2025-02-25 22:20:35 +00:00
|
|
|
# original version: https://github.com/Wan-Video/Wan2.1/blob/main/wan/modules/model.py
|
|
|
|
# Copyright 2024-2025 The Alibaba Wan Team Authors. All rights reserved.
|
|
|
|
import math
|
|
|
|
|
|
|
|
import torch
|
|
|
|
import torch.nn as nn
|
2025-02-26 00:11:14 +00:00
|
|
|
from einops import repeat
|
2025-02-25 22:20:35 +00:00
|
|
|
|
|
|
|
from comfy.ldm.modules.attention import optimized_attention
|
2025-02-26 00:11:14 +00:00
|
|
|
from comfy.ldm.flux.layers import EmbedND
|
|
|
|
from comfy.ldm.flux.math import apply_rope
|
2025-02-26 10:22:29 +00:00
|
|
|
from comfy.ldm.modules.diffusionmodules.mmdit import RMSNorm
|
2025-02-26 00:56:04 +00:00
|
|
|
import comfy.ldm.common_dit
|
2025-02-26 06:49:43 +00:00
|
|
|
import comfy.model_management
|
2025-02-25 22:20:35 +00:00
|
|
|
|
2025-02-26 10:22:29 +00:00
|
|
|
|
2025-02-25 22:20:35 +00:00
|
|
|
def sinusoidal_embedding_1d(dim, position):
|
|
|
|
# preprocess
|
|
|
|
assert dim % 2 == 0
|
|
|
|
half = dim // 2
|
2025-02-26 21:59:26 +00:00
|
|
|
position = position.type(torch.float32)
|
2025-02-25 22:20:35 +00:00
|
|
|
|
|
|
|
# calculation
|
|
|
|
sinusoid = torch.outer(
|
|
|
|
position, torch.pow(10000, -torch.arange(half).to(position).div(half)))
|
|
|
|
x = torch.cat([torch.cos(sinusoid), torch.sin(sinusoid)], dim=1)
|
|
|
|
return x
|
|
|
|
|
|
|
|
|
|
|
|
class WanSelfAttention(nn.Module):
|
|
|
|
|
|
|
|
def __init__(self,
|
|
|
|
dim,
|
|
|
|
num_heads,
|
|
|
|
window_size=(-1, -1),
|
|
|
|
qk_norm=True,
|
|
|
|
eps=1e-6, operation_settings={}):
|
|
|
|
assert dim % num_heads == 0
|
|
|
|
super().__init__()
|
|
|
|
self.dim = dim
|
|
|
|
self.num_heads = num_heads
|
|
|
|
self.head_dim = dim // num_heads
|
|
|
|
self.window_size = window_size
|
|
|
|
self.qk_norm = qk_norm
|
|
|
|
self.eps = eps
|
|
|
|
|
|
|
|
# layers
|
|
|
|
self.q = operation_settings.get("operations").Linear(dim, dim, device=operation_settings.get("device"), dtype=operation_settings.get("dtype"))
|
|
|
|
self.k = operation_settings.get("operations").Linear(dim, dim, device=operation_settings.get("device"), dtype=operation_settings.get("dtype"))
|
|
|
|
self.v = operation_settings.get("operations").Linear(dim, dim, device=operation_settings.get("device"), dtype=operation_settings.get("dtype"))
|
|
|
|
self.o = operation_settings.get("operations").Linear(dim, dim, device=operation_settings.get("device"), dtype=operation_settings.get("dtype"))
|
2025-02-26 10:22:29 +00:00
|
|
|
self.norm_q = RMSNorm(dim, eps=eps, elementwise_affine=True, device=operation_settings.get("device"), dtype=operation_settings.get("dtype")) if qk_norm else nn.Identity()
