BiFM: Bidirectional Flow Matching for Few-Step Image Editing and Generation

Yasong Dai; Zeeshan Hayder; David Ahmedt-Aristizabal; Hongdong Li

cs.CVMar 26, 2026

BiFM: Bidirectional Flow Matching for Few-Step Image Editing and Generation

Authors: Yasong Dai, Zeeshan Hayder, David Ahmedt-Aristizabal, Hongdong Li

Organizations: Australian National University, Australia · Data61-CSIRO, Australia

Abstract

Recent diffusion and flow matching models have demonstrated strong capabilities in image generation and editing by progressively removing noise through iterative sampling. While this enables flexible inversion for semantic-preserving edits, few-step sampling regimes suffer from poor forward process approximation, leading to degraded editing quality. Existing few-step inversion methods often rely on pretrained generators and auxiliary modules, limiting scalability and generalization across different architectures. To address these limitations, we propose BiFM (Bidirectional Flow Matching), a unified framework that jointly learns generation and inversion within a single model. BiFM directly estimates average velocity fields in both image $\to$ noise" and noise $\to$ image" directions, constrained by a shared instantaneous velocity field derived from either predefined schedules or pretrained multi-step diffusion models. Additionally, BiFM introduces a novel training strategy using continuous time-interval supervision, stabilized by a bidirectional consistency objective and a lightweight time-interval embedding. This bidirectional formulation also enables one-step inversion and can integrate seamlessly into popular diffusion and flow matching backbones. Across diverse image editing and generation tasks, BiFM consistently outperforms existing few-step approaches, achieving superior performance and editability.

Figures & tables

Figure 1: Inversion-Based Image Editing. (a) In training-free inversion, the process is approximated by numerically reversing the generation steps, leading to accumulated approximation errors; (b) An auxiliary inversion network is introduced on top of a pretrained generator, improving fidelity but increasing complexity and reducing generalization across architectures. (c) Our method, BiFM, jointly learns generation and inversion within a single flow matching model, enabling consistent few-step inversion and editing.

Figure 2: Overview of BiFM. (a) Our one-step generation architecture built upon MMDiT based flow matching model. (b) A single MMDiT block showing how time embedding modulation impact model output. (c) Naive DDIM inversion reuses DDIM update in reverse time, causing departures from original ODE trajectory in few-step regime. (d) Tuning based inversion introduces an auxiliary network $\Phi({\mathbf{x}}_{t^{\prime}},t^{\prime},t)$ (e) BiFM inversion (ours) learns a physically constrained bidirectional average velocity field.

Settings	Methods	NFE	Background Preservation				CLIP Semantics
Settings	Methods	NFE	LPIPS ${}_{\times 10^{3}}\downarrow$	SSIM ${}_{\%}\uparrow$	PSNR $\uparrow$	MSE ${}_{\times 10^{4}}\downarrow$	Whole $\uparrow$	Edited $\uparrow$
Multi-Step	NT Inv [ 24 ] (CVPR23)	50	60.67	84.11	27.03	35.86	23.61	21.64
	MasaCtrl [ 3 ] (ICCV23)	50	106.62	79.67	22.17	86.97	23.96	21.16
	PnP Inv [ 15 ] (ICLR24)	50	49.25	84.86	27.22	32.87	25.83	22.39
	EditFT [ 38 ] (CVPR25)	30	80.55	91.50	26.62	40.24	25.74	22.27
	FlowEdit [ 17 ] (ICCV25)	28	112.19	83.08	21.96	94.99	25.25	22.58

Table 1: PIE-Bench Image Editing Performance. We compare with baselines under different sampling budgets with evaluation metrics from PIE-Bench [ 15 ] . More baselines can be found in Appendix.

Method	MSE ${}_{\times 10^{4}}^{\downarrow}$	LPIPS ${}_{\times 10^{3}}^{\downarrow}$	SSIM ${}_{\%}^{\uparrow}$	PSNR $\uparrow$
DDIM [ 29 ] (ICLR21)	224.43	210.84	70.96	17.76
NT Inv [ 24 ] (CVPR23)	97.64	102.25	82.33	27.82
PnP Inv [ 15 ] (ICLR24)	105.66	95.95	87.20	28.79
RF-Solver [ 34 ] (ICML25)	94.80	106.15	86.36	28.26
BiFM (ours)	87.72	89.49	88.03	30.32

Table 2: Image Reconstruction Performance . We use 50 inversion steps to generate results for all methods. BiFM’s learned inversion process greatly reduces reconstruction error.

Panel of Figure 3 — Figure 3: Inversion and Reconstruction Quality. From left to right: original input image, PnP Inversion [ 15 ] , RF-Edit [ 34 ] , and BiFM (ours) . BiFM faithfully reconstructs image details, while RF-Edit exhibits semantic shift and PnP Inv fails to recover fine details in the source image.

Panel of Figure 4 — Figure 4: Image Editing Visualization. Given a source image, a source prompt and a target prompt (left illustrates difference between source prompt and target prompt), BiFM generates edits which follow more faithfully the intended concept while better preserving the original layout and fine details than other baselines. For example, BiFM engraves a clear lion pattern on the latte art without distorting the background, swaps the Statue of Liberty’s torch for a flower without geometry distortion, and maintains the lighthouse structure.

Table 3: Ablation Study on 1-NFE Image Generation. FID computed from 50K samples is reported. Defaults are shown in bottom row.

