Runway ML for people who have never edited a video in their life

I made a video of my dog turning into a cartoon wolf last Tuesday. The whole thing took eleven minutes. I have never opened Premiere Pro, Final Cut, or any video editing software in my life — unless you count cropping an Instagram Story, which you shouldn't.

That's the thing about Runway ML that nobody really explains well. It's not video editing software with AI bolted on. It's closer to a toy box where you throw in an image or a text prompt and something weird happens. Sometimes useful weird. Sometimes "why does my dog have six legs now" weird.

What Runway Actually Is (And What It Isn't)

Runway calls itself a creative toolkit, which is vague enough to mean nothing. Here's what it actually is: a browser-based platform where you can generate video clips from text prompts, animate still images, remove backgrounds from videos, and do a bunch of other AI tricks that used to require knowing what a keyframe was.

What it isn't: a place to edit your vacation footage into a montage. If you have 47 clips from your trip to Portugal and want to arrange them with music and transitions, Runway is not your tool. It generates and transforms. It doesn't organize or sequence.

I went in thinking it would be like Canva for video. Wrong. It's more like a weird lab where you experiment and occasionally get something you can actually use.

The interface is surprisingly clean for something this experimental. No intimidating timeline. No thirty-seven buttons. You pick a tool from a grid, upload something or type something, and wait. The waiting is the main activity, honestly.

The Part That Actually Worked

Gen-3 Alpha is their text-to-video model, and it's the reason most people show up. You type a prompt like "a golden retriever walking through autumn leaves, cinematic" and it generates a 5-10 second clip.

I typed: "a woman sitting alone at a diner counter at 2am, rain on the window, noir lighting." It took about 90 seconds to generate. The result looked like a frame from a movie I'd actually watch. The lighting was moody. The rain looked real. The woman's face was slightly off — that uncanny valley thing where the proportions are 2% wrong — but from a distance, at a glance, it worked.

The image-to-video feature is where I spent most of my time. You upload a still photo, and it animates it. My dog photo became a three-second clip of him turning his head. Not life-changing, but genuinely surprising the first time.

Background removal on video worked better than I expected. I uploaded a clip of myself talking (yes, I recorded myself talking to test this, don't judge) and it pulled me out of my messy office background in about two minutes. Not perfect edges around my hair, but good enough for something that required zero skill.

The Kick: What Nobody Mentions About Actually Using This

Here's the thing I only figured out after burning through most of my free credits: the prompt structure matters way more than the prompt content.

I kept getting garbage results until I realized you need to describe the shot, not just the scene. "A cat sleeping on a couch" gives you a weird floating cat situation. "Close-up shot of a tabby cat sleeping on a worn leather couch, shallow depth of field, warm afternoon light" gives you something usable.

The model thinks in cinematography terms. It wants to know: what's the camera doing? How far away is it? What's the lighting like? What's the texture of things? When I started writing prompts like a pretentious film student describing a shot list, my results improved dramatically.

Also — and this is annoying — the free tier gives you 125 credits. Generating a single 5-second video costs about 10-15 credits depending on settings. So you get maybe 8-10 real attempts before you're paying. That's not enough to learn what works. I burned my free credits making mistakes, then had to decide if I cared enough to pay.

I did pay for one month ($12). It helped. I wouldn't have understood the tool at all on free tier alone.

What Straight Up Didn't Work

Hands. Anything with hands. I tried to generate "a person writing in a journal at a coffee shop" and got a creature with seven fingers holding a pen like a claw machine grabbing a stuffed animal. This is a known AI thing, but it's still jarring every time.

Text in videos is also a disaster. Wanted a shot of a neon sign saying "OPEN." Got something that looked like "OGFN" in a font that doesn't exist. Don't ask for text. Just don't.

Motion can be unpredictable. I wanted a slow zoom on a landscape. Sometimes it zoomed. Sometimes the entire mountain started morphing like it was breathing. You don't have fine control over movement, and that's frustrating when you have something specific in mind.

The generation time varied wildly too. Same prompt, same settings — sometimes 40 seconds, sometimes three minutes. No explanation. You just wait.

So Is This Actually Useful?

For someone who's never edited video, Runway is genuinely interesting as a thing to play with. You can make stuff that looks surprisingly good for zero technical skill. The barrier to getting a short, weird, AI-generated clip is basically nonexistent now.

But "useful" depends on what you need. Social media b-roll? Maybe. A serious project? Probably not yet. The clips are short, the control is limited, and you're always one generation away from something unusable.

What surprised me most is that it changed how I think about video. I never considered making video content before because the learning curve felt vertical. Now I've made maybe thirty short clips of varying quality, and I understand camera angles and lighting better than I did two weeks ago — just from writing prompts about them.

Whether that's worth twelve bucks a month, I genuinely don't know yet. I'm still deciding if I'll keep paying next month, or if this was just an expensive week of clicking generate and waiting.

Heads up: Some links in this post may be affiliate links. I only recommend tools I've personally tested. Opinions are entirely my own.