|
|
|
|
self.norm_k = RMSNorm(dim, eps=eps, elementwise_affine=True, device=operation_settings.get("device"), dtype=operation_settings.get("dtype")) if qk_norm else nn.Identity()
|
2025-02-25 22:20:35 +00:00
|
|
|
|
2025-02-26 00:56:04 +00:00
|
|
|
def forward(self, x, freqs):
|
2025-02-25 22:20:35 +00:00
|
|
|
r"""
|
|
|
|
Args:
|
|
|
|
x(Tensor): Shape [B, L, num_heads, C / num_heads]
|
|
|
|
freqs(Tensor): Rope freqs, shape [1024, C / num_heads / 2]
|
|
|
|
"""
|
|
|
|
b, s, n, d = *x.shape[:2], self.num_heads, self.head_dim
|
|
|
|
|
|
|
|
# query, key, value function
|
|
|
|
def qkv_fn(x):
|
|
|
|
q = self.norm_q(self.q(x)).view(b, s, n, d)
|
|
|
|
k = self.norm_k(self.k(x)).view(b, s, n, d)
|
|
|
|
v = self.v(x).view(b, s, n * d)
|
|
|
|
return q, k, v
|
|
|
|
|
|
|
|
q, k, v = qkv_fn(x)
|
2025-02-26 00:11:14 +00:00
|
|
|
q, k = apply_rope(q, k, freqs)
|
2025-02-25 22:20:35 +00:00
|
|
|
|
|
|
|
x = optimized_attention(
|
2025-02-26 00:13:39 +00:00
|
|
|
q.view(b, s, n * d),
|
|
|
|
k.view(b, s, n * d),
|
|
|
|
v,
|
2025-02-25 22:20:35 +00:00
|
|
|
heads=self.num_heads,
|
|
|
|
)
|
|
|
|
|
|
|
|
x = self.o(x)
|
|
|
|
return x
|
|
|
|
|
|
|
|
|
|
|
|
class WanT2VCrossAttention(WanSelfAttention):
|
|
|
|
|
2025-02-26 00:56:04 +00:00
|
|
|
def forward(self, x, context):
|
2025-02-25 22:20:35 +00:00
|
|
|
r"""
|
|
|
|
Args:
|
|
|
|
x(Tensor): Shape [B, L1, C]
|
|
|
|
context(Tensor): Shape [B, L2, C]
|
|
|
|
"""
|
|
|
|
# compute query, key, value
|
|
|
|
q = self.norm_q(self.q(x))
|
|
|
|
k = self.norm_k(self.k(context))
|
|
|
|
v = self.v(context)
|
|
|
|
|
|
|
|
# compute attention
|
|
|
|
x = optimized_attention(q, k, v, heads=self.num_heads)
|
|
|
|
|
|
|
|
x = self.o(x)
|
|
|
|
return x
|
|
|
|
|
|
|
|
|
|
|
|
class WanI2VCrossAttention(WanSelfAttention):
|
|
|
|
|
|
|
|
def __init__(self,
|
|
|
|
dim,
|
|
|
|
num_heads,
|
|
|
|
window_size=(-1, -1),
|
|
|
|
qk_norm=True,
|
|
|
|
eps=1e-6, operation_settings={}):
|
2025-02-26 06:49:43 +00:00
|
|
|
super().__init__(dim, num_heads, window_size, qk_norm, eps, operation_settings=operation_settings)
|
2025-02-25 22:20:35 +00:00
|
|
|
|
|
|
|
self.k_img = operation_settings.get("operations").Linear(dim, dim, device=operation_settings.get("device"), dtype=operation_settings.get("dtype"))
|
|
|
|
self.v_img = operation_settings.get("operations").Linear(dim, dim, device=operation_settings.get("device"), dtype=operation_settings.get("dtype"))
|
|
|
|
# self.alpha = nn.Parameter(torch.zeros((1, )))
|
2025-02-26 10:22:29 +00:00
|
|
|
self.norm_k_img = RMSNorm(dim, eps=eps, elementwise_affine=True, device=operation_settings.get("device"), dtype=operation_settings.get("dtype")) if qk_norm else nn.Identity()
