditioning_data(
        self,
        context: InvocationContext,
        scheduler: Scheduler,
        unet: UNet2DConditionModel,
        seed: int,
    ) -> ConditioningData:
        positive_cond_data = context.conditioning.load(self.positive_conditioning.conditioning_name)
        c = positive_cond_data.conditionings[0].to(device=unet.device, dtype=unet.dtype)
        extra_conditioning_info = c.extra_conditioning

        negative_cond_data = context.conditioning.load(self.negative_conditioning.conditioning_name)
        uc = negative_cond_data.conditionings[0].to(device=unet.device, dtype=unet.dtype)

        conditioning_data = ConditioningData(
            unconditioned_embeddings=uc,
            text_embeddings=c,
            guidance_scale=self.cfg_scale,
            guidance_rescale_multiplier=self.cfg_rescale_multiplier,
            extra=extra_conditioning_info,
            postprocessing_settings=PostprocessingSettings(
                threshold=0.0,  # threshold,
                warmup=0.2,  # warmup,
                h_symmetry_time_pct=None,  # h_symmetry_time_pct,
                v_symmetry_time_pct=None,  # v_symmetry_time_pct,
            ),
        )

        conditioning_data = conditioning_data.add_scheduler_args_if_applicable(  # FIXME
            scheduler,
            # for ddim scheduler
            eta=0.0,  # ddim_eta
            # for ancestral and sde schedulers
            # flip all bits to have noise different from initial
            generator=torch.Generator(device=unet.device).manual_seed(seed ^ 0xFFFFFFFF),
        )
        return conditioning_data

    def create_pipeline(
        self,
        unet: UNet2DConditionModel,
        scheduler: Scheduler,
    ) -> StableDiffusionGeneratorPipeline:
        # TODO:
        # configure_model_padding(
        #    unet,
        #    self.seamless,
        #    self.seamless_axes,
        # )

        class FakeVae:
            class FakeVaeConfig:
                def __init__(self) -> None:
                    self.block_out_channels = [0]

            def __init__(self) -> None:
                self.config = FakeVae.FakeVaeConfig()

        return StableDiffusionGeneratorPipeline(
            vae=FakeVae(),  # TODO: oh...
            text_encoder=None,
            tokenizer=None,
            unet=unet,
            scheduler=scheduler,
            safety_checker=None,
            feature_extractor=None,
            requires_safety_checker=False,
        )

    def prep_control_data(
        self,
        context: InvocationContext,
        control_input: Optional[Union[ControlField, List[ControlField]]],
        latents_shape: List[int],
        exit_stack: ExitStack,
        do_classifier_free_guidance: bool = True,
    ) -> Optional[List[ControlNetData]]:
        # Assuming fixed dimensional scaling of LATENT_SCALE_FACTOR.
        control_height_resize = latents_shape[2] * LATENT_SCALE_FACTOR
        control_width_resize = latents_shape[3] * LATENT_SCALE_FACTOR
        if control_input is None:
            control_list = None
        elif isinstance(control_input, list) and len(control_input) == 0:
            control_list = None
        elif isinstance(control_input, ControlField):
            control_list = [control_input]
        elif isinstance(control_input, list) and len(control_input) > 0 and isinstance(control_input[0], ControlField):
            control_list = control_input
        else:
            control_list = None
        if control_list is None:
            return None
        # After above handling, any control that is not None should now be of type list[ControlField].

        # FIXME: add checks to skip entry if model or image is None
        #        and if weight is None, populate with default 1.0?
        controlnet_data = []
        for control_info in control_list:
            control_model = exit_stack.enter_context(context.models.load(key=control_info.control_model.key))

            # control_models.append(control_model)
            control_image_field = control_info.image
            input_image = context.images.get_pil(control_image_field.image_name)
            # self.image.image_type, self.image.image_name
            # FIXME: still need to test with different widths, heights, devices, dtypes
            #        and add in batch_size, num_images_per_prompt?
            #        and do real check for classifier_free_guidance?
            # prepare_control_image should return torch.Tensor of shape(batch_size, 3, height, width)
            control_image = prepare_control_image(
                image=input_image,
                do_classifier_free_guidance=do_classifier_free_guidance,
                width=control_width_resize,
                height=control_height_resize,
                # batch_size=batch_size * num_images_per_prompt,
                # num_images_per_prompt=num_images_per_prompt,
                device=control_model.device,
                dtype=control_model.dtype,
                control_mode=control_info.control_mode,
                resize_mode=control_info.resize_mode,
            )
            control_item = ControlNetData(
                model=control_model,  # model object
                image_tensor=control_image,
                weight=control_info.control_weight,
                begin_step_percent=control_info.begin_step_percent,
                end_step_percent=control_info.end_step_percent,
                control_mode=control_info.control_mode,
                # any resizing needed should currently be happening in prepare_control_image(),
                #    but adding resize_mode to ControlNetData in case needed in the future
                resize_mode=control_info.resize_mode,
            )
            controlnet_data.append(control_item)
            # MultiControlNetModel has been refactored out, just need list[ControlNetData]

