CharGen: Fast and Fluent Portrait Modification
Fuente:
arXiv
Salvato in:
| Autori principali: | , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866912615196786688 |
|---|---|
| author | Dihlmann, Jan-Niklas Killguss, Arnela Lensch, Hendrik P. A. |
| author_facet | Dihlmann, Jan-Niklas Killguss, Arnela Lensch, Hendrik P. A. |
| contents | Interactive editing of character images with diffusion models remains challenging due to the inherent trade-off between fine-grained control, generation speed, and visual fidelity. We introduce CharGen, a character-focused editor that combines attribute-specific Concept Sliders, trained to isolate and manipulate attributes such as facial feature size, expression, and decoration with the StreamDiffusion sampling pipeline for more interactive performance. To counteract the loss of detail that often accompanies accelerated sampling, we propose a lightweight Repair Step that reinstates fine textures without compromising structural consistency. Throughout extensive ablation studies and in comparison to open-source InstructPix2Pix and closed-source Google Gemini, and a comprehensive user study, CharGen achieves two-to-four-fold faster edit turnaround with precise editing control and identity-consistent results. Project page: https://chargen.jdihlmann.com/ |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_25058 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | CharGen: Fast and Fluent Portrait Modification Dihlmann, Jan-Niklas Killguss, Arnela Lensch, Hendrik P. A. Graphics Computer Vision and Pattern Recognition Interactive editing of character images with diffusion models remains challenging due to the inherent trade-off between fine-grained control, generation speed, and visual fidelity. We introduce CharGen, a character-focused editor that combines attribute-specific Concept Sliders, trained to isolate and manipulate attributes such as facial feature size, expression, and decoration with the StreamDiffusion sampling pipeline for more interactive performance. To counteract the loss of detail that often accompanies accelerated sampling, we propose a lightweight Repair Step that reinstates fine textures without compromising structural consistency. Throughout extensive ablation studies and in comparison to open-source InstructPix2Pix and closed-source Google Gemini, and a comprehensive user study, CharGen achieves two-to-four-fold faster edit turnaround with precise editing control and identity-consistent results. Project page: https://chargen.jdihlmann.com/ |
| title | CharGen: Fast and Fluent Portrait Modification |
| topic | Graphics Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2509.25058 |