Method	Model	FID $\downarrow$
U-Net [ 27 ]	Diffusion	7.32
U-ViT-S/2 [ 1 ]	Diffusion	5.95
U-ViT-S/2 (Deep) [ 1 ]	Diffusion	5.48
MMDiT	Flow Matching	6.05
MMDiT+REPA	Flow Matching	4.73
MMDiT+MeanFlow	Flow Matching	5.02

Table 4: MSCOCO-256 Text-to-Image Generation. FID computed from 40K samples is reported.

Setting	Model	FID $\downarrow$	NFE
Multi-Step	DDIM [ 29 ]	4.67	50
	Flow Matching [ 20 ]	2.63	50
	BiFM (ours)	2.17	50
One-Step	Rectified Flow [ 21 ]	4.85	1
	sCT [ 22 ]	2.85	1
	MeanFlow [ 11 ]	2.92	1

Table 5: Unconditional CIFAR-10 Results. We include both multi-step and one-step FID.

Model	Epochs	FID $\downarrow$
SiT-B/2	80	33.0
SiT-B/2 + BiFM	80	27.8
SiT-L/2	80	18.8
SiT-L/2 + BiFM	80	16.4
SiT-XL/2	80	17.2
SiT-XL/2 + BiFM	80	15.5

Table 6: ImageNet Performance across Model Size. Under multi-step sampling, BiFM training also consistently improves performance across model sizes.

Panel of Figure 11 — Figure A : CIFAR-10 Training Epochs vs. FID

Settings	Methods	Steps	Background Preservation				CLIP Semantics
Settings	Methods	Steps	LPIPS ${}_{\times 10^{3}}\downarrow$	SSIM ${}_{\%}\uparrow$	PSNR $\uparrow$	MSE ${}_{\times 10^{4}}\downarrow$	Whole $\uparrow$	Edited $\uparrow$
Multi-Step	DNAEdit	28	112.60	83.69	23.24	67.32	28.90	23.66
	FlowEdit	28	112.19	83.08	21.96	94.99	25.25	22.58
	EditFT	30	80.55	91.50	26.62	40.24	25.74	22.27
	BiFM (ours)	50	47.01	87.50	29.89	42.66	27.42	24.66
Few-Step	FireFlow	18	-	82.49	23.03	-	26.02	22.81

Table A: Additional PIE-Bench Image Editing Performance.

Table B: Image Generation Results. NFE=50.

Panel of Figure 14 — Figure B : Image Editing Visualization.

Figure C : MSCOCO-256 Text-to-Image Generation Visualization Results. Model trained for 100K iterations.

Panel of Figure 16 — Figure D : CIFAR-10 Generation Visualization Results. Model trained for 500 epochs.

Panel of Figure 17 — Figure E : ImageNet-32 Generation Visualization Results. Model trained for 80 epochs.

Flow Matching

Flow matching learns continuous-time vector fields that transport simple noise distributions into data for efficient generative modeling.

See all 656 papers

SteerFlow: Steering Rectified Flows for Faithful Inversion-Based Image Editing
Apr 2, 2026Thinh Dao, Zhen Wang, Kien T. Pham +1Text-Guided Image Editing Flow-Based Generative Modeling
The Hong Kong University of Science and Technology, Clear Water Bay, Hong Kong · VinUniversity, Ha Noi, Viet Nam
Perceptual Flow Matching for Few-Step Generative Modeling
Jul 3, 2026Chuyang Zhao, Yifei Song, Hongfa Wang +7Flow Matching Perceptual Loss
1Joy Future Academy · 2Fudan University · 3Tsinghua University +1
DirectEdit: Step-Level Accurate Inversion for Flow-Based Image Editing
May 4, 2026Desong Yang, Mang YeFlow Matching Text-Guided Image Editing
National Engineering Research Center for Multimedia Software, School of Computer Science, Wuhan University, Wuhan, China.

Apr 2, 2026J/K move · Enter open · S save

SteerFlow: Steering Rectified Flows for Faithful Inversion-Based Image Editing

Thinh Dao, Zhen Wang, Kien T. Pham, Long Chen

The Hong Kong University of Science and Technology, Clear Water Bay, Hong Kong · VinUniversity, Ha Noi, Viet Nam

Text-Guided Image Editing Flow-Based Generative Modeling Rectified Flow

Open paper AskPDF arXiv

Recent advances in flow-based generative models have enabled training-free, text-guided image editing by inverting an image into its latent noise and regenerating it under a new target conditional guidance. However, existing methods struggle to preserve source fidelity: higher-order solvers incur additional model inferences, truncated inversion constrains editability, and feature injection methods lack architectural transferability. To address these limitations, we propose SteerFlow, a model-agnostic editing framework with strong theoretical guarantees on source fidelity. In the forward process, we introduce an Amortized Fixed-Point Solver that implicitly straightens the forward trajectory by enforcing velocity consistency across consecutive timesteps, yielding a high-fidelity inverted latent. In the backward process, we introduce Trajectory Interpolation, which adaptively blends target-editing and source-reconstruction velocities to keep the editing trajectory anchored to the source. To further improve background preservation, we introduce an Adaptive Masking mechanism that spatially constrains the editing signal with concept-guided segmentation and source-target velocity differences. Extensive experiments on FLUX.1-dev and Stable Diffusion 3.5 Medium demonstrate that SteerFlow consistently achieves better editing quality than existing methods. Finally, we show that SteerFlow extends naturally to a complex multi-turn editing paradigm without accumulating drift.

Abstract

Flow Matching

Explore similar work

SteerFlow: Steering Rectified Flows for Faithful Inversion-Based Image Editing

Perceptual Flow Matching for Few-Step Generative Modeling

DirectEdit: Step-Level Accurate Inversion for Flow-Based Image Editing