        return controlnet_data

    def prep_ip_adapter_data(
        self,
        context: InvocationContext,
        ip_adapter: Optional[Union[IPAdapterField, list[IPAdapterField]]],
        conditioning_data: ConditioningData,
        exit_stack: ExitStack,
    ) -> Optional[list[IPAdapterData]]:
        """If IP-Adapter is enabled, then this function loads the requisite models, and adds the image prompt embeddings
        to the `conditioning_data` (in-place).
        """
        if ip_adapter is None:
            return None

        # ip_adapter could be a list or a single IPAdapterField. Normalize to a list here.
        if not isinstance(ip_adapter, list):
            ip_adapter = [ip_adapter]

        if len(ip_adapter) == 0:
            return None

        ip_adapter_data_list = []
        conditioning_data.ip_adapter_conditioning = []
        for single_ip_adapter in ip_adapter:
            ip_adapter_model: Union[IPAdapter, IPAdapterPlus] = exit_stack.enter_context(
                context.models.load(key=single_ip_adapter.ip_adapter_model.key)
            )

            image_encoder_model_info = context.models.load(key=single_ip_adapter.image_encoder_model.key)

            # `single_ip_adapter.image` could be a list or a single ImageField. Normalize to a list here.
            single_ipa_image_fields = single_ip_adapter.image
            if not isinstance(single_ipa_image_fields, list):
                single_ipa_image_fields = [single_ipa_image_fields]

            single_ipa_images = [context.images.get_pil(image.image_name) for image in single_ipa_image_fields]

            # TODO(ryand): With some effort, the step of running the CLIP Vision encoder could be done before any other
            # models are needed in memory. This would help to reduce peak memory utilization in low-memory environments.
            with image_encoder_model_info as image_encoder_model:
                # Get image embeddings from CLIP and ImageProjModel.
                image_prompt_embeds, uncond_image_prompt_embeds = ip_adapter_model.get_image_embeds(
                    single_ipa_images, image_encoder_model
                )

                conditioning_data.ip_adapter_conditioning.append(
                    IPAdapterConditioningInfo(image_prompt_embeds, uncond_image_prompt_embeds)
                )

            ip_adapter_data_list.append(
                IPAdapterData(
                    ip_adapter_model=ip_adapter_model,
                    weight=single_ip_adapter.weight,
                    begin_step_percent=single_ip_adapter.begin_step_percent,
                    end_step_percent=single_ip_adapter.end_step_percent,
                )
            )

        return ip_adapter_data_list

    def run_t2i_adapters(
        self,
        context: InvocationContext,
        t2i_adapter: Optional[Union[T2IAdapterField, list[T2IAdapterField]]],
        latents_shape: list[int],
        do_classifier_free_guidance: bool,
    ) -> Optional[list[T2IAdapterData]]:
        if t2i_adapter is None:
            return None

        # Handle the possibility that t2i_adapter could be a list or a single T2IAdapterField.
        if isinstance(t2i_adapter, T2IAdapterField):
            t2i_adapter = [t2i_adapter]

        if len(t2i_adapter) == 0:
            return None

        t2i_adapter_data = []
        for t2i_adapter_field in t2i_adapter:
            t2i_adapter_model_config = context.models.get_config(key=t2i_adapter_field.t2i_adapter_model.key)
            t2i_adapter_loaded_model = context.models.load(key=t2i_adapter_field.t2i_adapter_model.key)
            image = context.images.get_pil(t2i_adapter_field.image.image_name)

            # The max_unet_downscale is the maximum amount that the UNet model downscales the latent image internally.
            if t2i_adapter_model_config.base == BaseModelType.StableDiffusion1:
                max_unet_downscale = 8
            elif t2i_adapter_model_config.base == BaseModelType.StableDiffusionXL:
                max_unet_downscale = 4
            else:
                raise ValueError(f"Unexpected T2I-Adapter base model type: '{t2i_adapter_model_config.base}'.")

