If artificial intelligences can generate images, imitate voices, hold a conversation, and write code, sooner or later they will cross swords with video. That's how we arrive at QuickVid, a new platform developed by Daniel Habib to create TikTok videos, YouTube, Instagram, and Snapchat videos with minimal user intervention. The platform's availability is limited, and it still needs to improve in quality... but time is on its side.
There is no middle ground: creating videos on TikTok is a casual and natural process, or something that demands several hours of preparation and editing. We can definitely talk about "professionals" focused on applying content strategies for TikTok, keeping up with trends, maximizing the reach of their clips, and of course, growing their follower count. But what happens when we add artificial intelligence to the equation?
That possibility isn't exactly new. Google and Meta have explored video generation with Make-A-Video and Imagen Video respectively, but today they qualify as proof of concept. Daniel Habib's platform QuickVid wants to become a viable commercial product, making it easier for users to adopt artificial intelligence to create TikTok, Instagram, and any other social network with video support.
QuickVid, the new tool for creating TikTok videos with voiceovers
QuickVid's developer has suggested that his platform seeks to prioritize those who already produce content for the social network, with the promise of increasing the number of published videos without falling into the dreaded burnout. The current version of QuickVid is relatively basic but easy to use.
The first step is creating a prompt, a description of the topic the video will cover. From that prompt, QuickVid generates a script using the magic of GPT-3. The second phase is selecting a royalty-free video from Pexels (based on keywords in the script, or entered manually by the user), as well as copyright-free music from YouTube, and generates additional images for overlays with DALL-E 2. The voice in the videos is courtesy of Google Cloud's text-to-speech API, but Habib has said users will be able to clone their own voices in the future.
Logically, QuickVid carries very important limitations. Pexels' database reports fewer than 100,000 videos, and the images created by DALL-E 2 are far from works of art. To finish, QuickVid will have to face two enormous challenges: copyright claims, and malicious use to spread false information, something TikTok is already struggling with today. Want to try it? The official link is below, but you'll need to join a waitlist.
Official site: Click here
Source: TechCrunch
