From Diffusion to One-Step Generation: A Comparative Study of Flow-Based Models with Application to Image Inpainting
This work addresses the computational bottleneck in generative modeling for image synthesis and inpainting, offering a practical speed-up with competitive quality, though it is incremental in applying existing flow-based methods to inpainting.
The paper tackled the problem of slow iterative sampling in generative models by comparing diffusion and flow-based methods, showing that MeanFlow enables one-step generation with an FID of 29.15 on CIFAR-10, reducing inference time by 50X, and extended CFM to image inpainting with PSNR improvements up to 73%.
We present a comprehensive comparative study of three generative modeling paradigms: Denoising Diffusion Probabilistic Models (DDPM), Conditional Flow Matching (CFM), and MeanFlow. While DDPM and CFM require iterative sampling, MeanFlow enables direct one-step generation by modeling the average velocity over time intervals. We implement all three methods using a unified TinyUNet architecture (<1.5M parameters) on CIFAR-10, demonstrating that CFM achieves an FID of 24.15 with 50 steps, significantly outperforming DDPM (FID 402.98). MeanFlow achieves FID 29.15 with single-step sampling -- a 50X reduction in inference time. We further extend CFM to image inpainting, implementing mask-guided sampling with four mask types (center, random bbox, irregular, half). Our fine-tuned inpainting model achieves substantial improvements: PSNR increases from 4.95 to 8.57 dB on center masks (+73%), and SSIM improves from 0.289 to 0.418 (+45%), demonstrating the effectiveness of inpainting-aware training.