Browse Source

Stable Cascade Stage A.

pull/2811/head
comfyanonymous 9 months ago
parent
commit
5e06baf112
  1. 254
      comfy/ldm/cascade/stage_a.py
  2. 24
      comfy/sd.py

254
comfy/ldm/cascade/stage_a.py

@ -0,0 +1,254 @@
"""
This file is part of ComfyUI.
Copyright (C) 2024 Stability AI
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <https://www.gnu.org/licenses/>.
"""
import torch
from torch import nn
from torch.autograd import Function
class vector_quantize(Function):
@staticmethod
def forward(ctx, x, codebook):
with torch.no_grad():
codebook_sqr = torch.sum(codebook ** 2, dim=1)
x_sqr = torch.sum(x ** 2, dim=1, keepdim=True)
dist = torch.addmm(codebook_sqr + x_sqr, x, codebook.t(), alpha=-2.0, beta=1.0)
_, indices = dist.min(dim=1)
ctx.save_for_backward(indices, codebook)
ctx.mark_non_differentiable(indices)
nn = torch.index_select(codebook, 0, indices)
return nn, indices
@staticmethod
def backward(ctx, grad_output, grad_indices):
grad_inputs, grad_codebook = None, None
if ctx.needs_input_grad[0]:
grad_inputs = grad_output.clone()
if ctx.needs_input_grad[1]:
# Gradient wrt. the codebook
indices, codebook = ctx.saved_tensors
grad_codebook = torch.zeros_like(codebook)
grad_codebook.index_add_(0, indices, grad_output)
return (grad_inputs, grad_codebook)
class VectorQuantize(nn.Module):
def __init__(self, embedding_size, k, ema_decay=0.99, ema_loss=False):
"""
Takes an input of variable size (as long as the last dimension matches the embedding size).
Returns one tensor containing the nearest neigbour embeddings to each of the inputs,
with the same size as the input, vq and commitment components for the loss as a touple
in the second output and the indices of the quantized vectors in the third:
quantized, (vq_loss, commit_loss), indices
"""
super(VectorQuantize, self).__init__()
self.codebook = nn.Embedding(k, embedding_size)
self.codebook.weight.data.uniform_(-1./k, 1./k)
self.vq = vector_quantize.apply
self.ema_decay = ema_decay
self.ema_loss = ema_loss
if ema_loss:
self.register_buffer('ema_element_count', torch.ones(k))
self.register_buffer('ema_weight_sum', torch.zeros_like(self.codebook.weight))
def _laplace_smoothing(self, x, epsilon):
n = torch.sum(x)
return ((x + epsilon) / (n + x.size(0) * epsilon) * n)
def _updateEMA(self, z_e_x, indices):
mask = nn.functional.one_hot(indices, self.ema_element_count.size(0)).float()
elem_count = mask.sum(dim=0)
weight_sum = torch.mm(mask.t(), z_e_x)
self.ema_element_count = (self.ema_decay * self.ema_element_count) + ((1-self.ema_decay) * elem_count)
self.ema_element_count = self._laplace_smoothing(self.ema_element_count, 1e-5)
self.ema_weight_sum = (self.ema_decay * self.ema_weight_sum) + ((1-self.ema_decay) * weight_sum)
self.codebook.weight.data = self.ema_weight_sum / self.ema_element_count.unsqueeze(-1)
def idx2vq(self, idx, dim=-1):
q_idx = self.codebook(idx)
if dim != -1:
q_idx = q_idx.movedim(-1, dim)
return q_idx
def forward(self, x, get_losses=True, dim=-1):
if dim != -1:
x = x.movedim(dim, -1)
z_e_x = x.contiguous().view(-1, x.size(-1)) if len(x.shape) > 2 else x
z_q_x, indices = self.vq(z_e_x, self.codebook.weight.detach())
vq_loss, commit_loss = None, None
if self.ema_loss and self.training:
self._updateEMA(z_e_x.detach(), indices.detach())
# pick the graded embeddings after updating the codebook in order to have a more accurate commitment loss
z_q_x_grd = torch.index_select(self.codebook.weight, dim=0, index=indices)
if get_losses:
vq_loss = (z_q_x_grd - z_e_x.detach()).pow(2).mean()
commit_loss = (z_e_x - z_q_x_grd.detach()).pow(2).mean()
z_q_x = z_q_x.view(x.shape)
if dim != -1:
z_q_x = z_q_x.movedim(-1, dim)
return z_q_x, (vq_loss, commit_loss), indices.view(x.shape[:-1])
class ResBlock(nn.Module):
def __init__(self, c, c_hidden):
super().__init__()
# depthwise/attention
self.norm1 = nn.LayerNorm(c, elementwise_affine=False, eps=1e-6)
self.depthwise = nn.Sequential(
