2022-12-25 03:39:00 +08:00
|
|
|
import os
|
|
|
|
|
|
|
|
import torch
|
|
|
|
from torch import nn
|
2023-07-13 22:24:54 +08:00
|
|
|
from modules import devices, paths, shared
|
2022-12-25 03:39:00 +08:00
|
|
|
|
2023-07-13 22:24:54 +08:00
|
|
|
sd_vae_approx_models = {}
|
2022-12-25 03:39:00 +08:00
|
|
|
|
|
|
|
|
|
|
|
class VAEApprox(nn.Module):
|
|
|
|
def __init__(self):
|
|
|
|
super(VAEApprox, self).__init__()
|
|
|
|
self.conv1 = nn.Conv2d(4, 8, (7, 7))
|
|
|
|
self.conv2 = nn.Conv2d(8, 16, (5, 5))
|
|
|
|
self.conv3 = nn.Conv2d(16, 32, (3, 3))
|
|
|
|
self.conv4 = nn.Conv2d(32, 64, (3, 3))
|
|
|
|
self.conv5 = nn.Conv2d(64, 32, (3, 3))
|
|
|
|
self.conv6 = nn.Conv2d(32, 16, (3, 3))
|
|
|
|
self.conv7 = nn.Conv2d(16, 8, (3, 3))
|
|
|
|
self.conv8 = nn.Conv2d(8, 3, (3, 3))
|
|
|
|
|
|
|
|
def forward(self, x):
|
|
|
|
extra = 11
|
|
|
|
x = nn.functional.interpolate(x, (x.shape[2] * 2, x.shape[3] * 2))
|
|
|
|
x = nn.functional.pad(x, (extra, extra, extra, extra))
|
|
|
|
|
|
|
|
for layer in [self.conv1, self.conv2, self.conv3, self.conv4, self.conv5, self.conv6, self.conv7, self.conv8, ]:
|
|
|
|
x = layer(x)
|
|
|
|
x = nn.functional.leaky_relu(x, 0.1)
|
|
|
|
|
|
|
|
return x
|
|
|
|
|
|
|
|
|
2023-07-13 22:24:54 +08:00
|
|
|
def download_model(model_path, model_url):
|
|
|
|
if not os.path.exists(model_path):
|
|
|
|
os.makedirs(os.path.dirname(model_path), exist_ok=True)
|
|
|
|
|
|
|
|
print(f'Downloading VAEApprox model to: {model_path}')
|
|
|
|
torch.hub.download_url_to_file(model_url, model_path)
|
|
|
|
|
|
|
|
|
2022-12-25 03:39:00 +08:00
|
|
|
def model():
|
2023-07-13 22:24:54 +08:00
|
|
|
model_name = "vaeapprox-sdxl.pt" if getattr(shared.sd_model, 'is_sdxl', False) else "model.pt"
|
|
|
|
loaded_model = sd_vae_approx_models.get(model_name)
|
2022-12-25 03:39:00 +08:00
|
|
|
|
2023-07-13 22:24:54 +08:00
|
|
|
if loaded_model is None:
|
|
|
|
model_path = os.path.join(paths.models_path, "VAE-approx", model_name)
|
2023-03-11 12:22:59 +08:00
|
|
|
if not os.path.exists(model_path):
|
2023-07-13 22:24:54 +08:00
|
|
|
model_path = os.path.join(paths.script_path, "models", "VAE-approx", model_name)
|
|
|
|
|
|
|
|
if not os.path.exists(model_path):
|
|
|
|
model_path = os.path.join(paths.models_path, "VAE-approx", model_name)
|
|
|
|
download_model(model_path, 'https://github.com/AUTOMATIC1111/stable-diffusion-webui/releases/download/v1.0.0-pre/' + model_name)
|
|
|
|
|
|
|
|
loaded_model = VAEApprox()
|
|
|
|
loaded_model.load_state_dict(torch.load(model_path, map_location='cpu' if devices.device.type != 'cuda' else None))
|
|
|
|
loaded_model.eval()
|
|
|
|
loaded_model.to(devices.device, devices.dtype)
|
|
|
|
sd_vae_approx_models[model_name] = loaded_model
|
2022-12-25 03:39:00 +08:00
|
|
|
|
2023-07-13 22:24:54 +08:00
|
|
|
return loaded_model
|
2022-12-25 03:39:00 +08:00
|
|
|
|
|
|
|
|
|
|
|
def cheap_approximation(sample):
|
|
|
|
# https://discuss.huggingface.co/t/decoding-latents-to-rgb-without-upscaling/23204/2
|
|
|
|
|
2023-07-15 01:27:41 +08:00
|
|
|
if shared.sd_model.is_sdxl:
|
|
|
|
coeffs = [
|
|
|
|
[ 0.3448, 0.4168, 0.4395],
|
|
|
|
[-0.1953, -0.0290, 0.0250],
|
|
|
|
[ 0.1074, 0.0886, -0.0163],
|
|
|
|
[-0.3730, -0.2499, -0.2088],
|
|
|
|
]
|
|
|
|
else:
|
|
|
|
coeffs = [
|
|
|
|
[ 0.298, 0.207, 0.208],
|
|
|
|
[ 0.187, 0.286, 0.173],
|
|
|
|
[-0.158, 0.189, 0.264],
|
|
|
|
[-0.184, -0.271, -0.473],
|
|
|
|
]
|
|
|
|
|
|
|
|
coefs = torch.tensor(coeffs).to(sample.device)
|
2022-12-25 03:39:00 +08:00
|
|
|
|
2023-08-04 13:38:52 +08:00
|
|
|
x_sample = torch.einsum("...lxy,lr -> ...rxy", sample, coefs)
|
2022-12-25 03:39:00 +08:00
|
|
|
|
|
|
|
return x_sample
|