            t2i_adapter_model: T2IAdapter
            with t2i_adapter_loaded_model as t2i_adapter_model:
                total_downscale_factor = t2i_adapter_model.total_downscale_factor

                # Resize the T2I-Adapter input image.
                # We select the resize dimensions so that after the T2I-Adapter's total_downscale_factor is applied, the
                # result will match the latent image's dimensions after max_unet_downscale is applied.
                t2i_input_height = latents_shape[2] // max_unet_downscale * total_downscale_factor
                t2i_input_width = latents_shape[3] // max_unet_downscale * total_downscale_factor

                # Note: We have hard-coded `do_classifier_free_guidance=False`. This is because we only want to prepare
                # a single image. If CFG is enabled, we will duplicate the resultant tensor after applying the
                # T2I-Adapter model.
                #
                # Note: We re-use the `prepare_control_image(...)` from ControlNet for T2I-Adapter, because it has many
                # of the same requirements (e.g. preserving binary masks during resize).
                t2i_image = prepare_control_image(
                    image=image,
                    do_classifier_free_guidance=False,
                    width=t2i_input_width,
                    height=t2i_input_height,
                    num_channels=t2i_adapter_model.config["in_channels"],  # mypy treats this as a FrozenDict
                    device=t2i_adapter_model.device,
                    dtype=t2i_adapter_model.dtype,
                    resize_mode=t2i_adapter_field.resize_mode,
                )

                adapter_state = t2i_adapter_model(t2i_image)

            if do_classifier_free_guidance:
                for idx, value in enumerate(adapter_state):
                    adapter_state[idx] = torch.cat([value] * 2, dim=0)

            t2i_adapter_data.append(
                T2IAdapterData(
                    adapter_state=adapter_state,
                    weight=t2i_adapter_field.weight,
                    begin_step_percent=t2i_adapter_field.begin_step_percent,
                    end_step_percent=t2i_adapter_field.end_step_percent,
                )
            )

        return t2i_adapter_data

    # original idea by https://github.com/AmericanPresidentJimmyCarter
    # TODO: research more for second order schedulers timesteps
    def init_scheduler(
        self,
        scheduler: Union[Scheduler, ConfigMixin],
        device: torch.device,
        steps: int,
        denoising_start: float,
        denoising_end: float,
    ) -> Tuple[int, List[int], int]:
        assert isinstance(scheduler, ConfigMixin)
        if scheduler.config.get("cpu_only", False):
            scheduler.set_timesteps(steps, device="cpu")
            timesteps = scheduler.timesteps.to(device=device)
        else:
            scheduler.set_timesteps(steps, device=device)
            timesteps = scheduler.timesteps

        # skip greater order timesteps
        _timesteps = timesteps[:: scheduler.order]

        # get start timestep index
        t_start_val = int(round(scheduler.config["num_train_timesteps"] * (1 - denoising_start)))
        t_start_idx = len(list(filter(lambda ts: ts >= t_start_val, _timesteps)))

        # get end timestep index
        t_end_val = int(round(scheduler.config["num_train_timesteps"] * (1 - denoising_end)))
        t_end_idx = len(list(filter(lambda ts: ts >= t_end_val, _timesteps[t_start_idx:])))

        # apply order to indexes
        t_start_idx *= scheduler.order
        t_end_idx *= scheduler.order

        init_timestep = timesteps[t_start_idx : t_start_idx + 1]
        timesteps = timesteps[t_start_idx : t_start_idx + t_end_idx]
        num_inference_steps = len(timesteps) // scheduler.order

        return num_inference_steps, timesteps, init_timestep

    def prep_inpaint_mask(
        self, context: InvocationContext, latents: torch.Tensor
    ) -> Tuple[Optional[torch.Tensor], Optional[torch.Tensor]]:
        if self.denoise_mask is None:
            return None, None

        mask = context.tensors.load(self.denoise_mask.mask_name)
        mask = tv_resize(mask, latents.shape[-2:], T.InterpolationMode.BILINEAR, antialias=False)
        if self.denoise_mask.masked_latents_name is not None:
            masked_latents = context.tensors.load(self.denoise_mask.masked_latents_name)
        else:
            masked_latents = None

        return 1 - mask, masked_latents

    @torch.no_grad()
    def invoke(self, context: InvocationContext) -> LatentsOutput:
        with SilenceWarnings():  # this quenches NSFW nag from diffusers
            seed = None
            noise = None
            if self.noise is not None:
                noise = context.tensors.load(self.noise.latents_name)
                seed = self.noise.seed

            if self.latents is not None:
                latents = context.tensors.load(self.latents.latents_name)
                if seed is None:
                    seed = self.latents.seed

                if noise is not None and noise.shape[1:] != latents.shape[1:]:
                    raise Exception(f"Incompatable 'noise' and 'latents' shapes: {latents.shape=} {noise.shape=}")

            elif noise is not None:
                latents = torch.zeros_like(noise)
            else:
                raise Exception("'latents' or 'noise' must be provided!")

