Salvato in:
| Autori principali: | , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2509.06499 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866918143253807104 |
|---|---|
| author | Lin, Jibai Ma, Bo Yang, Yating Zhou, Xi Ma, Rong Osman, Turghun Ahmat, Ahtamjan Dong, Rui Wang, Lei |
| author_facet | Lin, Jibai Ma, Bo Yang, Yating Zhou, Xi Ma, Rong Osman, Turghun Ahmat, Ahtamjan Dong, Rui Wang, Lei |
| contents | Subject-driven image generation (SDIG) aims to manipulate specific subjects within images while adhering to textual instructions, a task crucial for advancing text-to-image diffusion models. SDIG requires reconciling the tension between maintaining subject identity and complying with dynamic edit instructions, a challenge inadequately addressed by existing methods. In this paper, we introduce the Target-Instructed Diffusion Enhancing (TIDE) framework, which resolves this tension through target supervision and preference learning without test-time fine-tuning. TIDE pioneers target-supervised triplet alignment, modelling subject adaptation dynamics using a (reference image, instruction, target images) triplet. This approach leverages the Direct Subject Diffusion (DSD) objective, training the model with paired "winning" (balanced preservation-compliance) and "losing" (distorted) targets, systematically generated and evaluated via quantitative metrics. This enables implicit reward modelling for optimal preservation-compliance balance. Experimental results on standard benchmarks demonstrate TIDE's superior performance in generating subject-faithful outputs while maintaining instruction compliance, outperforming baseline methods across multiple quantitative metrics. TIDE's versatility is further evidenced by its successful application to diverse tasks, including structural-conditioned generation, image-to-image generation, and text-image interpolation. Our code is available at https://github.com/KomJay520/TIDE. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_06499 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | TIDE: Achieving Balanced Subject-Driven Image Generation via Target-Instructed Diffusion Enhancement Lin, Jibai Ma, Bo Yang, Yating Zhou, Xi Ma, Rong Osman, Turghun Ahmat, Ahtamjan Dong, Rui Wang, Lei Computer Vision and Pattern Recognition Subject-driven image generation (SDIG) aims to manipulate specific subjects within images while adhering to textual instructions, a task crucial for advancing text-to-image diffusion models. SDIG requires reconciling the tension between maintaining subject identity and complying with dynamic edit instructions, a challenge inadequately addressed by existing methods. In this paper, we introduce the Target-Instructed Diffusion Enhancing (TIDE) framework, which resolves this tension through target supervision and preference learning without test-time fine-tuning. TIDE pioneers target-supervised triplet alignment, modelling subject adaptation dynamics using a (reference image, instruction, target images) triplet. This approach leverages the Direct Subject Diffusion (DSD) objective, training the model with paired "winning" (balanced preservation-compliance) and "losing" (distorted) targets, systematically generated and evaluated via quantitative metrics. This enables implicit reward modelling for optimal preservation-compliance balance. Experimental results on standard benchmarks demonstrate TIDE's superior performance in generating subject-faithful outputs while maintaining instruction compliance, outperforming baseline methods across multiple quantitative metrics. TIDE's versatility is further evidenced by its successful application to diverse tasks, including structural-conditioned generation, image-to-image generation, and text-image interpolation. Our code is available at https://github.com/KomJay520/TIDE. |
| title | TIDE: Achieving Balanced Subject-Driven Image Generation via Target-Instructed Diffusion Enhancement |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2509.06499 |