Miller's Rig

Make a video

Prompt

25 and 30 second clips are experiments. Longer clips need more GPU memory, so if one fails, drop back a step.

Start image Reference pictures (pick as many as you like) Audio (a song or a voice track) Cut the audio into clips of the length above and make every clip in one run Use the last frame of each clip as the start of the next one

More controls

Steps

Frames per second

Seed (blank = random)

Things to avoid

Load a video model on the Models tab first.

Progress

Finished clips

On the GPU right now

Power on the instance to see this.

Pick a model

Model

Loading a model for the first time downloads it, which can take several minutes. After that it stays on the instance's disk.

Talk to it

Load a chatbot on the Models tab to start. For voice, also load a speech-to-text and a text-to-speech helper.

Read replies out loud Hands-free: keep listening after each reply

Personality and instructions

System prompt You are a clever, warm, a little wry conversation partner and a strong coder. Keep spoken answers short and natural. When you write code, put it in code blocks.

Start a new chat