The generation of videos through artificial intelligence is taking a new approach with this generative AI, which in turn sources its material from databases with images and clips.
This will also interest you
Take a close look at the following video. We posted it on YouTube, a channel dedicated to artificial intelligence. It’s in English, but the French subtitles display easily.
This is a short clip that tells how the video game market has overtaken the Hollywood market over the years. The essential point is that this 2 minutes and 42 seconds long clip was produced solely from a simple wish: “The video game market has become bigger than Hollywood.” It was enough to enter this simple request for the Invideo application:
- write a suitable video script;
- Look for images and videos that can illustrate this point.
- Make the video you want.
In the free version we use, the video is time-limited, but if you choose a paid version you can produce a 15-minute clip!
This 2 minute and 42 second video was created from a simple question: “The video game market has overtaken Hollywood.” © Invideo.io
Master the art of the “prompt”
If you want your video to have strong originality, like ChatGPT or Midjourney, you can write a long-form prompt with lots of details:
- What is the target audience?
- the clip style (minimalist, professional, energetic, etc.);
- the desired duration;
- the target medium (depending on whether you choose Youtube, Facebook or Linkedin, the approach will be different), etc.
Here, as elsewhere, the more detailed the “prompt” is, the more original the final video will be.
A different approach than the usual generative video AIs
Some may complain about the clips created by default (in the free version). Therefore, in this case, the mention “IStock” appears above certain images. It turns out that Invideo sources its material from such databases of images and clips. In order for such mentions to disappear, you would need to take out a paid subscription (at least $20 per month).
Furthermore, the voices in the current version are in English and it would therefore be necessary to re-record a French voice instead. However, such an application can save valuable time for those who need to manage a thematic channel and update it regularly.
Classic looking videos by default
It should be noted that this approach differs from applications such as Pika Labs or Gen-2, which in turn invent a video from a text, usually of amazing quality. As a reminder, these applications are presented with sample clips on these pages:
The downside to such apps is that they are currently limited to very short durations (three or four seconds).
Invideo doesn’t bother creating animations: it stitches together existing images or videos, creating a relatively classically shaped clip. However, from Invideo it is possible to change the prompt or the parameters of the video (including the sentences of the script or certain images and clips that we can replace with those we submit) and regenerate the clip. In short, the application already seems to be very mature.
Let’s remember that we are still at the beginning of this new era of artificial intelligence and that this approach opens up further potential for video generation.