            if seed is None:
                seed = 0

            mask, masked_latents = self.prep_inpaint_mask(context, latents)

            # TODO(ryand): I have hard-coded `do_classifier_free_guidance=True` to mirror the behaviour of ControlNets,
            # below. Investigate whether this is appropriate.
            t2i_adapter_data = self.run_t2i_adapters(
                context,
                self.t2i_adapter,
                latents.shape,
                do_classifier_free_guidance=True,
            )

            # get the unet's config so that we can pass the base to dispatch_progress()
            unet_config = context.models.get_config(self.unet.unet.key)

            def step_callback(state: PipelineIntermediateState) -> None:
                context.util.sd_step_callback(state, unet_config.base)

            def _lora_loader() -> Iterator[Tuple[LoRAModelRaw, float]]:
                for lora in self.unet.loras:
                    lora_info = context.models.load(**lora.model_dump(exclude={"weight"}))
                    yield (lora_info.model, lora.weight)
                    del lora_info
                return

            unet_info = context.models.load(**self.unet.unet.model_dump())
            assert isinstance(unet_info.model, UNet2DConditionModel)
            with (
                ExitStack() as exit_stack,
                ModelPatcher.apply_freeu(unet_info.model, self.unet.freeu_config),
                set_seamless(unet_info.model, self.unet.seamless_axes),  # FIXME
                unet_info as unet,
                # Apply the LoRA after unet has been moved to its target device for faster patching.
                ModelPatcher.apply_lora_unet(unet, _lora_loader()),
            ):
                assert isinstance(unet, UNet2DConditionModel)
                latents = latents.to(device=unet.device, dtype=unet.dtype)
                if noise is not None:
                    noise = noise.to(device=unet.device, dtype=unet.dtype)
                if mask is not None:
                    mask = mask.to(device=unet.device, dtype=unet.dtype)
                if masked_latents is not None:
                    masked_latents = masked_latents.to(device=unet.device, dtype=unet.dtype)

                scheduler = get_scheduler(
                    context=context,
                    scheduler_info=self.unet.scheduler,
                    scheduler_name=self.scheduler,
                    seed=seed,
                )

                pipeline = self.create_pipeline(unet, scheduler)
                conditioning_data = self.get_conditioning_data(context, scheduler, unet, seed)

                controlnet_data = self.prep_control_data(
                    context=context,
                    control_input=self.control,
                    latents_shape=latents.shape,
                    # do_classifier_free_guidance=(self.cfg_scale >= 1.0))
                    do_classifier_free_guidance=True,
                    exit_stack=exit_stack,
                )

                ip_adapter_data = self.prep_ip_adapter_data(
                    context=context,
                    ip_adapter=self.ip_adapter,
                    conditioning_data=conditioning_data,
                    exit_stack=exit_stack,
                )

                num_inference_steps, timesteps, init_timestep = self.init_scheduler(
                    scheduler,
                    device=unet.device,
                    steps=self.steps,
                    denoising_start=self.denoising_start,
                    denoising_end=self.denoising_end,
                )

                (
                    result_latents,
                    result_attention_map_saver,
                ) = pipeline.latents_from_embeddings(
                    latents=latents,
                    timesteps=timesteps,
                    init_timestep=init_timestep,
                    noise=noise,
                    seed=seed,
                    mask=mask,
                    masked_latents=masked_latents,
                    num_inference_steps=num_inference_steps,
                    conditioning_data=conditioning_data,
                    control_data=controlnet_data,
                    ip_adapter_data=ip_adapter_data,
                    t2i_adapter_data=t2i_adapter_data,
                    callback=step_callback,
                )

            # https://discuss.huggingface.co/t/memory-usage-by-later-pipeline-stages/23699
            result_latents = result_latents.to("cpu")
            torch.cuda.empty_cache()
            if choose_torch_device() == torch.device("mps"):
                mps.empty_cache()

            name = context.tensors.save(tensor=result_latents)
        return LatentsOutput.build(latents_name=name, latents=result_latents, seed=seed)


