DuplexBridge: Coordinating Understanding and Generation for Unified Image Editing

Yidi Liu; Jie Xiao; Chaojie Mao; Jingfeng Zhang; Xueyang Fu

cs.CVOct 13, 2026

DuplexBridge: Coordinating Understanding and Generation for Unified Image Editing

Authors: Yidi Liu, Jie Xiao, Chaojie Mao, Jingfeng Zhang, Xueyang Fu

Organizations: University of Science and Technology of China · Wan Team, Alibaba Group

Abstract

Complex image editing requires faithful task interpretation and accurate execution, but a failed edit does not reveal which responsibility failed. We present DuplexBridge, which uses a unified model's own understanding capability to enhance prompts for its own generation capability, without an external understanding module. A source-grounded bridge specifies targets, final-state changes, and preservation constraints. Understanding is supervised on fidelity to the original request without observing the generated image; generation is supervised on execution of the condition it actually receives. Reciprocal trust checks selectively admit global task feedback to each role while retaining independently valid local supervision. We introduce CE-Bench, a 1,200-task benchmark measuring joint satisfaction of task-derived change and preservation requirements. Across three model configurations, the full system improves complete compliance over direct editors by 21.75-25.75 percentage points, with gains also observed on KRIS-Bench, RISEBench, ImgEdit, and GEdit-Bench. From a common supervised Wan-Image initialization, gated joint training reaches 77.08% complete compliance, compared with 70.92% without gates and 75.00% when generation-local supervision instead checks the original task with both gates retained. These results support distinct local responsibilities and selective shared feedback for joint understanding-generation adaptation.

Figures & tables

Figure 1: DuplexBridge overview. (a) Task failures with different local responsibilities. (b) Native U–G communication and three distinct checks. (c) CE-Bench coverage and full-system CCR gains, including PE, SFT, and RL. Primary sectors encode 300/240/240/240/180 tasks; outer subdivisions illustrate capabilities rather than subtype frequencies.

Figure 2: From understanding to a checkable bridge. U resolves the clock task and separates the editable analog clock from protected digital and scene content. Only the public condition enters G and the execution compiler; change and invariant checks use that same condition.

Figure 3: Bidirectional partner-gated feedback. Execution trust gates U-global; prompt trust gates G-global. Local feedback retains independent validity masks. The diagram abbreviates $e_{U}^{\rm glob},e_{G}^{\rm glob}$ as $e_{U},e_{G}$ and local rewards as $r_{U}^{\rm loc},r_{G}^{\rm loc}$ . The routing matrix assumes usable judgments. $Y_{r}$ is a baseline edit, distinct from task-provided reference images.

Primary category	$N$	Secondary categories
A. Source-Grounded	300	A1 Instance / Ordinal Grounding; A2 Spatial / Relational Grounding; A3 Attribute-Conditioned Selection; A4 Part / Region Grounding.
B. Text & Structured Content	240	B1 Exact Text Editing; B2 Multi-Text & Text Preservation; B3 Document / UI Layout Editing; B4 Numeric / Symbolic / Structured Visual Editing.
C. Reasoning-Grounded	240	C1 Visual Question-to-Edit; C2 Logical / Spatial / Quantitative Reasoning; C3 Temporal / Causal / Physical Inference; C4 Knowledge / Conceptual / Procedural Reasoning.
D. Coupled & Compositional	240	D1 Multi-Target Composition; D2 Dependent / Coupled Edits; D3 Relational Rearrangement; D4 Mixed Operations + Preservation.
E. Reference-Conditioned Multi-Image	180	E1 Entity / Subject Transfer; E2 Attribute / Appearance Transfer; E3 Pose / Layout / Structural Transfer; E4 Multi-Reference Composition & Reasoning.

Table 1: CE-Bench taxonomy. The five primary categories are mutually exclusive; the secondary categories describe capability coverage.

Model	A	B	C	D	E	AC $\uparrow$	Pres. $\uparrow$	CCR $\uparrow$
Commercial Models
GPT-Image-1.5	83.00	78.33	72.92	82.50	68.89	93.47	94.12	77.83
Gemini-3-pro-image-preview	82.33	77.08	81.67	84.17	72.78	94.08	93.87	80.08
Seedream 4.5	80.33	77.50	62.92	72.50	58.89	91.36	94.43	71.50
Gemini-2.5-Flash-Image	73.00	69.58	64.17	70.83	59.44	89.67	90.72	68.08
Generation-only Editors

Table 2: CE-Bench comparison. A–E are category CCR ( table 1 ); scores are percentages. Full systems include native PE, structured SFT, and gated RL. Dark/light pink marks the best/second-best values within each model group, including ties.

