Explain the loss takes x and output(x) to evaluate in Diffusion

Question

x = x.flatten(start_dim=0, end_dim=1) if x.ndim == 5 else x
self.step += 1
x = x.to(self.device)
output = self.model(x)
noise_loss, photo_loss, frequency_loss = self.estimation_loss(x, output)

why the loss is calculated by (x, output), and think the output as the gt