nn.ReplicationPad2d(1),
nn.Conv2d(c, c, kernel_size=3, groups=c)
)
# channelwise
self.norm2 = nn.LayerNorm(c, elementwise_affine=False, eps=1e-6)
self.channelwise = nn.Sequential(
nn.Linear(c, c_hidden),
nn.GELU(),
nn.Linear(c_hidden, c),
)
self.gammas = nn.Parameter(torch.zeros(6), requires_grad=True)
# Init weights
def _basic_init(module):
if isinstance(module, nn.Linear) or isinstance(module, nn.Conv2d):
torch.nn.init.xavier_uniform_(module.weight)
if module.bias is not None:
nn.init.constant_(module.bias, 0)
self.apply(_basic_init)
def _norm(self, x, norm):
return norm(x.permute(0, 2, 3, 1)).permute(0, 3, 1, 2)
def forward(self, x):
mods = self.gammas
x_temp = self._norm(x, self.norm1) * (1 + mods[0]) + mods[1]
x = x + self.depthwise(x_temp) * mods[2]
x_temp = self._norm(x, self.norm2) * (1 + mods[3]) + mods[4]
x = x + self.channelwise(x_temp.permute(0, 2, 3, 1)).permute(0, 3, 1, 2) * mods[5]
return x
class StageA(nn.Module):
def __init__(self, levels=2, bottleneck_blocks=12, c_hidden=384, c_latent=4, codebook_size=8192,
scale_factor=0.43): # 0.3764
super().__init__()
self.c_latent = c_latent
self.scale_factor = scale_factor
c_levels = [c_hidden // (2 ** i) for i in reversed(range(levels))]
# Encoder blocks
self.in_block = nn.Sequential(
nn.PixelUnshuffle(2),
nn.Conv2d(3 * 4, c_levels[0], kernel_size=1)
)
down_blocks = []
for i in range(levels):
if i > 0:
down_blocks.append(nn.Conv2d(c_levels[i - 1], c_levels[i], kernel_size=4, stride=2, padding=1))
block = ResBlock(c_levels[i], c_levels[i] * 4)
down_blocks.append(block)
down_blocks.append(nn.Sequential(
nn.Conv2d(c_levels[-1], c_latent, kernel_size=1, bias=False),
nn.BatchNorm2d(c_latent), # then normalize them to have mean 0 and std 1
))
self.down_blocks = nn.Sequential(*down_blocks)
self.down_blocks[0]
self.codebook_size = codebook_size
self.vquantizer = VectorQuantize(c_latent, k=codebook_size)
# Decoder blocks
up_blocks = [nn.Sequential(
nn.Conv2d(c_latent, c_levels[-1], kernel_size=1)
)]
for i in range(levels):
for j in range(bottleneck_blocks if i == 0 else 1):
block = ResBlock(c_levels[levels - 1 - i], c_levels[levels - 1 - i] * 4)
up_blocks.append(block)
if i < levels - 1:
up_blocks.append(
nn.ConvTranspose2d(c_levels[levels - 1 - i], c_levels[levels - 2 - i], kernel_size=4, stride=2,
padding=1))
self.up_blocks = nn.Sequential(*up_blocks)
self.out_block = nn.Sequential(
nn.Conv2d(c_levels[0], 3 * 4, kernel_size=1),
nn.PixelShuffle(2),
)
def encode(self, x, quantize=False):
x = self.in_block(x)
x = self.down_blocks(x)
if quantize:
qe, (vq_loss, commit_loss), indices = self.vquantizer.forward(x, dim=1)
return qe / self.scale_factor, x / self.scale_factor, indices, vq_loss + commit_loss * 0.25
else:
return x / self.scale_factor
def decode(self, x):
x = x * self.scale_factor
x = self.up_blocks(x)
x = self.out_block(x)
return x
def forward(self, x, quantize=False):
qe, x, _, vq_loss = self.encode(x, quantize)
x = self.decode(qe)
return x, vq_loss
class Discriminator(nn.Module):
def __init__(self, c_in=3, c_cond=0, c_hidden=512, depth=6):
super().__init__()
d = max(depth - 3, 3)
layers = [
nn.utils.spectral_norm(nn.Conv2d(c_in, c_hidden // (2 ** d), kernel_size=3, stride=2, padding=1)),
nn.LeakyReLU(0.2),
]
for i in range(depth - 1):
c_in = c_hidden // (2 ** max((d - i), 0))
c_out = c_hidden // (2 ** max((d - 1 - i), 0))
layers.append(nn.utils.spectral_norm(nn.Conv2d(c_in, c_out, kernel_size=3, stride=2, padding=1)))
layers.append(nn.InstanceNorm2d(c_out))
layers.append(nn.LeakyReLU(0.2))
self.encoder = nn.Sequential(*layers)
self.shuffle = nn.Conv2d((c_hidden + c_cond) if c_cond > 0 else c_hidden, 1, kernel_size=1)
self.logits = nn.Sigmoid()
def forward(self, x, cond=None):
x = self.encoder(x)
if cond is not None:
cond = cond.view(cond.size(0), cond.size(1), 1, 1, ).expand(-1, -1, x.size(-2), x.size(-1))
x = torch.cat([x, cond], dim=1)
x = self.shuffle(x)
x = self.logits(x)
return x