Method	KRIS $\uparrow$	RISE $\uparrow$	ImgEdit $\uparrow$	GEdit EN $\uparrow$
PromptEnhancerV2	71.54	18.90	4.18	7.59
DARS	80.72	27.50	4.39	7.86
BAGEL	56.21	5.80	3.45	6.40
+ DuplexBridge	76.48	26.11	4.04	7.41
Qwen-2511	64.42	17.50	4.36	6.97
+ DuplexBridge	82.16	31.11	4.43	7.89

Table 3: Public editing benchmarks. Higher is better; metric scales follow section 5 . Dark/light pink marks the best/second-best value per column among displayed systems.

Method	Temporal	Causal	Spatial	Logical	Overall
BAGEL (B0)	14.12	23.33	23.00	5.88	16.94
BAGEL + DuplexBridge	24.71	33.33	31.00	14.12	26.11
Qwen-2511 (B0)	21.18	26.67	29.00	9.41	21.94
Qwen-2511 + DuplexBridge	31.76	38.89	34.00	18.82	31.11
Wan-Image-7B (B0)	16.47	23.33	24.00	7.06	18.06
Wan-Image-7B + DuplexBridge	29.41	35.56	34.00	16.47	29.17

Table 4: Native UMM configurations on RISEBench-360. Qwen-2511 denotes Qwen-Image-Edit-2511. Overall is task-weighted across 85 temporal, 90 causal, 100 spatial, and 85 logical items. Dark/light pink mark the best/second-best value in each column, including ties.

Figure 4: Qualitative comparison on CE-Bench. Rows cover sign translation, image-grounded question answering, geometric reasoning, spatially specified slicing, and composition from multiple references. The geometry example requires an answer within the existing diagram; the final row combines entity placement with preservation of the base scene.

Model	A	B	C	D	E	AC	Pres.	CCR
BAGEL (B0)	68.67	49.58	47.92	62.92	47.78	83.91	88.63	56.42
Qwen-2511 (B0)	73.00	61.25	55.00	64.58	51.11	86.62	90.14	62.08
Wan-Image-7B (B0)	70.00	57.92	52.50	62.08	49.44	85.28	89.21	59.42

Table 5: Supervised initialization controls.

B_{0}

uses native PE and structured SFT, before RL. A–E follow table 1 ; public metrics follow section 5.4 . Dark/light pink mark the best/second-best values, including ties.

Interface	CE AC	CE CCR
Fact-matched prose	84.38	57.00
Flat key-value	84.16	56.67
Flat JSON	84.55	57.42
Grouped key-value	85.09	59.00
CE-v1 grouped JSON	85.28	59.42
CE-v1 without redundant goal	85.17	59.17

Table 6: Wan-Image interface ablations at SFT. Rows compare prose, flat and grouped structures, and CE-v1 variants without the redundant goal or with shuffled fields. AC and CCR are percentages.

Construction	Fidelity	Recall	Dup.	Human agr.	Retrained CCR
Same-contract prose rubric	95.11	88.73	4.28	87.14	74.92
Same-contract atomic reconstruction	95.46	89.12	3.17	87.36	75.58
Compiled execution checks	98.21	91.68	0.73	87.92	77.08

Table 7: Wan-Image execution-check construction. Fidelity, task recall, and duplicates use all rubric attempts. Human agreement uses an adjudicated content-matched subset. The final column reports CCR after retraining with each construction method.

Variant	U/G RL	CE AC	CE CCR	Pres.	KRIS	RISE	ImgEdit	GEdit EN
Direct, no PE	–/–	82.70	53.83	88.47	60.84	13.06	4.09	6.88
Native PE	–/–	83.86	56.67	88.92	63.41	15.83	4.11	6.97
$B_{0}$ : matched SFT	–/–	85.28	59.42	89.21	66.82	18.06	4.16	7.09
U-only RL	$\checkmark$ /–	87.96	66.67	88.74	73.28	24.44	4.17	7.21
G-only RL	–/ $\checkmark$	88.61	67.50	92.28	71.94	21.94	4.31	7.52
Dual-role local-only	$\checkmark$ / $\checkmark$	–	68.25	–	–	–	–	–

