MiniMax H3 on Magnific: Director-like workflow with 2K video and multimodal references
s_mohinii · x · 2026-08-15
MiniMax H3 is now available on Magnific, featuring a multimodal workflow that integrates text, images, video, and audio. Key capabilities include support for up to 9 images, 3 videos, and 3 audio references, generating 2K videos up to 15 seconds. It offers native synchronized voice, music, and sound effects, along with motion transfer and instruction-based video editing. Users can control scenes using moodboards, characters, and environment references, with sound generated alongside video to create a directing experience rather than simple prompting.
More from Multimodal
- XGRIDS builds world's largest 3DGS town: 23.5 km² and 6.2 billion Gaussian Splats — bilawalsidhu · 2026-10-16
- IMMERSION: AI film compilation links 12 shorts with one continuous camera move — AI_Andrew · 2026-10-16
- A song with vocals, lyrics and 3D head animation created entirely by Opus 5.5 — repligate · 2026-10-16
- Alibaba open-sources CommerceGen-Studio for commercial-grade e-commerce images — alibabagroup · 2026-10-16
- EVA-Cap: Fudan and OPPO align video captioning with atomic audiovisual events via GRPO — Fudan-University · 2026-10-16
- SCOPED: skill-decoupled expert distillation beats mixed-capability RL across 8 benchmarks — Yuqi Xiong · 2026-10-16