CapField-OPD: Learning Continuous Capability Fields via Joint-Anchored Multi-Teacher On-Policy Distillation for Flow Models

Pengyang Ling; Jiazi Bu; Yujie Zhou; Yibin Wang; Zeqiang Lai; Xiaoxiao Ma; Yi Jin; Huaian Chen

cs.CVSep 28, 2026

CapField-OPD: Learning Continuous Capability Fields via Joint-Anchored Multi-Teacher On-Policy Distillation for Flow Models

Authors: Pengyang Ling, Jiazi Bu, Yujie Zhou, Yibin Wang, Zeqiang Lai, Xiaoxiao Ma, Yi Jin, Huaian Chen, +1 more

Organizations: University of Science and Technology of China · Shanghai Jiao Tong University · Fudan University · The Chinese University of Hong Kong · Shanghai Artificial Intelligence Laboratory

Abstract

Reward-specialized post-training produces strong experts for flow-based generative models, while multi-teacher on-policy distillation (OPD) consolidates their capabilities into a single student. Existing methods, however, route each prompt to a single teacher according to its semantic category, implicitly binding the desired capability to prompt content. This coupling makes capability invocation vulnerable to prompt perturbations and prevents users from explicitly adjusting the strength of the desired capability at inference time. In this work, we introduce CapField-OPD, an OPD framework that integrates multiple teachers into a continuous capability field through explicit capability coordinates. We use teacher models as anchors to construct this field, with the coordinates determining how their outputs are combined. Each capability configuration thus receives a unique supervision target, and capability control no longer depends on prompt semantics. Since the training anchors may not be optimal at inference time, we further profile the learned field on a small calibration set. The coordinate with the highest mean reward serves as the recommended default, while coordinates that are frequently optimal offer a promising candidate set for test-time scaling. Extensive experiments on compositional generation, text rendering, and visual aesthetics demonstrate that CapField-OPD consolidates multiple specialized teachers into a single student while preserving or surpassing their performance, reliably invokes the desired capabilities under semantics-preserving prompt variations, and supports continuous capability control and coordinate-based test-time scaling.

Figures & tables

Figure 1: Overview of CapField-OPD. (a) Teacher anchors define a continuous capability field controlled by explicit, potentially extrapolative coordinates. (b-d) The learned field supports continuous control, robust capability invocation under prompt rewriting, and coordinate-based test-time scaling.

Figure 2: Visualization under prompt rewriting.

Method	Composition-task prompts				Text-rendering prompts				Aesthetic prompts
Method	GenEval	HPSv3	CLIP	PickScore	OCR	HPSv3	CLIP	PickScore	HPSv3	CLIP	PickScore
FLUX.1-dev	0.6616	8.65	0.3966	23.40	0.5735	13.00	0.4466	22.91	13.28	0.3868	22.58
GenEval teacher	0.9417	9.16	0.4092	23.27	0.6170	13.43	0.4585	22.97	13.49	0.3946	22.47
OCR teacher	0.7166	8.90	0.4030	23.59	0.9417	12.74	0.4567	22.88	13.57	0.3834	22.64
Aesthetic teacher	0.3228	10.62	0.4117	24.04	0.4824	15.14	0.4688	23.95	15.23	0.4181	23.68
GenEval+Aesthetic teacher	0.9253	11.75	0.4210	24.11	0.6235	14.74	0.4525	23.32	14.65	0.3968	22.71

Table 1: Quantitative comparison. Single-only and Single+Joint distill from the three single-capability teachers without and with the two joint-capability teachers, respectively. The two CapField-OPD modes share one student model and differ only in their inference coordinates. Bold and underlined values denote the best and second-best results among unified models. Since aesthetic prompts contain no composition target or text-rendering target, the joint mode does not apply, and the corresponding entries are marked “–”.

Figure 3: Visual demonstration of different methods, from left to right: FLUX.1-dev, Multi-task GRPO, DiffusionOPD(Single-only), DiffusionOPD(Single+Joint), DanceOPD(Single-only), DanceOPD(Single+Joint), CapField-OPD(Single mode), and CapField-OPD(Joint mode).

Metric	FLUX.1-dev	Multi-task GRPO	DiffusionOPD		DanceOPD		CapField-OPD Single mode
Metric	FLUX.1-dev	Multi-task GRPO	Single	Single+Joint	Single	Single+Joint	CapField-OPD Single mode
GenEval $\uparrow$	0.6550	0.8461	0.6241	0.6132	0.6838	0.5560	0.9267
OCR $\uparrow$	0.6115	0.9242	0.8662	0.8718	0.8879	0.8097	0.9677
HPSv3 $\uparrow$	13.46	14.45	15.13	15.15	15.08	15.17	15.28

Table 2: Robustness evaluation under semantics-preserving prompt rewriting.

Figure 4: Illustration of continuous capability control at inference time.

Figure 5: Demonstration of coordinate-based and seed-based test-time scaling.