24
comfy/sd.py

@ -2,6 +2,8 @@ import torch
from comfy import model_management
from .ldm.models.autoencoder import AutoencoderKL, AutoencodingEngine
from .ldm.cascade.stage_a import StageA
import yaml
import comfy.utils
@ -156,6 +158,8 @@ class VAE:
self.memory_used_decode = lambda shape, dtype: (2178 * shape[2] * shape[3] * 64) * model_management.dtype_size(dtype)
self.downscale_ratio = 8
self.latent_channels = 4
self.process_input = lambda image: image * 2.0 - 1.0
self.process_output = lambda image: torch.clamp((image + 1.0) / 2.0, min=0.0, max=1.0)
if config is None:
if "decoder.mid.block_1.mix_factor" in sd:
@ -168,6 +172,14 @@ class VAE:
decoder_config={'target': "comfy.ldm.modules.temporal_ae.VideoDecoder", 'params': decoder_config})
elif "taesd_decoder.1.weight" in sd:
self.first_stage_model = comfy.taesd.taesd.TAESD()
elif "vquantizer.codebook.weight" in sd: #VQGan: stage a of stable cascade
self.first_stage_model = StageA()
self.downscale_ratio = 4
#TODO
#self.memory_used_encode
#self.memory_used_decode
self.process_input = lambda image: image
self.process_output = lambda image: image
else:
#default SD1.x/SD2.x VAE parameters
ddconfig = {'double_z': True, 'z_channels': 4, 'resolution': 256, 'in_channels': 3, 'out_ch': 3, 'ch': 128, 'ch_mult': [1, 2, 4, 4], 'num_res_blocks': 2, 'attn_resolutions': [], 'dropout': 0.0}
@ -206,12 +218,12 @@ class VAE:
steps += samples.shape[0] * comfy.utils.get_tiled_scale_steps(samples.shape[3], samples.shape[2], tile_x * 2, tile_y // 2, overlap)
pbar = comfy.utils.ProgressBar(steps)
decode_fn = lambda a: (self.first_stage_model.decode(a.to(self.vae_dtype).to(self.device)) + 1.0).float()
output = torch.clamp((
decode_fn = lambda a: self.first_stage_model.decode(a.to(self.vae_dtype).to(self.device)).float()
output = self.process_output(
(comfy.utils.tiled_scale(samples, decode_fn, tile_x // 2, tile_y * 2, overlap, upscale_amount = self.downscale_ratio, output_device=self.output_device, pbar = pbar) +
comfy.utils.tiled_scale(samples, decode_fn, tile_x * 2, tile_y // 2, overlap, upscale_amount = self.downscale_ratio, output_device=self.output_device, pbar = pbar) +
comfy.utils.tiled_scale(samples, decode_fn, tile_x, tile_y, overlap, upscale_amount = self.downscale_ratio, output_device=self.output_device, pbar = pbar))
/ 3.0) / 2.0, min=0.0, max=1.0)
/ 3.0)
return output
def encode_tiled_(self, pixel_samples, tile_x=512, tile_y=512, overlap = 64):
@ -220,7 +232,7 @@ class VAE:
steps += pixel_samples.shape[0] * comfy.utils.get_tiled_scale_steps(pixel_samples.shape[3], pixel_samples.shape[2], tile_x * 2, tile_y // 2, overlap)
pbar = comfy.utils.ProgressBar(steps)
encode_fn = lambda a: self.first_stage_model.encode((2. * a - 1.).to(self.vae_dtype).to(self.device)).float()
encode_fn = lambda a: self.first_stage_model.encode((self.process_input(a)).to(self.vae_dtype).to(self.device)).float()
samples = comfy.utils.tiled_scale(pixel_samples, encode_fn, tile_x, tile_y, overlap, upscale_amount = (1/self.downscale_ratio), out_channels=self.latent_channels, output_device=self.output_device, pbar=pbar)
samples += comfy.utils.tiled_scale(pixel_samples, encode_fn, tile_x * 2, tile_y // 2, overlap, upscale_amount = (1/self.downscale_ratio), out_channels=self.latent_channels, output_device=self.output_device, pbar=pbar)
samples += comfy.utils.tiled_scale(pixel_samples, encode_fn, tile_x // 2, tile_y * 2, overlap, upscale_amount = (1/self.downscale_ratio), out_channels=self.latent_channels, output_device=self.output_device, pbar=pbar)
@ -238,7 +250,7 @@ class VAE:
pixel_samples = torch.empty((samples_in.shape[0], 3, round(samples_in.shape[2] * self.downscale_ratio), round(samples_in.shape[3] * self.downscale_ratio)), device=self.output_device)
for x in range(0, samples_in.shape[0], batch_number):
samples = samples_in[x:x+batch_number].to(self.vae_dtype).to(self.device)
pixel_samples[x:x+batch_number] = torch.clamp((self.first_stage_model.decode(samples).to(self.output_device).float() + 1.0) / 2.0, min=0.0, max=1.0)
pixel_samples[x:x+batch_number] = self.process_output(self.first_stage_model.decode(samples).to(self.output_device).float())
except model_management.OOM_EXCEPTION as e:
print("Warning: Ran out of memory when regular VAE decoding, retrying with tiled VAE decoding.")
pixel_samples = self.decode_tiled_(samples_in)
@ -261,7 +273,7 @@ class VAE:
batch_number = max(1, batch_number)
samples = torch.empty((pixel_samples.shape[0], self.latent_channels, round(pixel_samples.shape[2] // self.downscale_ratio), round(pixel_samples.shape[3] // self.downscale_ratio)), device=self.output_device)
for x in range(0, pixel_samples.shape[0], batch_number):
pixels_in = (2. * pixel_samples[x:x+batch_number] - 1.).to(self.vae_dtype).to(self.device)
pixels_in = self.process_input(pixel_samples[x:x+batch_number]).to(self.vae_dtype).to(self.device)
samples[x:x+batch_number] = self.first_stage_model.encode(pixels_in).to(self.output_device).float()
except model_management.OOM_EXCEPTION as e:

Loading…
Cancel
Save