Player not loading? Watch on YouTube
This tutorial tests Amuse AI 3.5.1, which the presenter describes as a Windows-only preview release for generating media on Nvidia and AMD GPUs. The source describes the app as free and open source. The demonstration uses an RTX 5060 Ti; Intel GPU support remains uncertain in the speaker's account.
Setup starts with the GitHub releases page. Amuse AI handles model component downloads through its interface and prompts users to create a Python environment. The presenter finds this easier than ComfyUI's manual component placement, though the revised interface removes the earlier easy mode and exposes more settings.
The image test uses a four-billion-parameter model with standard precision, a resolution setting of 768 and four steps. The presenter reports a fast result with readable large text but inaccurate small lettering. Memory presets and model-specific controls give users ways to adjust local AI generation, although the tutorial mostly uses defaults.
The music test initially returns an error without lyrics. Adding lyrics obtained through ChatGPT lets generation proceed, with the presenter reporting roughly five seconds of processing. Audio then keeps playing after a menu change until the item is deleted.
Video generation fares worse: the SkyReels test takes 23 minutes for a four-second clip that fails to resemble the requested yellow car on a bridge. The cause remains unresolved; suggested compatibility and settings issues are possibilities, not established diagnoses.
