Yume-1.5: A Text-Controlled Interactive World Generation Model
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_ | 1866911340117884928 |
|---|---|
| author | Mao, Xiaofeng Li, Zhen Li, Chuanhao Xu, Xiaojie Ying, Kaining He, Tong Pang, Jiangmiao Qiao, Yu Zhang, Kaipeng |
| author_facet | Mao, Xiaofeng Li, Zhen Li, Chuanhao Xu, Xiaojie Ying, Kaining He, Tong Pang, Jiangmiao Qiao, Yu Zhang, Kaipeng |
| contents | Recent approaches have demonstrated the promise of using diffusion models to generate interactive and explorable worlds. However, most of these methods face critical challenges such as excessively large parameter sizes, reliance on lengthy inference steps, and rapidly growing historical context, which severely limit real-time performance and lack text-controlled generation capabilities. To address these challenges, we propose \method, a novel framework designed to generate realistic, interactive, and continuous worlds from a single image or text prompt. \method achieves this through a carefully designed framework that supports keyboard-based exploration of the generated worlds. The framework comprises three core components: (1) a long-video generation framework integrating unified context compression with linear attention; (2) a real-time streaming acceleration strategy powered by bidirectional attention distillation and an enhanced text embedding scheme; (3) a text-controlled method for generating world events. We have provided the codebase in the supplementary material. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2512_22096 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Yume-1.5: A Text-Controlled Interactive World Generation Model Mao, Xiaofeng Li, Zhen Li, Chuanhao Xu, Xiaojie Ying, Kaining He, Tong Pang, Jiangmiao Qiao, Yu Zhang, Kaipeng Computer Vision and Pattern Recognition Recent approaches have demonstrated the promise of using diffusion models to generate interactive and explorable worlds. However, most of these methods face critical challenges such as excessively large parameter sizes, reliance on lengthy inference steps, and rapidly growing historical context, which severely limit real-time performance and lack text-controlled generation capabilities. To address these challenges, we propose \method, a novel framework designed to generate realistic, interactive, and continuous worlds from a single image or text prompt. \method achieves this through a carefully designed framework that supports keyboard-based exploration of the generated worlds. The framework comprises three core components: (1) a long-video generation framework integrating unified context compression with linear attention; (2) a real-time streaming acceleration strategy powered by bidirectional attention distillation and an enhanced text embedding scheme; (3) a text-controlled method for generating world events. We have provided the codebase in the supplementary material. |
| title | Yume-1.5: A Text-Controlled Interactive World Generation Model |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2512.22096 |