|
2025-02-25 22:20:35 +00:00
|
|
|
|
2025-02-26 00:56:04 +00:00
|
|
|
def forward(self, x, context):
|
2025-02-25 22:20:35 +00:00
|
|
|
r"""
|
|
|
|
Args:
|
|
|
|
x(Tensor): Shape [B, L1, C]
|
|
|
|
context(Tensor): Shape [B, L2, C]
|
|
|
|
"""
|
|
|
|
context_img = context[:, :257]
|
|
|
|
context = context[:, 257:]
|
|
|
|
|
|
|
|
# compute query, key, value
|
|
|
|
q = self.norm_q(self.q(x))
|
|
|
|
k = self.norm_k(self.k(context))
|
|
|
|
v = self.v(context)
|
|
|
|
k_img = self.norm_k_img(self.k_img(context_img))
|
|
|
|
v_img = self.v_img(context_img)
|
|
|
|
img_x = optimized_attention(q, k_img, v_img, heads=self.num_heads)
|
|
|
|
# compute attention
|
|
|
|
x = optimized_attention(q, k, v, heads=self.num_heads)
|
|
|
|
|
|
|
|
# output
|
|
|
|
x = x + img_x
|
|
|
|
x = self.o(x)
|
|
|
|
return x
|
|
|
|
|
|
|
|
|
|
|
|
WAN_CROSSATTENTION_CLASSES = {
|
|
|
|
't2v_cross_attn': WanT2VCrossAttention,
|
|
|
|
'i2v_cross_attn': WanI2VCrossAttention,
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
class WanAttentionBlock(nn.Module):
|
|
|
|
|
|
|
|
def __init__(self,
|
|
|
|
cross_attn_type,
|
|
|
|
dim,
|
|
|
|
ffn_dim,
|
|
|
|
num_heads,
|
|
|
|
window_size=(-1, -1),
|
|
|
|
qk_norm=True,
|
|
|
|
cross_attn_norm=False,
|
|
|
|
eps=1e-6, operation_settings={}):
|
|
|
|
super().__init__()
|
|
|
|
self.dim = dim
|
|
|
|
self.ffn_dim = ffn_dim
|
|
|
|
self.num_heads = num_heads
|
|
|
|
self.window_size = window_size
|
|
|
|
self.qk_norm = qk_norm
|
|
|
|
self.cross_attn_norm = cross_attn_norm
|
|
|
|
self.eps = eps
|
|
|
|
|
|
|
|
# layers
|
|
|
|
self.norm1 = operation_settings.get("operations").LayerNorm(dim, eps, elementwise_affine=False, device=operation_settings.get("device"), dtype=operation_settings.get("dtype"))
|
|
|
|
self.self_attn = WanSelfAttention(dim, num_heads, window_size, qk_norm,
|
|
|
|
eps, operation_settings=operation_settings)
|
|
|
|
self.norm3 = operation_settings.get("operations").LayerNorm(
|
|
|
|
dim, eps,
|
|
|
|
elementwise_affine=True, device=operation_settings.get("device"), dtype=operation_settings.get("dtype")) if cross_attn_norm else nn.Identity()
|
|
|
|
self.cross_attn = WAN_CROSSATTENTION_CLASSES[cross_attn_type](dim,
|
|
|
|
num_heads,
|
|
|
|
(-1, -1),
|
|
|
|
qk_norm,
|
|
|
|
eps, operation_settings=operation_settings)
|
|
|
|
self.norm2 = operation_settings.get("operations").LayerNorm(dim, eps, elementwise_affine=False, device=operation_settings.get("device"), dtype=operation_settings.get("dtype"))
|
|
|
|
self.ffn = nn.Sequential(
|
|
|
|
