§ ÎÐjïãó$—ddlmZmZmZmZd„ZdS)é)Ú VIDEO_PROMPTÚVIDEO_PROMPT_SHORTÚ help_textÚ specialitiescóÆ—|dk}d}d}|rd}d}|dkr |dz }|dz }t|rd nd |t|t|rd nd zd g¦«}|rtd¦«ng|d<|S)NÚ longcat_videoa$Generate from text, animate a start image, or continue a source video. **Sliding Window** builds a longer video in successive sections. Keep subject and scene descriptions consistent, and inspect joins before extending further. **Auto** sampling selects the continuation behavior when needed.z—Text/image-to-video and source-video continuation. Sliding windows extend output; inspect joins and drift. Auto sampling chooses continuation behavior.ašAnimate a character using **Voice to Follow** and an optional **Anchor Reference Image** or start frame. The recording supplies speech and timing; the prompt describes the character, expression, gesture and scene. For two speakers, choose the two-voice mode and assign each voice to its visible person. **Sliding Window** supports longer performances; retain consistent anchor appearance and check transitions.zÅAudio-driven character video; audio supplies words/timing, prompt visual behavior. Optional image anchor/start frame; assign two voices to their speakers. Sliding windows support long performances.Úlongcat_avatar_v1_5zZ Avatar 1.5 uses the eight-step distilled preset. Keep its matching sampler and settings.z/ Avatar 1.5: use its eight-step Distill preset.zLongCat Avatarz LongCat VideozV The driving recording determines dialogue; do not substitute new words in the prompt.Úz0https://github.com/meituan-longcat/LongCat-Videozlip syncr)rrrr)Ú architectureÚavatarÚinfoÚcompactÚresults úÑØ €MóN)Úmodels.help_sharedrrrrr©rrúrsBðàXÐXÐXÐXÐXÐXÐXÐXÐXÐXÐXÐXð ð ð ð ð r