Video EditingBeginner

How to Make a Text-to-Video Clip in Runway Gen-3

Turn a written prompt into a short cinematic clip in Runway Gen-3 Alpha, then download it for editing.

7 minBeginner

Runway's Gen-3 Alpha can build a short clip from nothing but a text description. The model reads your prompt, picks a look and a camera move, and renders a few seconds of video. The quality you get depends almost entirely on how you write the prompt, so this guide focuses on getting a clean first clip and the prompt structure that makes it work.

What you need

  • A Runway account (the free tier includes a small pool of credits)
  • A clear idea of the shot you want: subject, setting, lighting, camera move
  • A browser; nothing to install
  • About 5 minutes plus render time

Step 1: Open the Gen-3 generator

Sign in at runwayml.com and go to your dashboard. Click Try Generative Video, then make sure the model selector at the top of the panel is set to Gen-3 Alpha. The left side holds your prompt box and settings; the right side is the preview and your render history.

Runway - Generative Video
Model: [ Gen-3 Alpha v ] Aspect: 16:9 Duration: 5s
------------------------------------------------------------
Prompt:
| A slow dolly-in on a foggy pine forest at dawn...
[ Image input ] [ Settings ] [ Generate ]
The Gen-3 prompt panel with the model selector at the top.

Step 2: Write a structured prompt

Gen-3 responds best to prompts written in three parts: the camera move, the scene, and the lighting or mood. Keep it to one or two sentences and avoid stacking too many actions. Describe one continuous moment rather than a sequence of events.

prompt.txt
A slow dolly-in shot through a foggy pine forest at dawn,
shafts of golden light cutting between the trees,
soft volumetric haze, cinematic, shallow depth of field.
Lead with the camera
Starting your prompt with a camera instruction such as 'slow dolly-in', 'static wide shot', or 'handheld tracking shot' gives you far more control over motion than describing the subject first.

Step 3: Set duration and aspect ratio

Pick 5 or 10 seconds and choose an aspect ratio that matches where the clip will go: 16:9 for landscape, 9:16 for vertical social posts. Longer clips cost more credits and give the model more room to drift off-prompt, so start at 5 seconds while you test wording.

Step 4: Generate and review

Click Generate. A render takes roughly one to three minutes depending on load. When it finishes, play it back and look for two things: did the camera move the way you asked, and did the subject stay coherent. If the motion is wrong, fix the camera phrase first before touching anything else.

Runway - Render result
[ > 0:00 / 0:05 ] forest_dawn_gen3.mp4
------------------------------------------------------------
[ Download ] [ Extend ] [ New variation ] [ Edit ]
Credits used: 50 Resolution: 1280x768
Finished clip with download and variation controls.

Step 5: Download the clip

Hit Download to save the MP4. Runway exports at 720p on the standard plans; you can upscale separately if you need more resolution. Keep the original file plus your exact prompt text in a notes file so you can reproduce or refine the look later.

Result

You now have a five second forest clip rendered from a single prompt, ready to drop into a timeline. The same three-part structure, camera then scene then lighting, will carry over to every other text-to-video tool you try.

Watch related tutorials

Tags
#runway#text-to-video#gen-3#prompting