operation_settings.get("operations").Linear(dim, ffn_dim, device=operation_settings.get("device"), dtype=operation_settings.get("dtype")), nn.GELU(approximate='tanh'),
|
|
|
|
operation_settings.get("operations").Linear(ffn_dim, dim, device=operation_settings.get("device"), dtype=operation_settings.get("dtype")))
|
|
|
|
|
|
|
|
# modulation
|
|
|
|
self.modulation = nn.Parameter(torch.empty(1, 6, dim, device=operation_settings.get("device"), dtype=operation_settings.get("dtype")))
|
|
|
|
|
|
|
|
def forward(
|
|
|
|
self,
|
|
|
|
x,
|
|
|
|
e,
|
|
|
|
freqs,
|
|
|
|
context,
|
|
|
|
):
|
|
|
|
r"""
|
|
|
|
Args:
|
|
|
|
x(Tensor): Shape [B, L, C]
|
|
|
|
e(Tensor): Shape [B, 6, C]
|
|
|
|
freqs(Tensor): Rope freqs, shape [1024, C / num_heads / 2]
|
|
|
|
"""
|
|
|
|
# assert e.dtype == torch.float32
|
|
|
|
|
2025-02-26 06:49:43 +00:00
|
|
|
e = (comfy.model_management.cast_to(self.modulation, dtype=x.dtype, device=x.device) + e).chunk(6, dim=1)
|
2025-02-25 22:20:35 +00:00
|
|
|
# assert e[0].dtype == torch.float32
|
|
|
|
|
|
|
|
# self-attention
|
|
|
|
y = self.self_attn(
|
2025-02-26 00:56:04 +00:00
|
|
|
self.norm1(x) * (1 + e[1]) + e[0],
|
2025-02-25 22:20:35 +00:00
|
|
|
freqs)
|
|
|
|
|
|
|
|
x = x + y * e[2]
|
|
|
|
|
2025-02-27 12:22:42 +00:00
|
|
|
# cross-attention & ffn
|
|
|
|
x = x + self.cross_attn(self.norm3(x), context)
|
|
|
|
y = self.ffn(self.norm2(x) * (1 + e[4]) + e[3])
|
|
|
|
x = x + y * e[5]
|
2025-02-25 22:20:35 +00:00
|
|
|
return x
|
|
|
|
|
|
|
|
|
|
|
|
class Head(nn.Module):
|
|
|
|
|
|
|
|
def __init__(self, dim, out_dim, patch_size, eps=1e-6, operation_settings={}):
|
|
|
|
super().__init__()
|
|
|
|
self.dim = dim
|
|
|
|
self.out_dim = out_dim
|
|
|
|
self.patch_size = patch_size
|
|
|
|
self.eps = eps
|
|
|
|
|
|
|
|
# layers
|
|
|
|
out_dim = math.prod(patch_size) * out_dim
|
|
|
|
self.norm = operation_settings.get("operations").LayerNorm(dim, eps, elementwise_affine=False, device=operation_settings.get("device"), dtype=operation_settings.get("dtype"))
|
|
|
|
self.head = operation_settings.get("operations").Linear(dim, out_dim, device=operation_settings.get("device"), dtype=operation_settings.get("dtype"))
|
|
|
|
|
|
|
|
# modulation
|
|
|
|
self.modulation = nn.Parameter(torch.empty(1, 2, dim, device=operation_settings.get("device"), dtype=operation_settings.get("dtype")))
|
|
|
|
|
|
|
|
def forward(self, x, e):
|
|
|
|
r"""
|
|
|
|
Args:
|
|
|
|
x(Tensor): Shape [B, L1, C]
|
|
|
|
e(Tensor): Shape [B, C]
|
|
|
|
"""
|
|
|
|
# assert e.dtype == torch.float32
|
2025-02-26 06:49:43 +00:00
|
|
|
e = (comfy.model_management.cast_to(self.modulation, dtype=x.dtype, device=x.device) + e.unsqueeze(1)).chunk(2, dim=1)