Training anchors	Composition + Aesthetics				Text Rendering + Aesthetics
Training anchors	GenEval $\uparrow$	HPSv3 $\uparrow$	CLIP $\uparrow$	PickScore $\uparrow$	OCR $\uparrow$	HPSv3 $\uparrow$	CLIP $\uparrow$	PickScore $\uparrow$
Base + single	0.7723	9.38	0.4034	22.23	0.8716	13.50	0.4481	22.84
Base + single + joint	0.9322	11.76	0.4201	23.94	0.9256	14.08	0.4538	23.32

Table 3: Effect of joint-teacher anchors at jointly activated tasks.

Coordinate selection	Composition		Text Rendering		Aesthetics
Coordinate selection	$\hat{\lambda}^{S}_{\mathrm{peak}}$	GenEval $\uparrow$	$\hat{\lambda}^{S}_{\mathrm{peak}}$	OCR $\uparrow$	$\hat{\lambda}^{S}_{\mathrm{peak}}$	HPSv3 $\uparrow$	CLIP $\uparrow$	PickScore $\uparrow$
Teacher model	/	0.9417	/	0.9417	/	15.23	0.4181	23.68
In-range profiling	(1.0, 0.0, 0.0)	0.9487	(0.0, 1.0, 0.0)	0.9390	(0.0, 0.0, 1.0)	15.25	0.4171	23.70
Expanded profiling	(1.05, 0.0, 0.0)	0.9531	(0.0, 1.5, 0.0)	0.9592	(0.0, 0.0, 1.3)	15.32	0.4182	23.71

Table 4: Effect of coordinate extrapolation. Each

\hat{\lambda}^{S}_{\mathrm{peak}}

(Eq. 11 ) reports the full selected coordinate over (GenEval, OCR, aesthetics), with

S

being the single capability of the corresponding metric.

Appendix figures & tables4 assets

Supplementary material from the paper’s appendix.

Appendix

Parameter	Value	Parameter	Value
Random seed	$42$	Learning rate	$2\!\times\!10^{-4}$
Optimizer	AdamW	Weight decay	$1\!\times\!10^{-4}$
AdamW $\beta_{1},\beta_{2}$	$0.9,\,0.999$	AdamW $\epsilon$	$1\!\times\!10^{-8}$
LR scheduler	Constant	Text length (CLIP / T5)	$77$ / $256$
LoRA rank / $\alpha$	$64$ / $128$	Cond. dimensions	$3\!\rightarrow\!256\!\rightarrow\!3072$
Batch size	$8$	Number of GPUs	$8$

Appendix

Table 5: Hyperparameter settings used for CapField-OPD.

Figure 6: Capability curves during CapField-OPD training.

Figure 7: Additional coordinate sweeps at inference time from GenEval to aesthetics.

Figure 8: Additional coordinate sweeps at inference time from OCR to aesthetics.

Explore similar work

DanceOPD: On-Policy Generative Field Distillation
Jun 25, 2026Wei Zhou, Xiongwei Zhu, Zelin Xu +8Multi-Reference Image Generation Duet
ByteDance Seed · NUS · UMD +1
Poly-OPD: Heterogeneous Multi-Teacher On-Policy Distillation for Capability-Selectable Flow Models
Aug 5, 2026Siming Fu, Haojun Xu, Ruizhe He +9Aware Heterogeneous Multi-Teacher Multimodal On-Policy Distillation Uni-Opd
1Joy Future Academy · 2Zhejiang University · 3Beihang University
FlowMap-OPD: Rollout--Kernel Separation for On-Policy Distillation of Few-Step Flow-Map Generators
Sep 29, 2026Zhiqi Li, Bo ZhuFlow Map Uni-Opd
Georgia Institute of Technology

Jun 25, 2026, cs.CVJ/K move · Enter open · S save

DanceOPD: On-Policy Generative Field Distillation

Wei Zhou, Xiongwei Zhu, Zelin Xu, Bo Dong+7

ByteDance Seed · NUS · UMD+1

Multi-Reference Image Generation Duet Edit

Open paper AskPDF arXiv

Modern image generation demands a single model that unifies diverse capabilities, including text-to-image (T2I), local editing, and global editing. However, these capabilities are rarely naturally aligned and often conflict. For instance, editing tends to degrade T2I performance, while global and local editing interfere with each other. Consequently, effectively composing these capabilities has become a central challenge for image generation model training. To tackle this, we introduce DanceOPD, an on-policy generative field distillation framework for flow-matching models that routes each sample to one capability field, queries one low-noise student-induced state, and trains with a simple velocity MSE objective. With each capability source defined as a velocity field over the shared flow state space, the student learns from fields queried on its own rollout states to compose expert capabilities. This formulation also absorbs operator-defined fields such as classifier-free guidance. Comprehensive experiments on T2I, editing, realism-field absorption, and CFG absorption show that our approach improves multi-capability composition, strengthening target capabilities while preserving anchor generation quality. We believe this work establishes a practical route for generative field distillation in flow-matching models.

Abstract

Explore similar work

DanceOPD: On-Policy Generative Field Distillation

Poly-OPD: Heterogeneous Multi-Teacher On-Policy Distillation for Capability-Selectable Flow Models

FlowMap-OPD: Rollout--Kernel Separation for On-Policy Distillation of Few-Step Flow-Map Generators