تولید ویدیو و سه‌بعدی

vid = torch.randn(1, 4, 8, 16, 16)                # (N, C, T, H, W)
model = TinyVideoDiT(in_channels=4, dim=64, depth=2, heads=2)
out, grid = model(vid)            # شبکه (4, 8, 8)، خروجی (1, 256, 32)
tokens, grid = self.patch(x)      # Conv3d، استراید برابر کرنل
xt = x.view(n, T, H * W, d).permute(0, 2, 1, 3).reshape(n * H * W, T, d)
a, _ = self.time_attn(self.ln1(xt), self.ln1(xt), self.ln1(xt), need_weights=False)
xs = xt.view(n, T, H * W, d).reshape(n * T, H * W, d)
a, _ = self.space_attn(self.ln2(xs), self.ln2(xs), self.ln2(xs), need_weights=False)
xs = xs + self.mlp(self.ln3(xs))  # MLP پس از زمان و مکان

model = Splats2D(num_splats=64, image_size=64)
opt = torch.optim.Adam(model.parameters(), lr=0.05)
for step in range(200):
    pred = model((64, 64))        # rasterise_2d درون
    loss = F.mse_loss(pred, target)
    opt.zero_grad(); loss.backward(); opt.step()
alphas = alphas.clamp(0.0, 0.99)  # درون rasterise_2d
order = torch.argsort(depths)     # اسکالر عمق یادگرفته‌شده در 2 بعد
out = out + (T * a)[..., None] * colours_sorted[i][None, None, :]
T = T * (1.0 - a)                 # ضریب عبور نور