Toys “R” Us debut mind blowing Open AI Sora commercial at Cannes Lions Festival, claiming to be the first brand to debut a film using the technology.
David Skerrett
Digital | Social | Strategy | AI | CX | BIMA Hot 100 | Drum Mobile Top 50 | Drum Digerati Top 100 x2 | Author/Speaker
OpenAI's Sora is back in the news, as it was showcased by Toys “R” Us at the 2024 Cannes Lions Festival, highlighting the cutting-edge capabilities of this AI-powered text-to-video generator.
Context:
?
Sora is rumoured to be more widely available late-Summer ’24 and is in the hands of just a hand-full of creators as at late June ‘24. When Sora was first announced it caused a whole lotta hype, with it being predicted to have the potential to revolutionise the digital entertainment industry.
?
The Toys “R” Us Sora-created content explores the 1930s origin story of Toys “R” Us founder, Charles Lazarus, his son and his toy giraffe Geoffrey - which is the brand logo. With the idea of toys coming to him in a dream. People love a comeback story: Toys “R” Us having famously filed for bankruptcy in 2018, are now coming back with physical stores.
?
You can find the new ad on their site here, which also offers various Geoffrey the Giraffe toys for sale online and talks to Toys “R” Us being in every Macy’s. To be honest, it reminded me a little of the John Lewis commercials that had merchandise from the ads characters each Christmas time.
?
It was created by Toys “R” Us Studios and Native Foreign AI Labs.
?
Here’s a breakdown of this strategic collaboration, with some lessons for marketers around strategy, exploration, proactivity and workflows – with rapid production that made this content possible. Here is the ad:
?
The strategy:
?
Toys 'R' Us were interested in leveraging cutting-edge technology to reinvigorate their brand. Once the leader in toy retailing – btw the backwards R is to appear seemingly handwritten by a child – this is a storied brand on the comeback trail.
Of course, whenever I hear the phrase “debuted at Cannes” it is natural to think: is it for customer’s or marketers?
?
Taking a proactive mind set to invent the future is something I really admire about this…
?
According to CNN, Kim Miller, CMO of Toys “R” Us Creative studios, said the concept was born after she attended a brand storytelling group and told the host she wanted to do something “fun” and “different” for their next project, which involved the origin story of the founder. After Native Foreign became an early tester of Sora, CCO Nik Kleverov, called Miller with the proposal to join forces.
?
The workflow and content tear down:
?
Let’s start with “Once upon a time”: Films, novels and adverts all tend to follow the 3-act structure, which is a classic storytelling framework that divides a plot into three parts:?setup, confrontation, and resolution.
?
The opening tends to answer: “Why should an audience care”, so this usually means having a hook, so the audience are interested which can be via mystery, a protagonist, antagonist, a setting or context. Some stories start with something in full action or serve as a taste of what is to come - the key is to get the audience invested.
?
This ad sets up a golden age and gets us connected to the founder, his family and his imagination in a Pixar-style way.
?
The collaboration has enabled the rapid production of the video, demonstrating the agility and efficiency of using AI in creative processes. The way it is made is as interesting, if not more so, than the end-result. Let’s go through the 3 acts.
?
The 66-second promo, starts with a nostalgic American scene, straight-away note the glitching bicycle seat in front of the store, but the lettering on the storefront is pristine and not weird which makes me think it was tweaked and focussed on.
?
Inside the store we meet Charles and his son, the lighting effects are more impressive than most alternative AI text-to-video tools today, but the CNN article does make it clear a lot of iteration, timing tweaks and blocking changes were done in the production (by hand one would assume). The character design of the people is quite contrived, and I see this a lot when creating static and video from prompts myself. They don’t feel very original or edgy that’s for certain, as they are trained on something they can be un-original and derivative.
?
The other-worldly scene of a dream is shown with the giraffe showing incredible character consistency, the star shapes on the neck for example are remarkably polished and considered - no glitching unlike the bicycle seat in the intro. The toy filled starry sky is a signature moment in the production, as the idea for the logo came to the founder in a dream, which is a nice truth.
?
The ending is then kind-of-bolted on to then talk to the Macy’s presence for Toys “R” Us.
?
