- Rating
- 2(+1)
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 1
Read review
Paper summary
Bind-and-Route targets multi-subject personalized image generation, where increasing reference subjects causes text-reference misalignment and subject confusion. It introduces bidirectional cross-attention resamplers and contrastive binding losses to associate each placeholder with its matched reference. It further proposes a spatial router that uses noise-to-text attention priors to redistribute image-to-reference attention without inference-time masks. Experiments on PSRBench, XVerseBench, qualitative comparisons, ablations, and a small user study show competitive overall performance.
Critical strengths
-
The paper offers a mechanistic decomposition of multi-subject failures. Explicit Reference Binding addresses incorrect text-reference correspondence, while Attention Routing addresses incorrect spatial assignment, making the proposed interventions interpretable and modular.
-
The method targets the four-reference setting, where existing approaches degrade substantially. It improves the PSRBench average and aesthetic scores, while remaining mask-free at inference and outperforming the same-data OminiControl baseline.
Critical weaknesses
-
The Bind contrastive loss does not define the source of and precisely. The paper omits the layer, heads, timestep, and aggregation procedure, leaving a key training signal under-specified.
-
The Bind resampler is not described sufficiently for reproduction. The paper does not clarify its timestep execution or caching, the training timestep schedule, or basic architecture details such as dimensions, head count, FFN configuration, depth, parameter count, and weight sharing.
-
Route is evaluated only through a coarse cumulative ablation. The paper does not isolate the contributions of the text and background priors, offset correction, three loss terms, loss weights, mask thresholds, block and timestep ranges, or text-prior detachment. Therefore, the claimed gains cannot be attributed to specific design choices.
-
The experimental comparison is not fully controlled or fair. The baselines use different backbones and inference step counts, while Bind-and-Route introduces LoRA, bidirectional resamplers, a spatial router, and placeholders with 285M trainable parameters. Without matched parameter budgets, training compute, and adaptation procedures, the reported gains cannot be cleanly attributed to the proposed binding and routing mechanisms.
Questions for authors
See Weakness.
6 comments
Response to reviewer MyB6 (part 2 of 2)
5. Parameter-count ablation. FLUX.1-dev transformer contains about 12B parameters, and our 285M trainable parameters account for 2.38% of its total parameter count. To provide a more capacity-controlled comparison, we additionally evaluate a baseline with rank-95 LoRA, comprising 344M trainable parameters:
| Method | 3-ref Subject | 3-ref Semantic | 3-ref Aesthetic | 3-ref Average | 4-ref Subject | 4-ref Semantic | 4-ref Aesthetic | 4-ref Average |
|---|---|---|---|---|---|---|---|---|
| Baseline | 0.5190 | 0.8638 | 0.9469 | 0.7766 | 0.4026 | 0.8543 | 0.9462 | 0.7344 |
| Baseline, 344M LoRA | 0.5236 | 0.8665 | 0.9878 | 0.7926 | 0.4394 | 0.7319 | 0.7764 | 0.6492 |
| Ours (full) | 0.5275 | 0.8961 | 0.9639 | 0.7958 | 0.4892 | 0.8644 | 0.9879 | 0.7805 |
Increasing the baseline’s trainable parameter count improves its performance on the 3-reference subset compared with our current baseline. However, on the four-reference subset, semantic alignment and aesthetic quality deteriorate substantially, resulting in an Average score below that of the original baseline. The 344M LoRA baseline also underperforms our full method on every reported metric except Aesthetic in the 3-reference setting, despite having more trainable parameters. These results suggest that increasing model capacity alone is insufficient to reproduce the gains achieved by our method in the evaluated settings.
Thank you for the detailed responses and the additional results, which have partly addressed my concerns. However, considering the overall quality of the paper, and since I agree with some of the points raised by Reviewer 3Nna, I have decided to raise my final score to 2.
Response to reviewer MyB6 (part 1 of 2)
Please refer to our revised manuscript and the Summary of Changes in the Revised Manuscript comment. Below is our response to the weaknesses:
1. Attention source and aggregation for the binding contrastive loss. For the token-level attention aggregation AP2R and AR2P defined after Eq. (1) and Eq. (3), we average the attention probabilities over all heads within each block. The bidirectional contrastive loss is computed and averaged across all 19 double-stream and 38 single-stream blocks. There is no block-range or timestep-range restriction for this objective.
2. Binding resampler architecture and execution details. Appendix A specifies that the binding resamplers are applied before the first transformer block at each timestep, while Section 3.1 and Figure 3 describe their bidirectional residual feature injection. Both resamplers use input, internal, and output dimensions of 3072 and consist of one pre-LayerNorm cross-attention layer with eight heads, followed by a pre-LayerNorm FFN, both with residual connections. The FFN has dimensions 3072–12288–3072 and uses GELU activation; all dropout rates are zero. The detailed architecture is shown in Figure 8. The two directions are instantiated independently, while subjects share the mapper within each direction. Each mapper contains 113,298,432 parameters, giving 226,596,864 parameters for the two resamplers together. All trainable modules, including the semantic resampler, are optimized using FLUX. 1-dev’s default timestep sampling schedule and rectified-flow training objective. Resampler outputs are recomputed without caching during both training and inference. We have supplemented Appendix A with numerical configurations and execution details.
3. Ablation study. We have expanded router range selection experiments in Appendix A (Table 4) and ablation studies in Appendix D (Table 7).
Text and background priors. The prior-only variant (Prior only) directly uses the text priors to guide the reference image attention. Compared with the Direct attention mask loss variant, it achieves better aesthetic quality by using text attention to guide reference image attention. However, without the router to further distinguish the regions associated with different subjects, Subject Confusion remains, resulting in lower subject consistency than the full method. This distinction is also illustrated by the mean text prior and mean router alpha visualized in the last two columns of Fig. 9, where the router produces more spatially distinct localized reference-region assignments. The base prior serves as a background branch that absorbs routing probability in background regions. Removing it makes reference image attention more diffuse (Router w/o pbasetext). Although this variant yields a marginal improvement in subject consistency in the 4-reference setting, it reduces semantic alignment and aesthetic quality.
Offset correction. During training, we observe that omitting the offset causes the total reference attention mass after routing to decrease substantially, as illustrated in Fig. 10. Router w/o offset slightly improves semantic alignment but reduces subject consistency, consistent with the decrease in reference attention mass.
Lrouter weights. The ablation of Lrouter coefficients shows that Lce (ce only (0.5/0/0)) improves subject consistency, compared with Router w/o Lrouter, but a large λce=0.6 (ce only (0.6/0/0)) can compromise semantic alignment and aesthetic quality. Adding the negative (ce + neg(0.5/0.1/0)) and margin losses (ce + margin (0.5/0/0.1)) also improves subject consistency.
Router range selection. Following the reviewers’ suggestions, we found that applying the router throughout the entire generation process yielded larger gains. Rather than testing every possible routing range, we used the analysis in Appendix A to select candidate ranges and adopted the best-performing configuration among those tested. An exhaustive search over all hyperparameter settings is infeasible; we therefore select a well-motivated, reasonable configuration rather than seek a globally optimal one.
4. Backbone fairness. Except for MS-Diffusion, which uses SDXL, all compared methods use DiT backbones. Qwen-Image-Edit and OmniGen2 do not use backbones inherently weaker than FLUX.1-dev. All FLUX.1-based methods are evaluated with the same number of denoising steps, while the other methods use their recommended default settings. Thus, the inference settings do not deliberately favor our method.