|
2025-02-25 22:20:35 +00:00
|
|
|
x = (self.head(self.norm(x) * (1 + e[1]) + e[0]))
|
|
|
|
return x
|
|
|
|
|
|
|
|
|
|
|
|
class MLPProj(torch.nn.Module):
|
|
|
|
|
|
|
|
def __init__(self, in_dim, out_dim, operation_settings={}):
|
|
|
|
super().__init__()
|
|
|
|
|
|
|
|
self.proj = torch.nn.Sequential(
|
|
|
|
operation_settings.get("operations").LayerNorm(in_dim, device=operation_settings.get("device"), dtype=operation_settings.get("dtype")), operation_settings.get("operations").Linear(in_dim, in_dim, device=operation_settings.get("device"), dtype=operation_settings.get("dtype")),
|
|
|
|
torch.nn.GELU(), operation_settings.get("operations").Linear(in_dim, out_dim, device=operation_settings.get("device"), dtype=operation_settings.get("dtype")),
|
|
|
|
operation_settings.get("operations").LayerNorm(out_dim, device=operation_settings.get("device"), dtype=operation_settings.get("dtype")))
|
|
|
|
|
|
|
|
def forward(self, image_embeds):
|
|
|
|
clip_extra_context_tokens = self.proj(image_embeds)
|
|
|
|
return clip_extra_context_tokens
|
|
|
|
|
|
|
|
|
|
|
|
class WanModel(torch.nn.Module):
|
|
|
|
r"""
|
|
|
|
Wan diffusion backbone supporting both text-to-video and image-to-video.
|
|
|
|
"""
|
|
|
|
|
|
|
|
def __init__(self,
|
|
|
|
model_type='t2v',
|
|
|
|
patch_size=(1, 2, 2),
|
|
|
|
text_len=512,
|
|
|
|
in_dim=16,
|
|
|
|
dim=2048,
|
|
|
|
ffn_dim=8192,
|
|
|
|
freq_dim=256,
|
|
|
|
text_dim=4096,
|
|
|
|
out_dim=16,
|
|
|
|
num_heads=16,
|
|
|
|
num_layers=32,
|
|
|
|
window_size=(-1, -1),
|
|
|
|
qk_norm=True,
|
|
|
|
cross_attn_norm=True,
|
|
|
|
eps=1e-6,
|
|
|
|
image_model=None,
|
|
|
|
device=None,
|
|
|
|
dtype=None,
|
|
|
|
operations=None,
|
|
|
|
):
|
|
|
|
r"""
|
|
|
|
Initialize the diffusion model backbone.
|
|
|
|
|
|
|
|
Args:
|
|
|
|
model_type (`str`, *optional*, defaults to 't2v'):
|
|
|
|
Model variant - 't2v' (text-to-video) or 'i2v' (image-to-video)
|
|
|
|
patch_size (`tuple`, *optional*, defaults to (1, 2, 2)):
|
|
|
|
3D patch dimensions for video embedding (t_patch, h_patch, w_patch)
|
|
|
|
text_len (`int`, *optional*, defaults to 512):
|
|
|
|
Fixed length for text embeddings
|
|
|
|
in_dim (`int`, *optional*, defaults to 16):
|
|
|
|
Input video channels (C_in)
|
|
|
|
dim (`int`, *optional*, defaults to 2048):
|
|
|
|
Hidden dimension of the transformer
|
|
|
|
ffn_dim (`int`, *optional*, defaults to 8192):
|
|
|
|
Intermediate dimension in feed-forward network
|
|
|
|
freq_dim (`int`, *optional*, defaults to 256):
|
|
|
|
Dimension for sinusoidal time embeddings
|
|
|
|
text_dim (`int`, *optional*, defaults to 4096):
|
|
|
|
Input dimension for text embeddings
|
|
|
|
out_dim (`int`, *optional*, defaults to 16):
|
|
|
|
Output video channels (C_out)
|
|
|
|
num_heads (`int`, *optional*, defaults to 16):
|
|
|
|
Number of attention heads
|
|
|
|
num_layers (`int`, *optional*, defaults to 32):
|
|
|
|
Number of transformer blocks
|
|
|
|
window_size (`tuple`, *optional*, defaults to (-1, -1)):
|
|
|
|
Window size for local attention (-1 indicates global attention)
|
|
|
|
qk_norm (`bool`, *optional*, defaults to True):
|
|
|
|
Enable query/key normalization
|
|
|
|
cross_attn_norm (`bool`, *optional*, defaults to False):
|
|
|
|
Enable cross-attention normalization
|
|
|
|
eps (`float`, *optional*, defaults to 1e-6):
|
|
|
|
Epsilon value for normalization layers
|
|
|
|
"""
|
|
|
|
|
|
|
|
super().__init__()
|
|
|
|
self.dtype = dtype
|
|
|
|
operation_settings = {"operations": operations, "device": device, "dtype": dtype}
|
|
|
|
|
|
|
|
assert model_type in ['t2v', 'i2v']
|
|
|
|
self.model_type = model_type
|
|
|
|
|
|
|
|
self.patch_size = patch_size
|
|
|
|
self.text_len = text_len
|
|
|
|
self.in_dim = in_dim
|
|
|
|
self.dim = dim
|
|
|
|
self.ffn_dim = ffn_dim
|
|
|
|
self.freq_dim = freq_dim
|
|
|
|
self.text_dim = text_dim
|
|
|
|
self.out_dim = out_dim
|
|
|
|
self.num_heads = num_heads
|
|
|
|
self.num_layers = num_layers
|
|
|
|
self.window_size = window_size
|
|
|
|
self.qk_norm = qk_norm
|
|
|
|
self.cross_attn_norm = cross_attn_norm
|
|
|
|
self.eps = eps
|
|
|
|
|
|
|
|
# embeddings
|
|
|
|
self.patch_embedding = operations.Conv3d(
|
2025-02-26 21:59:26 +00:00
|
|
|
in_dim, dim, kernel_size=patch_size, stride=patch_size, device=operation_settings.get("device"), dtype=torch.float32)
|
2025-02-25 22:20:35 +00:00
|
|
|
self.text_embedding = nn.Sequential(
|
|
|
|
operations.Linear(text_dim, dim, device=operation_settings.get("device"), dtype=operation_settings.get("dtype")), nn.GELU(approximate='tanh'),
|
|
|
|
operations.Linear(dim, dim, device=operation_settings.get("device"), dtype=operation_settings.get("dtype")))
|
|
|
|
|
|
|
|
self.time_embedding = nn.Sequential(
|
|
|
|
operations.Linear(freq_dim, dim, device=operation_settings.get("device"), dtype=operation_settings.get("dtype")), nn.SiLU(), operations.Linear(dim, dim, device=operation_settings.get("device"), dtype=operation_settings.get("dtype")))
|
|
|
|
self.time_projection = nn.Sequential(nn.SiLU(), operations.Linear(dim, dim * 6, device=operation_settings.get("device"), dtype=operation_settings.get("dtype")))
|
|
|
|
|
|
|
|
# blocks
|
|
|
|
cross_attn_type = 't2v_cross_attn' if model_type == 't2v' else 'i2v_cross_attn'
|
|
|
|
self.blocks = nn.ModuleList([
|
|
|
|