@invocation(
    "l2i",
    title="Latents to Image",
    tags=["latents", "image", "vae", "l2i"],
    category="latents",
    version="1.2.1",
)
class LatentsToImageInvocation(BaseInvocation, WithMetadata, WithBoard):
    """Generates an image from latents."""

    latents: LatentsField = InputField(
        description=FieldDescriptions.latents,
        input=Input.Connection,
    )
    vae: VaeField = InputField(
        description=FieldDescriptions.vae,
        input=Input.Connection,
    )
    tiled: bool = InputField(default=False, description=FieldDescriptions.tiled)
    fp32: bool = InputField(default=DEFAULT_PRECISION == "float32", description=FieldDescriptions.fp32)

    @torch.no_grad()
    def invoke(self, context: InvocationContext) -> ImageOutput:
        latents = context.tensors.load(self.latents.latents_name)

        vae_info = context.models.load(**self.vae.vae.model_dump())

        with set_seamless(vae_info.model, self.vae.seamless_axes), vae_info as vae:
            assert isinstance(vae, torch.nn.Module)
            latents = latents.to(vae.device)
            if self.fp32:
                vae.to(dtype=torch.float32)

                use_torch_2_0_or_xformers = isinstance(
                    vae.decoder.mid_block.attentions[0].processor,
                    (
                        AttnProcessor2_0,
                        XFormersAttnProcessor,
                        LoRAXFormersAttnProcessor,
                        LoRAAttnProcessor2_0,
                    ),
                )
                # if xformers or torch_2_0 is used attention block does not need
                # to be in float32 which can save lots of memory
                if use_torch_2_0_or_xformers:
                    vae.post_quant_conv.to(latents.dtype)
                    vae.decoder.conv_in.to(latents.dtype)
                    vae.decoder.mid_block.to(latents.dtype)
                else:
                    latents = latents.float()

            else:
                vae.to(dtype=torch.float16)
                latents = latents.half()

            if self.tiled or context.config.get().tiled_decode:
                vae.enable_tiling()
            else:
                vae.disable_tiling()

            # clear memory as vae decode can request a lot
            torch.cuda.empty_cache()
            if choose_torch_device() == torch.device("mps"):
                mps.empty_cache()

            with torch.inference_mode():
                # copied from diffusers pipeline
                latents = latents / vae.config.scaling_factor
                image = vae.decode(latents, return_dict=False)[0]
                image = (image / 2 + 0.5).clamp(0, 1)  # denormalize
                # we always cast to float32 as this does not cause significant overhead and is compatible with bfloat16
                np_image = image.cpu().permute(0, 2, 3, 1).float().numpy()

                image = VaeImageProcessor.numpy_to_pil(np_image)[0]

        torch.cuda.empty_cache()
        if choose_torch_device() == torch.device("mps"):
            mps.empty_cache()

        image_dto = context.images.save(image=image)

        return ImageOutput.build(image_dto)


LATENTS_INTERPOLATION_MODE = Literal["nearest", "linear", "bilinear", "bicubic", "trilinear", "area", "nearest-exact"]


@invocation(
    "lresize",
    title="Resize Latents",
    tags=["latents", "resize"],
    category="latents",
    version="1.0.1",
)
class ResizeLatentsInvocation(BaseInvocation):
    """Resizes latents to explicit width/height (in pixels). Provided dimensions are floor-divided by 8."""

    latents: LatentsField = InputField(
        description=FieldDescriptions.latents,
        input=Input.Connection,
    )
    width: int = InputField(
        ge=64,
        multiple_of=LATENT_SCALE_FACTOR,
        description=FieldDescriptions.width,
    )
    height: int = InputField(
        ge=64,
        multiple_of=LATENT_SCALE_FACTOR,
        description=FieldDescriptions.width,
    )
    mode: LATENTS_INTERPOLATION_MODE = InputField(default="bilinear", description=FieldDescriptions.interp_mode)
    antialias: bool = InputField(default=False, description=FieldDescriptions.torch_antialias)

    def invoke(self, context: InvocationContext) -> LatentsOutput:
        latents = context.tensors.load(self.latents.latents_name)

        # TODO:
        device = choose_torch_device()

        resized_latents = torch.nn.functional.interpolate(
            latents.to(device),
            size=(self.height // LATENT_SCALE_FACTOR, self.width // LATENT_SCALE_FACTOR),
            mode=self.mode,
            antialias=self.antialias if self.mode in ["bilinear", "bicubic"] else False,
        )

