AdaEdit: Adaptive Temporal and Channel Modulation for Flow-Based Image Editing

Guandong Li; Zhaobin Chu

cs.CVMar 23, 2026

AdaEdit: Adaptive Temporal and Channel Modulation for Flow-Based Image Editing

Authors: Guandong Li, Zhaobin Chu

Organizations: iFLYTEK

Abstract

Inversion-based image editing in flow matching models has emerged as a powerful paradigm for training-free, text-guided image manipulation. A central challenge in this paradigm is the injection dilemma: injecting source features during denoising preserves the background of the original image but simultaneously suppresses the model's ability to synthesize edited content. Existing methods address this with fixed injection strategies -- binary on/off temporal schedules, uniform spatial mixing ratios, and channel-agnostic latent perturbation -- that ignore the inherently heterogeneous nature of injection demand across both the temporal and channel dimensions. In this paper, we present AdaEdit, a training-free adaptive editing framework that resolves this dilemma through two complementary innovations. First, we propose a Progressive Injection Schedule that replaces hard binary cutoffs with continuous decay functions (sigmoid, cosine, or linear), enabling a smooth transition from source-feature preservation to target-feature generation and eliminating feature discontinuity artifacts. Second, we introduce Channel-Selective Latent Perturbation, which estimates per-channel importance based on the distributional gap between the inverted and random latents and applies differentiated perturbation strengths accordingly -- strongly perturbing edit-relevant channels while preserving structure-encoding channels. Extensive experiments on the PIE-Bench benchmark (700 images, 10 editing types) demonstrate that AdaEdit achieves an 8.7% reduction in LPIPS, a 2.6% improvement in SSIM, and a 2.3% improvement in PSNR over strong baselines, while maintaining competitive CLIP similarity. AdaEdit is fully plug-and-play and compatible with multiple ODE solvers including Euler, RF-Solver, and FireFlow. Code is available at https://github.com/leeguandong/AdaEdit

Figures & tables

Figure 1: Comparison of injection schedule functions. The Progressive Injection Schedule (sigmoid, cosine, linear) provides smooth decay from full injection to zero, eliminating the discontinuity artifact of the binary schedule.

Figure 2: Pipeline of AdaEdit. Our method consists of three phases: (1) Inversion with feature caching, where source attention features $K_{s}^{l},V_{s}^{l}$ are cached and an editing mask $M$ is extracted; (2) Channel-Selective Latent Perturbation, which estimates per-channel importance and applies differentiated AdaIN strengths—edit-relevant channels receive strong perturbation while structure channels are preserved; (3) Sampling with Progressive Injection, where cached source features are mixed with target features using a smoothly decaying weight $w(t)$ . Sub-diagrams show (b) the progressive sigmoid schedule vs. binary cutoff, and (c) channel-selective perturbation strengths.

Method	LPIPS $\downarrow$	SSIM $\uparrow$	PSNR $\uparrow$	CLIP $\uparrow$
ProEdit [ 17 ]	0.2960	0.7244	19.13	0.2617
AdaEdit (Ours)	0.2703	0.7433	19.58	0.2593
Improvement	-8.7%	+2.6%	+2.3%	-0.9%

Table 1: Main results on PIE-Bench (700 images). Best results in bold .

Figure 3: Main results comparison on PIE-Bench (700 images). AdaEdit achieves significant improvements in background preservation metrics (LPIPS, SSIM, PSNR) with minimal impact on editing quality (CLIP).

Method	Model	Struct. Dist. $\downarrow$ ( $\times 10^{3}$ )	PSNR $\uparrow$	SSIM $\uparrow$ ( $\times 10^{2}$ )	CLIP Whole $\uparrow$	CLIP Edited $\uparrow$
P2P [ 4 ]	Diffusion	69.43	17.87	71.14	25.01	22.44
PnP [ 20 ]	Diffusion	28.22	22.28	79.05	25.41	22.55
PnP-Inv [ 9 ]	Diffusion	24.29	22.46	79.68	25.41	22.62
RF-Solver [ 21 ]	Flow	31.10	22.90	81.90	26.00	22.88
FireFlow [ 2 ]	Flow	28.30	23.28	82.82	25.98	22.94
UniEdit [ 8 ] ( $\alpha{=}0.6$ )	Flow	10.14	29.54	90.42	25.80	22.33

Table 2: Comparison with additional baselines on PIE-Bench. Results for P2P through FireFlow+ProEdit are from [ 17 ] . ProEdit and AdaEdit rows show our re-evaluation using the same protocol.

Edit Type	LPIPS $\downarrow$		CLIP $\uparrow$
Edit Type	ProEdit	Ours	ProEdit	Ours
Random	0.295	0.271	0.258	0.257
Change Obj.	0.280	0.246	0.260	0.260
Add Obj.	0.152	0.123	0.259	0.256
Delete Obj.	0.307	0.284	0.248	0.248
Change Attr.	0.267	0.245	0.259	0.260

Table 3: Per-type results on PIE-Bench (700 images).

