From Text to Terrain: Prompting an FPV Glacier Flight

RUNWAY GEN-4 TURBO

“Continuous hyperspeed FPV footage: The camera seamlessly flies through a glacial canyon to a dreamy cloudscape.”

This prompt was originally used to show off what Runway GEN-3 could do when it launched last year. With GEN-3, you could generate a full video clip using text alone — just type in a prompt, and you’d get a moving image.

Things have changed with GEN-4. Now, video generation requires a still image as your starting point. Instead of going straight from words to video, you now build from a visual foundation — which opens up new creative possibilities, but also changes your workflow.

ChatGPT Image Generation

To prep for the GEN-4 video, I used the same prompt from the GEN-3 demo — but this time, I asked Jen (aka ChatGPT) to generate a cinematic still image as my input frame.

That image became the launchpad for the 5-second FPV-style hyperspeed canyon flight. This hybrid workflow — using ChatGPT for the image, then Runway GEN-4 for animation — lets you create something uniquely tailored at every step.

Final Thoughts

GEN-4 might require more upfront effort than GEN-3, but it puts more creative control in your hands. You’re no longer just typing and hoping. Instead, you’re designing your shots visually and directing your camera with purpose — one image at a time.

If you’re experimenting with AI video workflows, this is a great time to start combining the strengths of different tools. Let ChatGPT handle the visuals. Let Runway handle the motion. And let your creativity do the rest.