Make your next creation.
Turn your ideas into 3D assets. Start creating with Hyper3D.
Train a Pen-Spinning Policy with the Sharpa Dexterous Hand
The author says this Chinese prompt was provided to GPT-6 Astra to use Isaac Lab for reinforcement learning on the Sharpa dexterous hand, create a 3D mesh of the pen from scratch, and deliver the trained policy and a visualization video.
Download unavailable. Open the original video to save it.
0:00 / 0:00
Video playback is unavailable. Open original video
Prompt
The author says this Chinese prompt was provided to GPT-6 Astra to use Isaac Lab for reinforcement learning on the Sharpa dexterous hand, create a 3D mesh of the pen from scratch, and deliver the trained policy and a visualization video.
Keywords
- GPT-6 Astra
- AI Will
Follow us
Prompt optimization tips
1
Specify editability
State which objects must remain separate, named and editable; otherwise a visually convincing shell may still be unusable.
2
Close the test loop
Ask the model to run the scene, inspect visible failures and repeat until concrete frame-rate, interaction and layout checks pass.
3
Stage long builds
For large worlds, split the work into districts, assemblies or milestones and approve each section before expanding scope.
More ideas to explore
All prompts
GPT-6.1 Sol
Soda Bottle Rocket Launch
A kitchen soda bottle launches like a miniature Starship. Sol 6.1 was asked to scale real flight speed and acceleration to the bottle and keep everyday objects visible for scale.
GPT-6.1 Sol / Sonnet 5.5 / GPT-6 Astra
Minion Turntable in Blender
EvoLink compares GPT-6.1 Sol, Sonnet 5.5 and GPT-6 Astra on a Blender Minion model with a 5-second, square, 360-degree turntable animation. The author supplies the full prompt in a reply.
GPT-6.1 Sol / Opus 5.5
Reference-Matched Boar Animation
A Meshy-rigged boar is refined in Blender against a synchronized video reference over five iterations. The public prompt specifies a 144-frame, 24 fps comparison and detailed gait and contact checks.