Figure 4: Per-type LPIPS comparison. AdaEdit consistently improves background preservation across all 10 editing types, with the largest gains on spatially localized edits.

Figure 5: Radar chart showing SSIM scores across all editing types. AdaEdit (orange) consistently outperforms ProEdit (blue) across all categories.

Configuration	LPIPS $\downarrow$	SSIM $\uparrow$	PSNR $\uparrow$	CLIP $\uparrow$
Baseline	0.297	0.729	20.16	0.267
+ Prog. Sched.	0.259	0.766	20.91	0.268
+ Soft Mask	0.174	0.812	23.33	0.251
+ Adapt. KV	0.285	0.739	20.61	0.266
+ Chan. LS	0.298	0.727	20.13	0.268

Table 4: Ablation of individual components (20 samples).

Figure 6: Ablation study of individual components. Progressive Schedule provides the best single-component trade-off between background preservation (LPIPS) and editing quality (CLIP).

Configuration	LPIPS $\downarrow$	SSIM $\uparrow$	PSNR $\uparrow$	CLIP $\uparrow$
SM only ( $\gamma{=}15$ )	0.222	0.788	22.60	0.265
Sched.+SM+CLS	0.139	0.840	24.78	0.237
Sched.+CLS (AdaEdit)	0.260	0.765	20.90	0.269

Table 5: Ablation of component combinations (20 samples).

$\gamma$	LPIPS $\downarrow$	SSIM $\uparrow$	PSNR $\uparrow$	CLIP $\uparrow$
5	0.174	0.812	23.33	0.251
15	0.222	0.788	22.60	0.265

Table 6: Effect of soft mask sharpness

\gamma

(20 samples).

Figure 7: Preservation-editing trade-off analysis. AdaEdit (orange star) achieves an optimal balance, while aggressive configurations like Soft Mask sacrifice editing quality for preservation.

Text-Guided Image Editing

Text-guided image editing changes selected image content from natural-language instructions while aiming to preserve everything else.

See all 165 papers

Sep 28, 2026

DecFlowEdit: Self-Localized Flow-based Image Editing via Guidance Decoupling

Flow-based image editing (FlowEdit) enables inversion-free semantic changes through the difference between source and target velocities. In this paper, we observe that FlowEdit's default classifier-free guidance (CFG) configuration, with asymmetric source and target scales, causes substantial background leakage. Matching these guidance scales, for example by removing CFG, improves edit-relevant localization but severely degrades editability. To get the best of both worlds, we propose DecFlowEdit, which decouples the optimal guidance scales for localization and for editing in flow-based generative models. In particular, DecFlowEdit first extracts an edit-relevant prior by temporally aggregating velocity differences evaluated without CFG, and then uses this prior to reweight the original updates under default CFG. Our method remains training-free and inversion-free, requiring neither external spatial masks nor attention manipulation. Experiments on PIE-Bench across FLUX, SD3, and SD3.5 show that DecFlowEdit improves background preservation, reducing structure distance by approximately 61 to 73 percent and background LPIPS by 68 to 80 percent relative to FlowEdit at comparable editing fidelity.

Zheyuan Zhan, Can Wang, Jiawei Chen +4

Zhejiang University · University at Buffalo · University of California, Berkeley

Flow Matching Classifier-Free Guidance Text-Guided Image Editing Flow-Based Generative Modeling Image Editing

Ask PDF

Apr 2, 2026

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

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.

Thinh Dao, Zhen Wang, Kien T. Pham +1

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

Ask PDF

Jul 12, 2026

h-Flow: Flexible Flow-based Image Editing via Doob's h-Transform

Editing images with pre-trained text-to-image flow models typically requires carefully balancing target alignment with the desired prompt and source consistency with the original image. Existing approaches either rely on inversion-based pipelines or heuristic source-to-target trajectory constructions, which often depend on architecture-specific designs or are sensitive to hyperparameters. In this paper, we propose h-Flow, a training-free and theoretically grounded flow-based editing framework. Inspired by Doob's

h

-Transform, we reformulate image editing as conditional generation under multiple terminal events corresponding to source consistency and target alignment. We first extend the classical

h

-Transform from SDE-based models to the deterministic RF framework by constructing an equivalent SDE with identical marginals. Within this formulation, we design dedicated

h

-functions for source consistency and target alignment, yielding closed-form reconstruction guidance and velocity-based semantic editing signals. We further introduce a velocity orthogonal decomposition to decouple reconstruction and editing directions, enabling a controllable trade-off between the two objectives. Extensive experiments demonstrate that h-Flow achieves effective, robust, and flexible editing across diverse scenarios. The code will be released soon.

Zehui Guo, Zhen Wang, Junwei Shu +3

East China Normal University, Shanghai, China · The Hong Kong University of Science and Technology, Hong Kong SAR, China

Doob H-Transform Text-Guided Image Editing Image Editing Training-Free Image Editing

Ask PDF

Abstract

Text-Guided Image Editing

Explore similar work

DecFlowEdit: Self-Localized Flow-based Image Editing via Guidance Decoupling

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

h-Flow: Flexible Flow-based Image Editing via Doob's h-Transform