Descript’s Background Removal

There are 2 ways to remove the background from a video in Descript.  Chroma key – removing a single color (i.e., green) from the video. This is perfect when you can record your subject in front of a green screen or some other solid color background.  But, recording in front of a solid color background like a green screen isn’t always practical or even possible.

 

That’s where Descript’s other background removal tool comes in handy.  It’s called green screen – and that’s a terrible name for the tool because it doesn’t depend on a green screen or any solid color background.  Instead, Descript’s green screen (I cringe calling it that!) uses AI to identify the subject of the video and then remove everything else. 

 

Suppose you’re recording a video of yourself at your kitchen table but don’t necessarily want your refrigerator and kitchen counter in the background.  That’s when you need a smarter tool… to eliminate everything in the background even though it’s not all the same color.  Then, you can substitute a background image or background video and put yourself, or whoever your subject is, on the beach, in a library, or at a news desk. 

 

Using this tool in Descript is literally two clicks.  Just make sure you’ve selected the video clip you want to work on and then, from the properties panel over on the right side, click the + icon next to Effects and then click Green Screen.  That’s it.  There’s nothing to adjust and no settings to fiddle with.  Just give it a minute or two (depending on how long your video is) and the background is gone. 

 

Is It Any Good?

Descript’s background removal (again, they call it green screen) is good, but it’s not perfect. In my experience it does a decent job identifying the subject and eliminating the background when I haven’t put any thought or effort into making that task easy.

 

For instance, when the video is me sitting in my high-back chair talking to the camera, Descript does very well at eliminating the things behind the chair. Sometimes it seems to think the chair itself is part of the subject and keep it in the video. In it’s defense, this happens most when I’m wearing a shirt that isn’t different enough from my black chair given the lighting situation I have going on.

 

It wouldn’t be the end of the world, at least for my needs, if it left the chair in. Instead, what I end up with is the chair developing transparent spots here and there throughout the video when I turn in certain directions. That’s distracting for the viewer, which is the entire reason I want to swap the background in the first.

 

There are some things that help Descript isolate the subject from the background – and they both involve contrast.

 

The obvious change is to wear a shirt color that is very different than my black chair. That definitely helped when I tested it by wearing a lighter blue, red, or even gray shirt against the black chair. But, I got tired of having to do a wardrobe change when it was time to record. Solution: I put a green t-shirt on the back of my chair. Since I rarely wear green, this works great because it creates an obvious contrast between me (the subject) and the chair (part of the background.)

 

The second thing that really helps is lighting. I know, lighting is a real pain for normal people (like me) who want a good looking video but don’t have the patience, time, or money to set up the 25 kinds of lights that photographers and videographers use.

 

I didn’t do all that. All I want to do is brighten up the places where my shoulders (part of the subject) meet the chair (the background) so the AI robots can see the distinction. I opened the blind on the window in front of me and I have 2 other lights in front but off to the sides. (think 10 and 2 o’clock.) That did the trick and made it much more likely to get the desired background removal effect.

Background Removal in Descript [Video]

Leave a Reply

Your email address will not be published. Required fields are marked *

/SinglePost

Other creator tools you might like...

A Google Labs experiment, Whisk is an AI image generation platform that lets you use reference images for subject, scene, and style - whisk will try to capture the essence of those elements, but not the details.
AI Image, video and sound generation platform with editing and upscaling tools and AI-generated stock media.
Create, translate and personalize video at Hollywood standards with LipDub AI's proprietary AI lip sync video generator.
DreamFace's "Avatar Video" is a capable AI lip sync tool, when a source video of the subject (not a still image) is used. Popular for its ease of use and pricing.
Image & video generation platform based on models by Bytedance, with an AI lip sync tool that syncs mouth movements, facial expressions, and natural gestures
AI lip sync tool that animates a subject from a photo or video, syncing mouth movements to uploaded or generated audio. Adapts to the speaker’s vocal style for more expressive results.