YouTube Shorts uses machine learning to make vertical ads from landscape videos

Youtube Shorts
(Image credit: Daniel Bader / Android Central)

What you need to know

  • YouTube is introducing new tools to introduce ads optimized for Shorts.
  • The methods include customized templates and Machine Learning techniques.
  • Using machine learning, current landscape ads dramatically change into vertical video adverts.

YouTube Shorts is a growing means of content consumption in short video format, particularly on smartphones. The platform has generated over 30 billion daily views alongside 1.5 billion monthly logged-in users. It is undeniably an all-in vertical watching experience similar to TikTok, but one problem is optimizing existing YouTube ads for the format.

The user count here is likely enticing to marketers creating video action campaigns since they can generate additional revenue on YouTube Shorts In a new blog post, YouTube has announced that it is introducing new solutions and creative guidelines "to help advertisers of all sizes make effective vertical video ads."

To make this happen, Google is implementing its machine learning prowess that "reformats landscape video ads into square or vertical formats based on how someone is watching YouTube. "

It works by detecting elements from a landscape ad, including faces, key objects, logos, text, and motion. It further breaks the video into scenes, and the ML ensures it is cropped at centered properly, recreating the vertical ad video from the landscape version.

YouTube Video Reformat

(Image credit: Google)

YouTube is also providing four customizable vertical video ad templates that are already available in the video creation tool of Google Ads. These templates are likely to follow YouTube's best practices, which offer a vertical viewing experience, are fast-paced, and include transitions and music. Users can navigate to the Asset Library to try them out.

The company adds that users "can customize these templates to create a vertical or square ad for your brand in minutes."

The machine learning technology creation outlined above is catered to app campaigns first, followed by Video action and Performance Max campaigns. Additionally, a new set of vertical video ad templates can help companies create these types of videos from existing assets based on text and images.

With the tools given, YouTube also suggests tips for creating successful campaigns, like being mindful of pacing and evoking emotion.

The company believes more people are streaming across multiple screens, channels, and different video lengths across streaming devices. In the eyes of the Google-owned video platform, users are coming to YouTube to consume different types of content.

And amongst them, Gen Z viewers are said to be watching a mix of short and long video content. In fact, Google notes how short-form videos help these viewers discover longer videos. This makes a win-win for the video platform, which, to varying degrees, excels in both. This also helps marketers identify the audience they like to share their innovative ad videos with.

YouTube's new machine learning tool is naturally a huge advantage to multiple small businesses with limited resources to create various video formats. YouTube states that video campaigns, including video action, app, and performance action campaigns, deliver 10-20% more conversions per dollar through YouTube Shorts. And adding vertical video assets can help marketers generate their ad content in vertical video format. Now, the problem of adopting their ads for Shorts seems to be solved thanks to ML.

Vishnu Sarangapurkar
News Writer

Vishnu works as a freelance News Writer for Android Central. For the past four years, he's been writing about consumer technology, primarily involving smartphones, laptops, and every other gizmo connected to the Internet. When he is away from keyboard, you can see him going on a long drive or chilling on a couch binge-watching some crime series.