AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

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

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

CVPR 2026 · Image generation › Fast diffusion sampling · approach: Flow matching distillation · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
train jointly for generation and inversion (BiFM) vs training-free inversion at 50 stepsStable Diffusion 3 editing · PIE-Bench, 50 NFELPIPS x10^3: 49.25 -> 47.01 (vs PnP Inv)Improved
BiFM few-step inversion at 4 steps vs strongest baseline InstantEditStable Diffusion 3 editing · PIE-Bench, 4 stepsLPIPS x10^3: 44.39 -> 67.25Hurt
BiFM vs SwiftEdit in one-step editingone-step editing · PIE-Bench, 1 stepSSIM %: 81.05 -> 85.88 SSIM; LPIPS 91.04 -> 92.30Mixed
learned bidirectional inversion vs baseline inversion methodsimage reconstruction · PIE-Bench reconstructionPSNR: 28.79 -> 30.32 (vs PnP Inv)Improved
add BiFM training to MMDiTMMDiT · MSCOCO-256FID: 6.05 -> 4.57Improved
add BiFM training to SiT modelsSiT-B/2, L/2, XL/2 · ImageNet-256, 80 epochsFID: 33.0 -> 27.8; 18.8 -> 16.4; 17.2 -> 15.5Improved
BiFM vs MeanFlow for one-step generationCIFAR-10 U-Net · CIFAR-10, 1 NFEFID: 2.92 -> 2.75Improved
condition on (t, t'-t) instead of (t, t') (ablation)BiFM 1-NFE · 1-NFE image generationFID: 59.37 -> 55.22Improved
add discrete direction flag to time input (ablation)BiFM 1-NFE · 1-NFE image generationFID: 55.22 -> 69.01Hurt
use warm-up weighting on consistency term instead of linear (ablation)BiFM 1-NFE · 1-NFE image generationFID: 67.37 -> 55.22Improved
sample (t,t') from lognorm(-0.4, 1.0) instead of uniform (ablation)BiFM 1-NFE · 1-NFE image generationFID: 63.25 -> 55.22Improved
use robust loss norm (p=1.0 default, p=0.5) instead of squared L2 (ablation)BiFM 1-NFE · 1-NFE image generationFID: 59.66 -> 55.22Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 3 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy