Amuse AI 3.7.0: Windows setup and RTX generation tests

Hugo Paredes walks through Amuse AI 3.7.0 on Windows, using its installer and built-in model downloads to set up local AI generation. He also identifies a ZIP download as the portable option. The application prompts him to create a stable-diffusion.cpp environment for images and a Python environment for music, rather than configure those environments manually.

For images, he selects Z Image Turbo with automatic memory settings and 8-bit precision. His RTX 3070 with 8 GB of VRAM produces a 1024 × 1024 image in a reported 30 seconds, followed by another in 27 seconds. Image-to-image edits change a kitten's instrument and a robot's appearance, taking 23 and 24 seconds respectively.

The LTX 2.3 video tests show why the settings matter when interpreting these results. The desktop run takes 49 minutes and 46 seconds at 40 steps and guidance 6, and Paredes dislikes the output. A laptop with an RTX 3050 and 6 GB of VRAM takes 13 minutes and 35 seconds at 8 steps and guidance 1. He prefers that result, but the different settings prevent a direct GPU performance comparison.

The final demonstration uses ACE Step 1.5 Turbo for text-to-music. Paredes reports generating a one-minute instrumental track in six seconds. These timings describe his particular tests; the tutorial does not establish general hardware requirements or expected performance for every model.