Table 8: Wan-Image-7B training-direction ablation. All RL variants start at

B_{0}

with common supervision and reference access. Native PE omits CE-v1 SFT; U-only and G-only are trained independently. UT/DARS-style rows transplant mechanisms into Wan-Image. Dashes in the local-only row denote unreported metrics. GEdit EN uses Full-set

G_{O}

Condition	CCR $\uparrow$	Full minus control
Count-matched random gate	71.33	5.75
Group-structure-matched random gate	72.25	4.83
Original-task G-local source	75.00	2.08
Full: semantic gates + received-condition G-local	77.08	—

Table 9: Additional mechanism endpoints on Wan-Image. CCR is a percentage; the final column gives full minus control (percentage points). Count-matched and group-structure-matched random gates are separate controls. The G-local comparison changes the reference for local supervision from the original task to the received condition. These endpoints compare training variants.

Gate	CE CCR	KRIS	RISE
Neither global gate	70.92	76.27	26.11
Prompt trust only	74.17	78.62	28.06
Execution trust only	73.33	78.14	27.50
Both global gates	77.08	80.13	29.17
Count-matched random	71.33	76.48	26.11
Soft semantic weighting	76.33	79.76	28.89

Table 10: Wan-Image global-reward gate ablations. The first four rows vary the two semantic trust decisions, with local supervision and reward-validity conditions retained. Oracle gates use extra annotations.

Interface	Ungated	Gated
Flat	67.42	70.75
Structured	70.92	77.08
Interaction: $+2.83$ points

Table 11: Structure

\times

global gating on Wan-Image. CE CCR (%) with matched task facts, local checks, initialization, data, and training budget.

Gate	G-global cov.	Wrong / G-global	U-global cov.	Unexec. / U-global
Neither semantic gate	93.60	18.59	91.40	27.41
Prompt trust only	71.20	2.90	91.40	27.41
Execution trust only	93.60	18.59	64.80	4.60
Both trust gates	71.20	2.90	64.80	4.60
Count-matched random	71.20	18.12	64.80	27.11

Table 12: Global-feedback selection on a frozen pool of 6,000 rollouts. Coverage uses all rollouts; error rates use the corresponding global-eligible subset. Wrong-prompt and unexecuted-prompt labels combine external-VLM judgments and human annotation.

Pair	CE CCR	KRIS	RISE
U0 G0	59.42	66.82	18.06
Ut G0	65.92	72.48	22.22
U0 Gt	68.83	73.94	24.17
Ut Gt	77.08	80.13	29.17

Table 13: Wan-Image checkpoint cross-pairing. U0/G0 are the common B0 initialization; Ut/Gt are the adapted checkpoints from the same full training run.

Figure 5: Requirement load and joint change–preservation outcomes. (a) CCR for full and ungated on the same 1,200 public tasks, grouped by their observed total number

K

of change and preservation requirements; each stratum’s task count is shown below its tick. (b) Counts and percentages of the same tasks in the four joint states (ungated above, full below), derived from task-level change and invariant failure counts.

C=1

means every requested change passes, and

P=1

means every preservation requirement passes;

P

is an all-pass indicator, distinct from the mean preservation fraction reported as Pres. The fraction of tasks in the

C=P=1

state equals CCR.

$K$	$N$	U CCR	F CCR	$\Delta$	95% interval	U AC	F AC
3	300	85.00	90.33	+5.33	[+2.33, +8.33]	95.00	96.78
5	360	74.44	80.83	+6.39	[+3.33, +9.44]	89.78	92.33
7	360	66.39	74.44	+8.06	[+4.72, +11.39]	88.53	92.58
9	180	49.44	52.78	+3.33	[-0.56, +7.22]	83.15	84.26
All	1200	70.92	77.08	+6.17	[+4.50, +7.83]	89.72	92.31

Table 14: Complete and average compliance by public requirement load. U/F denote ungated/full Wan-Image; CCR and AC are percentages.

\Delta

is full minus ungated CCR in points, calculated before rounding. Pointwise paired bootstrap intervals follow the protocol in section 6.7 .

