Next-level AI video generators: what can AI video tools offer?
Simply the fact alone that a clip has been created using an AI video generator is often enough to spark curiosity among marketers. Read here how a Toys“R”Us AI commercial has been causing a stir and which AI video tools are ahead of the pack.
Artificial intelligence in video marketing
Short video clips and reels ensure high reach on almost all popular social media platforms. To boost content production, why not also use AI video generators?
Toys“R”Us and its AI-generated commercial: the toy retailer’s dream
Charles, the young son of a bike shop owner, dreams of having his own store someday. And not any old store, but a very special one: for enchanted toys. The boy’s dream is a glittering paradise of magical plush animals, toy cars, dolls, musical instruments and building blocks.
The touching commercial from 2024 shows the dream of Charles Lazarus, founder of Toys“R”Us. However, the company itself had had to file for bankruptcy seven years earlier, as the toy giant could no longer hold its own against online competitors. The twist in this tale is that the video is AI-generated. The message: the boy’s dream may be a product of its time, but modern technology is rescuing the company and helping Toys”R”Us to make a remarkable comeback.
AI for your promotional video: the latest generation of AI video generators
In addition to this heart-warming tale’s unusual idea and remarkable timing, there’s another reason why the clip has been garnering attention – the video was generated using Sora, a tool from OpenAI that was not yet available to the general public at the time. The Toys”R”Us commercial was therefore simultaneously an advert for the next generation of AI video production tools.
Although OpenAI announced the release of Sora for May 2024, the AI tool is still not freely available. However, we’ve taken a look at Sora’s capabilities and the other AI video generators out there.
But first, let’s tackle this question: what can you use AI video generators for? Here are some examples of how AI can be used in video marketing:
- Advertising and marketing: promotional clips, product presentations and explainer videos
- Education and training: instructional videos and tutorials
- News and media: news items, short videos and content for social media (AI tools are becoming increasingly important for creating reels in particular)
- Movie production: development of special effects, creation of computer-generated imagery (CGI) and simplification of movie and TV series production
- Game development: animated cutscenes
That’s the theory, now for the practice: four AI video generators
Which AI can create reels? And which AI generates the best videos?
Sora: social media reels with AI
Sora can generate videos from text prompts, similar to text-to-image generators like Dall-E. Users describe the storylines or scenes they would like to see, which are then rendered by the neural network.
However, the Toys”R”Us video also shows the weaknesses that still exist: for example, the AI has difficulty generating consistent character models. We can see this from the changes in young Charles’ appearance throughout the commercial. Despite these minor irregularities, however, Sora would appear to have really impressive potential – especially when you consider that it can generate videos up to a minute long and will also be entirely free to use.
Runway ML (Gen-2): AI that turns text and images into short clips
Runway ML has been around since 2023. It works like Sora, turning text commands into results. However, currently it can only be used to generate 18-second clips. Gen-2 is the patched version and far exceeds the limit of the popular Full HD resolution. However, Runway ML (Gen-2) can also generate videos based on photos and images.
Runway ML (Gen-2) can initially be used free of charge. Once you’ve used up your free 125 credits, you can opt for a subscription model that currently costs 15 U.S. dollars a month and gives you a total of 625 credits.
Like most other AI video generators, Runway ML (Gen-2) particularly impresses when it comes to non-realistic imagery. Animals and people, on the other hand, often appear distorted or inconsistent. The tool is therefore not (yet) flawless.
Stability AI: open-source AI for all
According to Stability AI, Stable Video Diffusion in its present form is more suitable for research purposes than commercial use. However, a subscription service was introduced in December to enable its commercial application. Stable Video Diffusion can still be used free of charge for non-commercial applications.
There are currently two video AI models, one capable of generating 14 and the other 25 frames per second. The clip length of just four seconds is not quite so impressive yet. However, Stability AI hides nothing from its competitors, instead making its source code and updates available as open source.
FLUX.1: the shooting start from Baden-Württemberg
FLUX.1 is truly the star of the moment. Although we’ll have to wait a while yet until text-to-video production is available, FLUX.1 is impressing the industry. As the manufacturer’s name – Black Forest Labs – already indicates, this product is a local. Its really major plus point is the hyper-realistic images it can generate. It can even accurately render hands – something most AI image generators still struggle with.
There are currently three different models. FLUX.1 [pro] is the most powerful of the three and can even be used to create commercially usable products. FLUX.1 [dev], meanwhile, is an open-weight model intended for research purposes, like Stable Video Diffusion. FLUX.1 [schnell] is an efficient solution for local development and personal use.
Not long now: AI rules reels
Although there are still some teething problems with most AI video tools, they are gradually improving every day. It’s therefore reasonable to assume that creating AI reels will soon be an everyday task for marketers, even if they have no knowledge of editing.
