How to use Runway: current guide for creators

Runbo Li
Runbo Li
·
· 4 min read
Runway

You want to make a video in Runway, but an older tutorial may send you to controls that are no longer the starting point. Tools, Apps, Agent and Workflows are different ways to begin, and the model still needs inputs that fit the shot.

This guide helps you choose the right surface, write one filmable scene and check settings before generating. You will then inspect the complete clip and revise the largest mismatch first. Start with a small project you can review, rather than treating a dashboard tour or one attractive frame as a finished video.

Quick answer

Choose Runway Tools for direct controls, Apps for guided tasks, Agent for a conversational workflow or Workflows for connected nodes. For a first clip, select a current video model, provide its required inputs, set duration and ratio, then review the full result.

1. Choose Tools, Apps, Agent or Workflows

Runway’s current getting-started guide describes four generation modes. Choose Tools when you need explicit model and setting control; Apps for a named use case; Agent when you want the system to plan and generate through conversation; or Workflows for a repeatable node-based pipeline.

For a first test, use Tools. It makes the source, model, prompt and settings easiest to record and compare.

2. Choose text-to-video or image-to-video

  • Text-to-video: describe the visible subject, action, setting, camera and treatment.

  • Image-to-video: upload a permitted starting image and focus the prompt on desired motion.

  • Existing-footage edit: use Edit Studio when the job is to change a supplied clip rather than generate a new one.

3. Select the exact current model

Runway’s Gen-4.5 guide documents text-to-video and image-to-video controls, supported durations, aspect ratios, frame rates and credit use. Availability and limits depend on the selected model and plan, so read the model card shown in the interface instead of carrying settings from an older model.

Watch Runway Gen-4.5 examples

Runway’s official Gen-4.5 introduction illustrates the model’s intended creative uses. The clips are vendor-selected demonstrations, not our quality or speed benchmark.

Source: Runway: Introducing Gen-4.5 | Runway

4. Write one filmable shot

Keep the first prompt to one shot. For text-to-video, define what appears and how it moves. For image-to-video, let the image establish appearance and describe the motion. Avoid multiple scene changes or instructions that cannot fit the selected duration.

5. Review settings and credit use

Confirm aspect ratio, duration, frame rate, resolution, output format and the estimate before generating. Use the current Runway pricing page for plans, then rely on the interface estimate for the exact model and settings. A historical monthly price table cannot represent every current mode.

6. Generate and inspect the complete clip

Review subject identity, anatomy, object and product geometry, text, camera behavior, scene continuity, motion and audio where applicable. A convincing thumbnail or first frame does not prove the full clip is usable.

7. Iterate from the largest mismatch

Change one meaningful input at a time: source image, action, camera, setting, treatment or hard setting. Keep the original output and record what changed. This makes the next result informative even when it is not accepted.

How to edit existing footage in Runway

Runway’s current Edit Studio guide describes an Aleph 2.0 workflow for prompted changes to uploaded or generated footage. Upload a supported clip, select a representative keyframe, describe the change, preview the keyframe edit and then apply it to the video or a selected range.

Inspect the result beyond the edited region. Product swaps, character changes, backgrounds, relighting and restyles can affect edges, identity, geometry or surrounding details. Preserve the source and every accepted version.

A practical first project

Runway changes quickly. This guide uses Runway’s current documentation checked September 13, 2026; older tutorials built around Gen-2, Green Screen or a “Video Projects” dashboard no longer describe the main starting workflow.

  • Write the brief. One subject, one action, one setting and one camera instruction.

  • Define pass/fail checks. Name required identity, product, text, motion and output properties.

  • Generate a short proof. Use the lowest practical duration and quality that still exposes the hard part.

  • Review and revise once. Change the largest mismatch and keep both versions.

  • Finish only after acceptance. Upscale, extend or edit the selected result, then review the final downloaded file.

When to compare Runway with another platform

Compare alternatives when the required model, input type, output, cost or connected workflow is unavailable or inefficient for the job. Our AI video generator comparison separates platforms from underlying models and lists current decision criteria.

Magic Hour’s AI Video Generator is another browser workflow for testing current video models and connected image, video and audio tools. Run the same permitted brief and compare accepted output rather than vendor claims.

One adult subject walks naturally toward the camera in soft afternoon light. Medium tracking shot, stable identity and clothing, realistic motion, one continuous scene, no readable text or logos.

Compare the same shot in Magic Hour

Use the same permitted source, shot brief and acceptance criteria. Compare accepted output, retries, manual fixes and total cost before choosing a production workflow.

Open AI Video Generator

Frequently asked questions

Runway currently lists a Free plan, but model access, credits, watermarks and exports vary. Check the live pricing page and the estimate for the exact workflow before promising a deliverable.

Use Tools for a controlled first generation because the model, input and settings are explicit. Move to Apps, Agent or Workflows when the job benefits from a guided or multi-step process.

Yes. Edit Studio accepts an existing clip and uses a selected keyframe plus a prompt to guide an Aleph 2.0 edit. Confirm the current input limits and review the complete output.

Build and approve short shots, then assemble or continue them with the current editing workflow. Do not ask one generation to exceed the selected model’s supported duration.

Runbo Li
Runbo Li
CEO of Magic Hour
Runbo Li is the Co-founder and CEO of Magic Hour, where he builds AI video and image tools for content creation. He is a Y Combinator W24 founder and former Data Scientist at Meta, where he worked on 0-1 consumer social products in New Product Experimentation. He writes about AI video generation, AI image creation, creative workflows, and creator tools.
View author →

Continue Reading

Runway ML pricing guide: plans, credits and API costs, illustrated with a video editing workstation

Runway ML free plan: video quality, 720p, credits & pricing

Runway ML free plan: video quality, 720p, credits & pricing
How to use FaceFusion 3.9.0: setup, settings and results guide, with a blue face illustration

FaceFusion 3.9.0 guide: official install, settings & fixes

FaceFusion 3.9.0 guide: official install, settings & fixes
Flux <> MH

FLUX.1 Kontext tutorial: edit images, prompts & FLUX.2

FLUX.1 Kontext tutorial: edit images, prompts & FLUX.2
Handmade collage showing a portrait becoming an animated character video with audio

Hedra AI guide: current focus and earlier Character 3 workflow

Hedra AI guide: current focus and earlier Character 3 workflow
pika

How to use Pika AI (2027): video, PikaSwaps, audio & API

How to use Pika AI (2027): video, PikaSwaps, audio & API
midjourney

How to use Midjourney to create AI art

How to use Midjourney to create AI art