        # https://discuss.huggingface.co/t/memory-usage-by-later-pipeline-stages/23699
        resized_latents = resized_latents.to("cpu")
        torch.cuda.empty_cache()
        if device == torch.device("mps"):
            mps.empty_cache()

        name = context.tensors.save(tensor=resized_latents)
        return LatentsOutput.build(latents_name=name, latents=resized_latents, seed=self.latents.seed)


@invocation(
    "lscale",
    title="Scale Latents",
    tags=["latents", "resize"],
    category="latents",
    version="1.0.1",
)
class ScaleLatentsInvocation(BaseInvocation):
    """Scales latents by a given factor."""

    latents: LatentsField = InputField(
        description=FieldDescriptions.latents,
        input=Input.Connection,
    )
    scale_factor: float = InputField(gt=0, description=FieldDescriptions.scale_factor)
    mode: LATENTS_INTERPOLATION_MODE = InputField(default="bilinear", description=FieldDescriptions.interp_mode)
    antialias: bool = InputField(default=False, description=FieldDescriptions.torch_antialias)

    def invoke(self, context: InvocationContext) -> LatentsOutput:
        latents = context.tensors.load(self.latents.latents_name)

        # TODO:
        device = choose_torch_device()

        # resizing
        resized_latents = torch.nn.functional.interpolate(
            latents.to(device),
            scale_factor=self.scale_factor,
            mode=self.mode,
            antialias=self.antialias if self.mode in ["bilinear", "bicubic"] else False,
        )

        # https://discuss.huggingface.co/t/memory-usage-by-later-pipeline-stages/23699
        resized_latents = resized_latents.to("cpu")
        torch.cuda.empty_cache()
        if device == torch.device("mps"):
            mps.empty_cache()

        name = context.tensors.save(tensor=resized_latents)
        return LatentsOutput.build(latents_name=name, latents=resized_latents, seed=self.latents.seed)


@invocation(
    "i2l",
    title="Image to Latents",
    tags=["latents", "image", "vae", "i2l"],
    category="latents",
    version="1.0.1",
)
class ImageToLatentsInvocation(BaseInvocation):
    """Encodes an image into latents."""

    image: ImageField = InputField(
        description="The image to encode",
    )
    vae: VaeField = InputField(
        description=FieldDescriptions.vae,
        input=Input.Connection,
    )
    tiled: bool = InputField(default=False, description=FieldDescriptions.tiled)
    fp32: bool = InputField(default=DEFAULT_PRECISION == "float32", description=FieldDescriptions.fp32)

    @staticmethod
    def vae_encode(vae_info: LoadedModel, upcast: bool, tiled: bool, image_tensor: torch.Tensor) -> torch.Tensor:
        with vae_info as vae:
            assert isinstance(vae, torch.nn.Module)
            orig_dtype = vae.dtype
            if upcast:
                vae.to(dtype=torch.float32)

                use_torch_2_0_or_xformers = isinstance(
                    vae.decoder.mid_block.attentions[0].processor,
                    (
                        AttnProcessor2_0,
                        XFormersAttnProcessor,
                        LoRAXFormersAttnProcessor,
                        LoRAAttnProcessor2_0,
                    ),
                )
                # if xformers or torch_2_0 is used attention block does not need
                # to be in float32 which can save lots of memory
                if use_torch_2_0_or_xformers:
                    vae.post_quant_conv.to(orig_dtype)
                    vae.decoder.conv_in.to(orig_dtype)
                    vae.decoder.mid_block.to(orig_dtype)
                # else:
                #    latents = latents.float()

            else:
                vae.to(dtype=torch.float16)
                # latents = latents.half()

            if tiled:
                vae.enable_tiling()
            else:
                vae.disable_tiling()

            # non_noised_latents_from_image
            image_tensor = image_tensor.to(device=vae.device, dtype=vae.dtype)
            with torch.inference_mode():
                latents = ImageToLatentsInvocation._encode_to_tensor(vae, image_tensor)

            latents = vae.config.scaling_factor * latents
            latents = latents.to(dtype=orig_dtype)

        return latents

    @torch.no_grad()
    def invoke(self, context: InvocationContext) -> LatentsOutput:
        image = context.images.get_pil(self.image.image_name)

        vae_info = context.models.load(**self.vae.vae.model_dump())

        image_tensor = image_resized_to_grid_as_tensor(image.convert("RGB"))
        if image_tensor.dim() == 3:
            image_tensor = einops.rearrange(image_tensor, "c h w -> 1 c h w")

        latents = self.vae_encode(vae_info, self.fp32, self.tiled, image_tensor)

        latents = latents.to("cpu")
        name = context.tensors.save(tensor=latents)
        return LatentsOutput.build(latents_name=name, latents=latents, seed=None)