RM	HPA $\uparrow$	Hard $\uparrow$	NLL $\downarrow$	Brier $\downarrow$	SwapErr $\downarrow$	A/B gap $\downarrow$	Latency $\downarrow$
M0: Base direct	77.6	66.5	0.489	0.158	0.058	2.2	$1.0\times$
M1: Explicit think	82.3	75.8	0.377	0.120	0.045	1.3	$6.0\times$
M2: Preference SFT	78.9	69.2	0.439	0.142	0.051	1.0	$1.0\times$
M3: Decision–reason	79.9	71.2	0.409	0.133	0.046	0.8	$1.0\times$
M4: Hard KD	80.4	72.3	0.501	0.154	0.062	1.6	$1.0\times$
M5: Soft KD	81.7	74.8	0.385	0.122	0.041	0.6	$1.0\times$

Table 15: Reward-model judgment quality and cost. HPA/Hard are accuracies (%); A/B gap is in percentage points. NLL/Brier/SwapErr follow equation 17 . Latency is one end-to-end pairwise call relative to M0; only M1 reasons online.

Category	M0	M1	M4	M5
Wrong target or ordinal	70.0	78.3	75.8	78.3
Near-miss text	60.0	72.5	68.3	70.8
Preservation damage	69.2	76.7	73.3	78.3
Coupled-edit omission	61.7	73.3	68.3	70.8
Visually good, task wrong	71.7	78.3	75.8	75.8
Reported overall	66.5	75.8	72.3	74.8

Table 16: Difficult-pair accuracy (%). Scores are shown for five error categories and the reported overall.

RM	MMRB2 edit pair acc. $\uparrow$	EditReward $K=2$ $\uparrow$	EditReward $K=3$ $\uparrow$	EditReward $K=4$ $\uparrow$
M0	66.2	57.2	35.9	10.8
M1	71.8	61.8	44.9	16.6
M2	67.1	58.3	37.8	11.9
M3	68.6	59.6	40.4	13.3
M4	69.3	60.1	41.5	13.8
M5	70.6	61.0	43.5	15.8

Table 17: External RM evaluation. MMRB2 editing pair accuracy and EditReward-Bench

K=2,3,4

ranking scores are reported separately (%).

ID / RM	Gates	CCR $\uparrow$	AC $\uparrow$	Pres. $\uparrow$
D0 / M0	Off	69.08	88.91	91.35
D1 / M0	On	74.25	91.28	93.12
D2 / M5	Off	70.92	89.72	91.82
D3 / M5	On	77.08	92.31	93.86
D4 / M1	On	77.50	92.42	93.73

Table 18: Reward

\times

gate comparisons on Wan-Image. D0–D3 form the

2\times 2

comparison; D4 uses explicit thinking. D0–D4 use equal update steps and rollout budgets. All metrics are percentages.

Method	$N$	Human CCR	Auto. CCR	Bias	Agree.	FA	FR	Human CCR bounds
Direct	195	53.33	54.36	+1.03	88.72	13.19	9.62	[52.0, 54.5]
Ungated	196	69.39	70.92	+1.53	91.33	16.67	5.15	[68.0, 70.0]
Full	197	76.14	76.65	+0.51	93.40	14.89	4.00	[75.0, 76.5]

Table 19: Final-output human evaluation on 200 CE-Bench tasks.

N

is the number of human-decidable outputs per method. CCR, agreement, false acceptance (FA), and false rejection (FR) are percentages; bias is automatic minus human CCR in points. FA conditions on human failures and FR on human successes. Identification bounds assign unresolved labels to failure or success over all 200 tasks.

Quantity	Estimate	95% interval
Human gain $\Delta_{H}$	+6.70	[+1.03, +12.37]
Automatic gain $\Delta_{A}$	+6.19	[+1.55, +10.82]
Gain bias $\Delta_{A}-\Delta_{H}$	-0.52	[-3.61, +2.58]

Table 20: Full versus ungated on 194 common tasks. Point estimates and 95% percentile intervals are in percentage points. Intervals use 50,000 paired task bootstrap resamples within categories, preserving the observed category counts of this common decidable set.

Figure 6: Human preferences for full versus ungated. Each dimension has 197 decidable task pairs and three unresolved pairs out of 200. Wins, ties, and losses retain their separate categories; interval estimates for net preference appear in table 21 .

Full comparison	Dimension	W/T/L	Unres.	Net	95% interval
vs. ungated	Instruction	65/102/30	3	+17.77	[+8.63, +26.90]
vs. ungated	Preservation	38/122/37	3	+0.51	[-8.12, +9.14]
vs. ungated	Quality	44/105/48	3	-2.03	[-11.68, +7.11]
vs. direct	Instruction	89/83/25	3	+32.49	[+22.84, +42.13]
vs. direct	Preservation	70/94/33	3	+18.78	[+9.14, +28.43]
vs. direct	Quality	61/87/48	4	+6.63	[-3.57, +16.84]

Table 21: Human preference for full. W/T/L are wins/ties/losses; Unres. counts unresolved judgments out of 200 task pairs per comparison. Net preference is 100(W-L)/(W+T+L), in points. The 95% intervals use category-stratified task bootstrap; ties remain in the denominator.