WanAttentionBlock(cross_attn_type, dim, ffn_dim, num_heads,
|
|
|
|
window_size, qk_norm, cross_attn_norm, eps, operation_settings=operation_settings)
|
|
|
|
for _ in range(num_layers)
|
|
|
|
])
|
|
|
|
|
|
|
|
# head
|
|
|
|
self.head = Head(dim, out_dim, patch_size, eps, operation_settings=operation_settings)
|
|
|
|
|
|
|
|
d = dim // num_heads
|
2025-02-26 00:11:14 +00:00
|
|
|
self.rope_embedder = EmbedND(dim=d, theta=10000.0, axes_dim=[d - 4 * (d // 6), 2 * (d // 6), 2 * (d // 6)])
|
2025-02-25 22:20:35 +00:00
|
|
|
|
|
|
|
if model_type == 'i2v':
|
|
|
|
self.img_emb = MLPProj(1280, dim, operation_settings=operation_settings)
|
2025-02-26 13:38:09 +00:00
|
|
|
else:
|
|
|
|
self.img_emb = None
|
2025-02-25 22:20:35 +00:00
|
|
|
|
|
|
|
def forward_orig(
|
|
|
|
self,
|
|
|
|
x,
|
|
|
|
t,
|
|
|
|
context,
|
|
|
|
clip_fea=None,
|
2025-02-26 00:11:14 +00:00
|
|
|
freqs=None,
|
2025-02-25 22:20:35 +00:00
|
|
|
):
|
|
|
|
r"""
|
|
|
|
Forward pass through the diffusion model
|
|
|
|
|
|
|
|
Args:
|
2025-02-26 00:56:04 +00:00
|
|
|
x (Tensor):
|
|
|
|
List of input video tensors with shape [B, C_in, F, H, W]
|
2025-02-25 22:20:35 +00:00
|
|
|
t (Tensor):
|
|
|
|
Diffusion timesteps tensor of shape [B]
|
|
|
|
context (List[Tensor]):
|
2025-02-26 00:56:04 +00:00
|
|
|
List of text embeddings each with shape [B, L, C]
|
2025-02-25 22:20:35 +00:00
|
|
|
seq_len (`int`):
|
|
|
|
Maximum sequence length for positional encoding
|
|
|
|
clip_fea (Tensor, *optional*):
|
|
|
|
CLIP image features for image-to-video mode
|
|
|
|
y (List[Tensor], *optional*):
|
|
|
|
Conditional video inputs for image-to-video mode, same shape as x
|
|
|
|
|
|
|
|
Returns:
|
|
|
|
List[Tensor]:
|
|
|
|
List of denoised video tensors with original input shapes [C_out, F, H / 8, W / 8]
|
|
|
|
"""
|
|
|
|
# embeddings
|
2025-02-26 21:59:26 +00:00
|
|
|
x = self.patch_embedding(x.float()).to(x.dtype)
|
2025-02-26 00:56:04 +00:00
|
|
|
grid_sizes = x.shape[2:]
|
|
|
|
x = x.flatten(2).transpose(1, 2)
|
2025-02-25 22:20:35 +00:00
|
|
|
|
|
|
|
# time embeddings
|
|
|
|
e = self.time_embedding(
|
|
|
|
sinusoidal_embedding_1d(self.freq_dim, t).to(dtype=x[0].dtype))
|
|
|
|
e0 = self.time_projection(e).unflatten(1, (6, self.dim))
|
|
|
|
|
|
|
|
# context
|
2025-02-27 01:34:02 +00:00
|
|
|
context = self.text_embedding(context)
|
2025-02-25 22:20:35 +00:00
|
|
|
|
2025-02-26 13:38:09 +00:00
|
|
|
if clip_fea is not None and self.img_emb is not None:
|
2025-02-25 22:20:35 +00:00
|
|
|
context_clip = self.img_emb(clip_fea) # bs x 257 x dim
|
|
|
|
context = torch.concat([context_clip, context], dim=1)
|
|
|
|
|
|
|
|
# arguments
|
|
|
|