The audio is not said to have been AI generated, so the voice over, music creation composition and the script itself is likely all human crafted.
?
Overall, the piece of content was created predominantly from text-to-video prompts and some great human creativity and editing together to make the ad feel consistent, considered with elements that feel other worldly, vivid, with a fresh nostalgic aesthetic.
?
领英推荐
Time to get a bit techie, on Sora
?
They used Sora, which is trained on Unreal Engine physics and light diffusion, its best in class basically.
?
An expert from Nvidia highlighted Sora's capabilities by stating, “If you think OpenAI Sora is a creative toy like DALLE, ... think again. Sora is a data-driven physics engine. It is a simulation of many worlds, real or fantastical. The simulator learns intricate rendering, 'intuitive' physics, long-horizon reasoning, and semantic grounding, all by some denoising and gradient maths.” Sora has a super sophisticated understanding of physics and rendering.
?
Despite its capabilities, Sora still struggles with complex interactions and accurate physical modelling. Issues like unnatural object movements and inconsistencies in spatial details can detract from the overall quality of the videos. Sora excels in maintaining character consistency and visual quality across multiple shots, which is crucial for coherent and engaging storytelling. ??
?
Sora is trained using a text-conditional diffusion model that learns from a vast dataset of text paired with video clips. This dataset includes various durations, resolutions, and aspect ratios, allowing the model to understand how words translate into visual narratives
?
Reactions are telling:
?
As ever with AI related topics fear for jobs, ethical concerns over artist rights and what data AI is trained on are all important subjects for debate. In this case the early reaction to this Toys “R” Us ad is mixed, with people on X commenting that it “looks creepy”. I would agree that it looks different.
The hype train for AI at Cannes Lions seemed to have well and truly left the station, so I will assume that much Rose was consumed raising a glass to Sora in advertising in the months and years to come!
?
A word of caution, the potential misuse of AI-generated content for creating deepfakes or misleading information is a significant risk. OpenAI is addressing various concerns through rigorous testing and the development of detection tools to ensure ethical usage, such as via AI watermarking.
?
Closing words:
?
The AI-generated video brings a unique and interesting level of creativity and engagement to Toys “R” Us comms, as honestly this phoenix brand is still beloved and rooted in nostalgia for many folk!
?
Sora serves as a foundation for models that can understand and simulate the real world, a capability deemed critical for achieving Artificial General Intelligence (AGI). This real-world simulation aspect is a key reason behind Sora's advanced performance, as it is trained to replicate realistic scenarios rather than merely imitating pixel patterns. I can’t wait to see what people make with Sora when it is more widely available.
?
Using Sora, like any AI tool, is iterative, and can be quite painful to be honest, as AI can be quite obstinate and dumb – much like a cute puppy it needs training, care, rapport and patience. Lots of patience.
?
The origin story for this piece is captivating too. The bits that aren’t made by AI are noteworthy, such as: the concept, setting, voice over, blocking, editing, iteration to get the timing right, the call to action at the end, and the voice over and music.
I am going to keep telling myself (and others) that strategic and creative imagination cannot be outsourced to a bot. If we are curious and embrace the new, by working out what new tools do well and less well, we can get make things differently, realise imaginations and still make a living.
?
In marketing we are quite like Magpies, we love to collect the next shiny thing and bring it proudly back to our nest, whether we need it or not. This is the case with the next gold rush: AI. I deeply admire creators and the AI tools that will change future storytelling, how things get made and how quickly they can be made. Making great content is a human endeavour rooted in story, craft and passion. Having a clear Why is paramount.
?
I think of these AI tools like any other creative tool that came beforehand, from the canvas and paint that sparked the Renaissance, to the typewriter and printing press that transformed education, to the camera that helped build Hollywood. And each without human collaboration and imagination would be just an object on a shelf.
?
Thanks for reading and I hope you enjoyed the deep dive.
?
-ends-
?
Sources:
?
?
?
?
?
Founder CEO & Chief Creative Officer at Tigress Tigress/ Author of The Me Era/ Multiple-time award winner, keynote speaker and jury member Cannes Lions/New York Festivals/One Show/ADC/D&AD/Clio/Caples
8 个月So well written David - didn’t know you were at Cannes too - would have loved to catch up!