mlsanityData to 14 Oct 2026

Self-Corrected Flow Distillation for Consistent One-Step and Few-Step Text-to-Image Generation

arXiv 2412.16906 ↗First posted 22 Dec 2024cs.CVcited by 16

On the map: Diffusion models → Flow matching, time series and science

Abstract (arXiv)

Flow matching has emerged as a promising framework for training generative models, demonstrating impressive empirical performance while offering relative ease of training compared to diffusion-based models. However, this method still requires numerous function evaluations in the sampling process. To address these limitations, we introduce a self-corrected flow distillation method that effectively integrates consistency models and adversarial training within the flow-matching framework. This work is a pioneer in achieving consistent generation quality in both few-step and one-step sampling. Our extensive experiments validate the effectiveness of our method, yielding superior results both quantitatively and qualitatively on CelebA-HQ and zero-shot benchmarks on the COCO dataset. Our implementation is released at https://github.com/hao-pt/SCFlow.git.

Key facts. Cited by 16 papers in the map (data through 14 Oct 2026). Placed in Flow matching, time series and science, one of 14 directions in Diffusion models. Builds on 28 of its 41 references that are in the map. First posted on arXiv 22 Dec 2024.

Builds on

28 of 41 references.

Closest ancestors · 3

Same direction, newest first

  1. 2022Efficient Video Prediction via Sparsely Conditioned Flow Matching26 Nov
  2. 2022Rectified Flow: A Marginal Preserving Approach to Optimal Transport29 Sep
  3. 2022Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flow7 Sep

Other references · 25

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2020Denoising Diffusion Probabilistic Modelscited by 21,076
  3. 2015Distilling the Knowledge in a Neural Networkcited by 10,324
  4. 2020Score-Based Generative Modeling through Stochastic Differential Equationscited by 9,122
  5. 2018A Style-Based Generator Architecture for Generative Adversarial Networkscited by 6,013
  6. 2021Zero-Shot Text-to-Image Generationcited by 4,288
  7. 2024Scaling Rectified Flow Transformers for High-Resolution Image Synthesiscited by 4,181
  8. 2019Generative Modeling by Estimating Gradients of the Data Distributioncited by 4,101
  9. 2020Taming Transformers for High-Resolution Image Synthesiscited by 3,253
  10. 2017High-Resolution Image Synthesis and Semantic Manipulation with Conditional GANscited by 1,179
  11. 2019Improved Precision and Recall Metric for Assessing Generative Modelscited by 944
  12. 2023Latent Consistency Models: Synthesizing High-Resolution Images with Few-Step Inferencecited by 846
13 more
  1. 2021Diffusion Probabilistic Models for 3D Point Cloud Generationcited by 735
  2. 2022On Distillation of Guided Diffusion Modelscited by 725
  3. 2021Knowledge Distillation in Iterative Generative Models for Improved Sampling Speedcited by 348
  4. 2023Understanding Diffusion Objectives as the ELBO with Simple Data Augmentationcited by 304
  5. 2023UFOGen: You Forward Once Large Scale Text-to-Image Generation via Diffusion GANscited by 166
  6. 2024Lumina-T2X: Transforming Text into Any Modality, Resolution, and Duration via Flow-based Large Diffusion Transformerscited by 137
  7. 2022Wavelet Diffusion Models are fast and scalable Image Generatorscited by 117
  8. 2023SwiftBrush: One-Step Text-to-Image Diffusion Model with Variational Score Distillationcited by 110
  9. 2023BOOT: Data-free Distillation of Denoising Diffusion Models with Bootstrappingcited by 98
  10. 2020Adversarial score matching and improved sampling for image generationcited by 96
  11. 2024SwiftBrush v2: Make Your One-step Diffusion Model Better Than Its Teachercited by 45
  12. 2023Latent Space Editing in Transformer-Based Flow Matchingcited by 32
  13. 2023A High-Quality Robust Diffusion Framework for Corrupted Datasetcited by 6

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 7; 2026: 9.

16 papers.

Sort:

By area:

  1. 2025Improved Training Technique for Latent Consistency Modelscited by 13
  2. 2025Learning Straight Flows: Variational Flow Matching for Efficient Generationcited by 8
  3. 2025AlignFlow: Improving Flow-based Generative Models with Semi-Discrete Optimal Transportcited by 7
  4. 2025Improved Training Technique for Shortcut Modelscited by 6
  5. 2026FlowCast: Trajectory Forecasting for Scalable Zero-Cost Speculative Flow Matchingcited by 4
  6. 2026FastFlow: Accelerating The Generative Flow Matching Models with Bandit Inferencecited by 4
  7. 2026Perceptual Flow Matching for Few-Step Generative Modelingcited by 4
  8. 2025Fast3Dcache: Training-free 3D Geometry Synthesis Accelerationcited by 3
  9. 2026Geo2: Geometry-Guided Cross-view Geo-Localization and Image Synthesiscited by 3
  10. 2025Flow Matching Calibration for Simulation-Based Inference under Model Misspecificationcited by 2
  11. 2026MPDiT: Multi-Patch Global-to-Local Transformer Architecture For Efficient Flow Matching and Diffusion Modelcited by 2
  12. 2025Accelerating Inference of Masked Image Generators via Reinforcement Learningcited by 1
4 more
  1. 2026ImageRAGTurbo: Towards One-step Text-to-Image Generation with Retrieval-Augmented Diffusion Modelscited by 0
  2. 2026Graphical conditional generative modeling for digital twin modelingcited by 0
  3. 2026Contextual Flow Matching: Adaptive Step Selection in Flow Models for Efficient Visual Generationcited by 0
  4. 2026Data-free Multistep Diffusion Distillationcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Align Your Flow: Scaling Continuous-Time Flow Map Distillation
  2. 2023Flow Matching in Latent Space
  3. 2025FlowSteer: Guiding Few-Step Image Synthesis with Authentic Trajectories
  4. 2024Flow Generator Matching
  5. 2026Alignment of Diffusion Model and Flow Matching for Text-to-Image Generation
  6. 2025TwinFlow: Realizing One-step Generation on Large Models with Self-adversarial Flows
  7. 2026The Amazing Stability of Flow Matching
  8. 2026DanceOPD: On-Policy Generative Field Distillation