kwargs = dict(
|
|
|
|
e=e0,
|
2025-02-26 00:11:14 +00:00
|
|
|
freqs=freqs,
|
2025-02-26 00:56:04 +00:00
|
|
|
context=context)
|
2025-02-25 22:20:35 +00:00
|
|
|
|
|
|
|
for block in self.blocks:
|
|
|
|
x = block(x, **kwargs)
|
|
|
|
|
|
|
|
# head
|
|
|
|
x = self.head(x, e)
|
|
|
|
|
|
|
|
# unpatchify
|
|
|
|
x = self.unpatchify(x, grid_sizes)
|
|
|
|
return x
|
|
|
|
|
2025-02-26 06:49:43 +00:00
|
|
|
def forward(self, x, timestep, context, clip_fea=None, **kwargs):
|
2025-02-26 00:11:14 +00:00
|
|
|
bs, c, t, h, w = x.shape
|
2025-02-26 00:56:04 +00:00
|
|
|
x = comfy.ldm.common_dit.pad_to_patch_size(x, self.patch_size)
|
2025-02-26 00:11:14 +00:00
|
|
|
patch_size = self.patch_size
|
|
|
|
t_len = ((t + (patch_size[0] // 2)) // patch_size[0])
|
|
|
|
h_len = ((h + (patch_size[1] // 2)) // patch_size[1])
|
|
|
|
w_len = ((w + (patch_size[2] // 2)) // patch_size[2])
|
|
|
|
img_ids = torch.zeros((t_len, h_len, w_len, 3), device=x.device, dtype=x.dtype)
|
|
|
|
img_ids[:, :, :, 0] = img_ids[:, :, :, 0] + torch.linspace(0, t_len - 1, steps=t_len, device=x.device, dtype=x.dtype).reshape(-1, 1, 1)
|
|
|
|
img_ids[:, :, :, 1] = img_ids[:, :, :, 1] + torch.linspace(0, h_len - 1, steps=h_len, device=x.device, dtype=x.dtype).reshape(1, -1, 1)
|
|
|
|
img_ids[:, :, :, 2] = img_ids[:, :, :, 2] + torch.linspace(0, w_len - 1, steps=w_len, device=x.device, dtype=x.dtype).reshape(1, 1, -1)
|
|
|
|
img_ids = repeat(img_ids, "t h w c -> b (t h w) c", b=bs)
|
|
|
|
|
|
|
|
freqs = self.rope_embedder(img_ids).movedim(1, 2)
|
2025-02-26 06:49:43 +00:00
|
|
|
return self.forward_orig(x, timestep, context, clip_fea=clip_fea, freqs=freqs)[:, :, :t, :h, :w]
|
2025-02-25 22:20:35 +00:00
|
|
|
|
|
|
|
def unpatchify(self, x, grid_sizes):
|
|
|
|
r"""
|
|
|
|
Reconstruct video tensors from patch embeddings.
|
|
|
|
|
|
|
|
Args:
|
|
|
|
x (List[Tensor]):
|
|
|
|
List of patchified features, each with shape [L, C_out * prod(patch_size)]
|
|
|
|
grid_sizes (Tensor):
|
|
|
|
Original spatial-temporal grid dimensions before patching,
|
|
|
|
shape [B, 3] (3 dimensions correspond to F_patches, H_patches, W_patches)
|
|
|
|
|
|
|
|
Returns:
|
|
|
|
List[Tensor]:
|
2025-02-26 00:56:04 +00:00
|
|
|
Reconstructed video tensors with shape [L, C_out, F, H / 8, W / 8]
|
2025-02-25 22:20:35 +00:00
|
|
|
"""
|
|
|
|
|
|
|
|
c = self.out_dim
|
2025-02-26 00:56:04 +00:00
|
|
|
u = x
|
|
|
|
b = u.shape[0]
|
|
|
|
u = u[:, :math.prod(grid_sizes)].view(b, *grid_sizes, *self.patch_size, c)
|
|
|
|
u = torch.einsum('bfhwpqrc->bcfphqwr', u)
|
|
|
|
u = u.reshape(b, c, *[i * j for i, j in zip(grid_sizes, self.patch_size)])
|
|
|
|
return u
|