Mirrored_Repos/InvokeAI
1
0
Fork 0
mirror of https://github.com/invoke-ai/InvokeAI synced 2024-08-30 20:32:17 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'main' into lstein/new-model-manager

Browse Source
This commit is contained in:
StAlKeR7779
2023-06-13 23:37:52 +03:00
committed by GitHub
parent e7db6d8120 46cac6468e
commit c9ae26a176
39 changed files with 784 additions and 181 deletions
Download Patch File Download Diff File Expand all files Collapse all files

51
invokeai/app/invocations/latent.py
View File

@ -3,8 +3,6 @@
from contextlib import ExitStack
from typing import List, Literal, Optional, Union
import einops
from pydantic import BaseModel, Field, validator
import torch
from diffusers import ControlNetModel
@ -173,23 +171,36 @@ class TextToLatentsInvocation(BaseInvocation):
negative_conditioning: Optional[ConditioningField] = Field(description="Negative conditioning for generation")
noise: Optional[LatentsField] = Field(description="The noise to use")
steps: int = Field(default=10, gt=0, description="The number of steps to use to generate the image")
cfg_scale: float = Field(default=7.5, ge=1, description="The Classifier-Free Guidance, higher values may result in a result closer to the prompt", )
cfg_scale: Union[float, List[float]] = Field(default=7.5, ge=1, description="The Classifier-Free Guidance, higher values may result in a result closer to the prompt", )
scheduler: SAMPLER_NAME_VALUES = Field(default="euler", description="The scheduler to use" )
seamless: bool = Field(default=False, description="Whether or not to generate an image that can tile without seams", )
seamless_axes: str = Field(default="", description="The axes to tile the image on, 'x' and/or 'y'")
unet: UNetField = Field(default=None, description="UNet submodel")
control: Union[ControlField, list[ControlField]] = Field(default=None, description="The control to use")
#seamless: bool = Field(default=False, description="Whether or not to generate an image that can tile without seams", )
#seamless_axes: str = Field(default="", description="The axes to tile the image on, 'x' and/or 'y'")
# fmt: on
@validator("cfg_scale")
def ge_one(cls, v):
"""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
# Schema customisation
class Config(InvocationConfig):
schema_extra = {
"ui": {
"tags": ["latents", "image"],
"tags": ["latents"],
"type_hints": {
"model": "model",
"control": "control",
# "cfg_scale": "float",
"cfg_scale": "number"
}
},
}
@ -210,10 +221,10 @@ class TextToLatentsInvocation(BaseInvocation):
uc, _ = context.services.latents.get(self.negative_conditioning.conditioning_name)
conditioning_data = ConditioningData(
uc,
c,
self.cfg_scale,
extra_conditioning_info,
unconditioned_embeddings=uc,
text_embeddings=c,
guidance_scale=self.cfg_scale,
extra=extra_conditioning_info,
postprocessing_settings=PostprocessingSettings(
threshold=0.0,#threshold,
warmup=0.2,#warmup,
@ -351,10 +362,10 @@ class TextToLatentsInvocation(BaseInvocation):
loras = [(stack.enter_context(context.services.model_manager.get_model(**lora.dict(exclude={"weight"}))), lora.weight) for lora in self.unet.loras]
print("type of control input: ", type(self.control))
control_data = self.prep_control_data(model=pipeline, context=context, control_input=self.control,
latents_shape=noise.shape,
do_classifier_free_guidance=(self.cfg_scale >= 1.0))
control_data = self.prep_control_data(model=model, context=context, control_input=self.control,
latents_shape=noise.shape,
# do_classifier_free_guidance=(self.cfg_scale >= 1.0))
do_classifier_free_guidance=True,)
with ModelPatcher.apply_lora_unet(pipeline.unet, loras):
# TODO: Verify the noise is the right size
@ -364,7 +375,7 @@ class TextToLatentsInvocation(BaseInvocation):
num_inference_steps=self.steps,
conditioning_data=conditioning_data,
control_data=control_data, # list[ControlNetData]
callback=step_callback
callback=step_callback,
)
# https://discuss.huggingface.co/t/memory-usage-by-later-pipeline-stages/23699
@ -391,6 +402,7 @@ class LatentsToLatentsInvocation(TextToLatentsInvocation):
"type_hints": {
"model": "model",
"control": "control",
"cfg_scale": "number",
}
},
}
@ -421,6 +433,12 @@ class LatentsToLatentsInvocation(TextToLatentsInvocation):
pipeline = self.create_pipeline(unet, scheduler)
conditioning_data = self.get_conditioning_data(context, scheduler)
control_data = self.prep_control_data(model=model, context=context, control_input=self.control,
latents_shape=noise.shape,
# do_classifier_free_guidance=(self.cfg_scale >= 1.0))
do_classifier_free_guidance=True,
)
# TODO: Verify the noise is the right size
initial_latents = latent if self.strength < 1.0 else torch.zeros_like(
@ -442,6 +460,7 @@ class LatentsToLatentsInvocation(TextToLatentsInvocation):
noise=noise,
num_inference_steps=self.steps,
conditioning_data=conditioning_data,
control_data=control_data, # list[ControlNetData]
callback=step_callback
)