    @singledispatchmethod
    @staticmethod
    def _encode_to_tensor(vae: AutoencoderKL, image_tensor: torch.FloatTensor) -> torch.FloatTensor:
        assert isinstance(vae, torch.nn.Module)
        image_tensor_dist = vae.encode(image_tensor).latent_dist
        latents: torch.Tensor = image_tensor_dist.sample().to(
            dtype=vae.dtype
        )  # FIXME: uses torch.randn. make reproducible!
        return latents

    @_encode_to_tensor.register
    @staticmethod
    def _(vae: AutoencoderTiny, image_tensor: torch.FloatTensor) -> torch.FloatTensor:
        assert isinstance(vae, torch.nn.Module)
        latents: torch.FloatTensor = vae.encode(image_tensor).latents
        return latents


@invocation(
    "lblend",
    title="Blend Latents",
    tags=["latents", "blend"],
    category="latents",
    version="1.0.1",
)
class BlendLatentsInvocation(BaseInvocation):
    """Blend two latents using a given alpha. Latents must have same size."""

    latents_a: LatentsField = InputField(
        description=FieldDescriptions.latents,
        input=Input.Connection,
    )
    latents_b: LatentsField = InputField(
        description=FieldDescriptions.latents,
        input=Input.Connection,
    )
    alpha: float = InputField(default=0.5, description=FieldDescriptions.blend_alpha)

    def invoke(self, context: InvocationContext) -> LatentsOutput:
        latents_a = context.tensors.load(self.latents_a.latents_name)
        latents_b = context.tensors.load(self.latents_b.latents_name)

        if latents_a.shape != latents_b.shape:
            raise Exception("Latents to blend must be the same size.")

        # TODO:
        device = choose_torch_device()

        def slerp(
            t: Union[float, npt.NDArray[Any]],  # FIXME: maybe use np.float32 here?
            v0: Union[torch.Tensor, npt.NDArray[Any]],
            v1: Union[torch.Tensor, npt.NDArray[Any]],
            DOT_THRESHOLD: float = 0.9995,
        ) -> Union[torch.Tensor, npt.NDArray[Any]]:
            """
            Spherical linear interpolation
            Args:
                t (float/np.ndarray): Float value between 0.0 and 1.0
                v0 (np.ndarray): Starting vector
                v1 (np.ndarray): Final vector
                DOT_THRESHOLD (float): Threshold for considering the two vectors as
                                    colineal. Not recommended to alter this.
            Returns:
                v2 (np.ndarray): Interpolation vector between v0 and v1
            """
            inputs_are_torch = False
            if not isinstance(v0, np.ndarray):
                inputs_are_torch = True
                v0 = v0.detach().cpu().numpy()
            if not isinstance(v1, np.ndarray):
                inputs_are_torch = True
                v1 = v1.detach().cpu().numpy()

            dot = np.sum(v0 * v1 / (np.linalg.norm(v0) * np.linalg.norm(v1)))
            if np.abs(dot) > DOT_THRESHOLD:
                v2 = (1 - t) * v0 + t * v1
            else:
                theta_0 = np.arccos(dot)
                sin_theta_0 = np.sin(theta_0)
                theta_t = theta_0 * t
                sin_theta_t = np.sin(theta_t)
                s0 = np.sin(theta_0 - theta_t) / sin_theta_0
                s1 = sin_theta_t / sin_theta_0
                v2 = s0 * v0 + s1 * v1

            if inputs_are_torch:
                v2_torch: torch.Tensor = torch.from_numpy(v2).to(device)
                return v2_torch
            else:
                assert isinstance(v2, np.ndarray)
                return v2

        # blend
        bl = slerp(self.alpha, latents_a, latents_b)
        assert isinstance(bl, torch.Tensor)
        blended_latents: torch.Tensor = bl  # for type checking convenience

        # https://discuss.huggingface.co/t/memory-usage-by-later-pipeline-stages/23699
        blended_latents = blended_latents.to("cpu")
        torch.cuda.empty_cache()
        if device == torch.device("mps"):
            mps.empty_cache()

        name = context.tensors.save(tensor=blended_latents)
        return LatentsOutput.build(latents_name=name, latents=blended_latents)


