direct-denoising-diffusion

name	direct-denoising-diffusion
title	Back to Basics: Let Denoising Generative Models Denoise—Direct x-Prediction in Diffusion
version	0.0.2
engine	skillxiv-v0.0.2-claude-opus-4.6
license	MIT
url	https://arxiv.org/abs/2511.13720
keywords	["Diffusion Models","Direct Denoising","x-Prediction","Generative Modeling","Capacity Efficiency"]
description	Improve diffusion model capacity efficiency by directly predicting clean data instead of noise—leverage the manifold assumption that natural data occupies low-dimensional space while noise spans full dimensionality.

Recover Capacity by Direct Denoising—Predict Clean Data Not Noise

Diffusion models typically predict noise (epsilon) or velocity at each denoising step. This choice seems natural but imposes a hidden cost: noise and velocity distributions span the full data dimensionality, wasting model capacity on high-dimensional structure. Direct x-prediction assumes natural data lies on a low-dimensional manifold—the model only needs to preserve essential information while filtering noise, using capacity more efficiently.

This paper demonstrates that limited-capacity networks can generate high-dimensional data via x-prediction where epsilon/v-prediction fails. The insight is practical: shift the prediction target to align with your data's intrinsic dimensionality, not the ambient space.

Core Concept

Diffusion models work by iteratively adding and removing noise. At each step, the model must predict either:

1. Noise (ε-prediction): The random noise added; distributed across full dimensionality
2. Velocity (v-prediction): A mixture; also spreads across full space
3. Clean data (x-prediction): The denoised output; concentrated on low-dimensional manifold

The mathematical relationships between these are equivalent—they're coordinate transformations of the same underlying process. However, they differ fundamentally in capacity requirement:

• High-dimensional noise: Requires network to preserve all directional information; capacity-intensive
• Low-dimensional data: Network only needs to capture essential structure; more efficient

This manifold assumption—natural data lives in lower-dimensional space than noise—creates practical capacity gaps observable at scale.

Architecture Overview

• Manifold Assumption: Natural data concentrates on low-dimensional manifold; noise fills full high-dimensional space
• x-Prediction Head: Network outputs clean data directly, not noise or velocity
• Matching Framework: Use standard diffusion loss (MSE) adapted for x-prediction via mathematical conversion
• Patch-Based Encoding: For image/video models, use large patches to reduce dimensionality and align with manifold structure
• Capacity Scaling: Lower capacity networks exhibit larger performance gaps; x-prediction shows more benefit at limited budgets

Implementation Steps

Step 1: Reformulate Diffusion Process for x-Prediction. Convert noise schedule to x-prediction formulation.

 torch
 torch.nn  nn

 :
    
     ():
        .timesteps = timesteps

        
        betas = torch.linspace(beta_start, beta_end, timesteps)
        alphas =  - betas
        alphas_cumprod = torch.cumprod(alphas, dim=)

        
        
        .register_buffer(, alphas_cumprod)
        .register_buffer(, torch.sqrt(alphas_cumprod))
        .register_buffer(, torch.sqrt( - alphas_cumprod))

     ():
        
         noise  :
            noise = torch.randn_like(x_0)

        sqrt_alphas = .sqrt_alphas_cumprod[t]
        sqrt_1m_alphas = .sqrt_1m_alphas_cumprod[t]

        
         (sqrt_alphas.shape) < (x_0.shape):
            sqrt_alphas = sqrt_alphas.unsqueeze(-)
            sqrt_1m_alphas = sqrt_1m_alphas.unsqueeze(-)

        
        x_t = sqrt_alphas * x_0 + sqrt_1m_alphas * noise

         x_t, noise

     ():
        
        
         model_prediction

     ():
        
         torch.nn.functional.mse_loss(model_pred, x_0, reduction=)