Model	GenEval2 AM $\uparrow$	GenEval2 GM $\uparrow$	R2I $\times 100$ $\uparrow$	WISE-V $\times 100$ $\uparrow$	DPG $\uparrow$
BAGEL + native CoT	70.9	23.1	48.0	62.8	85.1
BAGEL + DuplexBridge-T2I	77.8	31.5	56.0	70.5	86.8
Qwen-Image	80.8	33.8	52.0	51.0	88.3
Qwen-Image + DuplexBridge-T2I	85.0	42.0	61.0	66.0	89.7
Wan-Image-7B	79.4	32.2	51.0	54.0	87.7
Wan-Image-7B + DuplexBridge-T2I	84.2	40.6	59.5	64.5	89.3

Table 22: Text-to-image adaptation after retraining. Metrics follow section 6.10 ; dark/light pink mark the best/second-best values across all six systems, including ties.

Figure 9: Additional CE-Bench showcase. Arrows indicate the transition from source images to edited outputs; auxiliary images provide visual references for flower replacement and dog insertion. Examples include sign-text replacement and translation, localized color changes, a traffic-light state change, and object appearance edits. The paired images show the edited regions alongside the surrounding scene.

Image Editing

Image editing changes selected visual content while preserving the parts of an image that should remain unchanged.

See all 313 papers

Uni-Edit: Intelligent Editing Is A General Task For Unified Model Tuning
May 20, 2026Dian Zheng, Manyuan Zhang, Hongyu Li +4Text-Guided Image Editing Unified Multimodal Models
CUHK MMLab · Meituan · TJU +1
STBridge: Shared-Target Alignment for Bridging Understanding and Generation in UMMs
Jul 19, 2026Ye Wang, Hongjun Wang, Hao Fang +7Unified Multimodal Models Cross-Modal Alignment
Jilin University · Tencent YouTu Lab
Instruction-based Image Editing with Planning, Reasoning, and Generation
Feb 26, 2026Liya Ji, Chenyang Qi, Qifeng ChenText-Guided Image Editing Diffusion-Based Image Editing
HKUST

May 20, 2026J/K move · Enter open · S save

Uni-Edit: Intelligent Editing Is A General Task For Unified Model Tuning

Dian Zheng, Manyuan Zhang, Hongyu Li, Hongbo Liu+3

CUHK MMLab · Meituan · TJU+1

Text-Guided Image Editing Unified Multimodal Models Synthetic Instruction Generation Image Editing Multimodal Instruction Tuning

Open paper AskPDF arXiv

Currently, enhancing Unified Multimodal Models (UMMs) with image understanding, generation, and editing capabilities mainly relies on mixed multi-task training. Due to inherent task conflicts, such strategy requires complex multi-stage pipelines, massive data mixing, and balancing tricks, merely resulting in a performance trade-off rather than true mutual reinforcement. To break this paradigm, we propose Uni-Edit, an intelligent image editing task that serves as the first general task for UMM tuning. Unlike complex mixed pipelines, Uni-Edit improves performance across all three abilities at once using only one task, one training stage, and one dataset. Specifically, we first identify image editing as an inherently ideal general task, as it naturally demands both visual understanding and generation. However, existing editing data relies on simplistic instructions that severely underutilize a model's understanding capacity. To address this, we introduce the first automated and scalable data synthesis pipeline for intelligent editing, transforming diverse VQA data into complex and effective editing instructions with embedded questions and nested logic. This yields Uni-Edit-148k, pairing diverse reasoning-intensive instructions with high-quality edited images. Extensive experiments on BAGEL and Janus-Pro demonstrate that tuning solely on Uni-Edit achieves comprehensive enhancements across all three capabilities without any auxiliary operations.

Abstract

Image Editing

Explore similar work

Uni-Edit: Intelligent Editing Is A General Task For Unified Model Tuning

STBridge: Shared-Target Alignment for Bridging Understanding and Generation in UMMs

Instruction-based Image Editing with Planning, Reasoning, and Generation