# The Crop Latents node was copied from @skunkworxdark's implementation here:
# https://github.com/skunkworxdark/XYGrid_nodes/blob/74647fa9c1fa57d317a94bd43ca689af7f0aae5e/images_to_grids.py#L1117C1-L1167C80
@invocation(
    "crop_latents",
    title="Crop Latents",
    tags=["latents", "crop"],
    category="latents",
    version="1.0.1",
)
# TODO(ryand): Named `CropLatentsCoreInvocation` to prevent a conflict with custom node `CropLatentsInvocation`.
# Currently, if the class names conflict then 'GET /openapi.json' fails.
class CropLatentsCoreInvocation(BaseInvocation):
    """Crops a latent-space tensor to a box specified in image-space. The box dimensions and coordinates must be
    divisible by the latent scale factor of 8.
    """

    latents: LatentsField = InputField(
        description=FieldDescriptions.latents,
        input=Input.Connection,
    )
    x: int = InputField(
        ge=0,
        multiple_of=LATENT_SCALE_FACTOR,
        description="The left x coordinate (in px) of the crop rectangle in image space. This value will be converted to a dimension in latent space.",
    )
    y: int = InputField(
        ge=0,
        multiple_of=LATENT_SCALE_FACTOR,
        description="The top y coordinate (in px) of the crop rectangle in image space. This value will be converted to a dimension in latent space.",
    )
    width: int = InputField(
        ge=1,
        multiple_of=LATENT_SCALE_FACTOR,
        description="The width (in px) of the crop rectangle in image space. This value will be converted to a dimension in latent space.",
    )
    height: int = InputField(
        ge=1,
        multiple_of=LATENT_SCALE_FACTOR,
        description="The height (in px) of the crop rectangle in image space. This value will be converted to a dimension in latent space.",
    )

    def invoke(self, context: InvocationContext) -> LatentsOutput:
        latents = context.tensors.load(self.latents.latents_name)

        x1 = self.x // LATENT_SCALE_FACTOR
        y1 = self.y // LATENT_SCALE_FACTOR
        x2 = x1 + (self.width // LATENT_SCALE_FACTOR)
        y2 = y1 + (self.height // LATENT_SCALE_FACTOR)

        cropped_latents = latents[..., y1:y2, x1:x2]

        name = context.tensors.save(tensor=cropped_latents)

        return LatentsOutput.build(latents_name=name, latents=cropped_latents)


@invocation_output("ideal_size_output")
class IdealSizeOutput(BaseInvocationOutput):
    """Base class for invocations that output an image"""

    width: int = OutputField(description="The ideal width of the image (in pixels)")
    height: int = OutputField(description="The ideal height of the image (in pixels)")


@invocation(
    "ideal_size",
    title="Ideal Size",
    tags=["latents", "math", "ideal_size"],
    version="1.0.2",
)
class IdealSizeInvocation(BaseInvocation):
    """Calculates the ideal size for generation to avoid duplication"""

    width: int = InputField(default=1024, description="Final image width")
    height: int = InputField(default=576, description="Final image height")
    unet: UNetField = InputField(default=None, description=FieldDescriptions.unet)
    multiplier: float = InputField(
        default=1.0,
        description="Amount to multiply the model's dimensions by when calculating the ideal size (may result in initial generation artifacts if too large)",
    )

    def trim_to_multiple_of(self, *args: int, multiple_of: int = LATENT_SCALE_FACTOR) -> Tuple[int, ...]:
        return tuple((x - x % multiple_of) for x in args)

    def invoke(self, context: InvocationContext) -> IdealSizeOutput:
        unet_config = context.models.get_config(**self.unet.unet.model_dump())
        aspect = self.width / self.height
        dimension: float = 512
        if unet_config.base == BaseModelType.StableDiffusion2:
            dimension = 768
        elif unet_config.base == BaseModelType.StableDiffusionXL:
            dimension = 1024
        dimension = dimension * self.multiplier
        min_dimension = math.floor(dimension * 0.5)
        model_area = dimension * dimension  # hardcoded for now since all models are trained on square images

        if aspect > 1.0:
            init_height = max(min_dimension, math.sqrt(model_area / aspect))
            init_width = init_height * aspect
        else:
            init_width = max(min_dimension, math.sqrt(model_area * aspect))
            init_height = init_width / aspect

        scaled_width, scaled_height = self.trim_to_multiple_of(
            math.floor(init_width),
            math.floor(init_height),
        )

        return IdealSizeOutput(width=scaled_width, height=scaled_height)