Earlier this year, I predicted that with the expansion of AI, we would see brands and content creators rely more on AI to create videos. Well, that day is here! On February 15, OpenAI previewed Sora, which makes it possible to create videos from text prompts:
As OpenAI noted, Sora can create videos of up to 60 seconds featuring highly detailed scenes, complex camera motion, and multiple characters with emotions. Contrary to what you might have heard, OpenAI did not release Sora to the public. It’s still in development. As part of that, OpenAI gave access to a few visual artists, designers, and filmmakers to gain feedback on how to advance the model to be most helpful for creative professionals. Even still, the buzz about Sora has upended the advertising industry just as ChatGPT did in November 2022. Here are some of the main points advertisers should keep in mind:
What Exactly Is Sora?
Sora is an advanced AI model that can generate videos based on simple text instructions. You provide a descriptive prompt, and Sora will try its best to create a video that matches. It’s just like ChatGPT but for video creation. Prompts lead to creation. Sora is built on diffusion models, a powerful type of generative AI, but it uses the strengths of transformer architectures, known for their ability to handle complex relationships in data.
Sora’s Key Capabilities and Limitations
We’re not talking about creating stick figures, folks. Sora can develop amazing animations as you can see from the sample video courtesy of OpenAI that I shared above, which depicts everything from woolly mammoths to tourists in Tokyo. Sora isn’t limited to realistic videos. It can generate videos with stylized animations, artistic effects, and more.
One of Sora’s strengths is creating videos with fluid camera pans, zooms, and rotations, adding even more realism and a cinematic feel. Sora demonstrates improved handling of how objects move within a scene (for example, a person disappearing behind a tree and reappearing) and maintaining consistency over longer videos.
But as I noted, Sora is a research project and not yet publicly available. While impressive, it still produces some inconsistencies, demonstrating there’s room for improvement. Like any advanced AI tool, there’s potential for Sora to be used to create harmful or misleading content, such as deepfakes.
Sora Is Not the Only AI Tool Available
Sora adds to a roster of text-to-video tools currently available in the market. Gen-2 by Runway has been in circulation for a year. Lightricks, maker of apps like Facetune and Videoleap, announced an AI-powered filmmaking tool, LTX Studio.
Gen-2 offers capabilities such as image-to-image translation within video, storyboard generation, and the ability to control video generation with text prompts.
LTX Studio is still in its early stages, but it promises to bring the power of AI video editing to a simplified app-style interface. Its focus is likely to be accessibility for everyday users and social media content creators. Other tools include:
- Phenaki: this platform boasts speed as a primary advantage, promising quick video generation based on text descriptions. It also includes features for editing and refining the AI-generated output.
- Make-A-Video (Meta): while not publicly available yet, Meta’s AI video generator has gained attention for its ability to produce short, playful videos from simple text descriptions.
- Imagen Video (Google Research): like Make-A-Video, Imagen Video focuses on generating short clips and demonstrates strong visual quality and coherence. It’s also still within Google’s research labs.
But Sora is gaining attention because it demonstrates so well the possibilities of text-to-video creation – potentially democratizing content creation as ChatGPT has done.
Why Sora Is Significant
While AI image generation has become remarkably advanced, Sora takes things a step further by creating moving images. This opens a new world of possibilities for creative expression and problem-solving. Sora’s development hints at the increasing ability of AI to understand and simulate aspects of the real world, including physics and object interactions. This has potential for use in robotics, self-driving technology, and more.
Sora could also change industries like filmmaking, animation, and game development. Imagine easily generating storyboards, concept art, or even full animated sequences with a few lines of text. Well, it’s not hard to imagine. Filmmakers instantly saw its potential to change things. Filmmaker Tyler Perry immediately halted an $800 million studio expansion because of Sora’s potential to do work that humans in a studio could perform.
Reactions to Sora
Sora has ignited both excitement and anxiety in the advertising industry. Advocates for technology such as Sora believe it can be an asset that enhances creative potential.
Conversely, those involved in the hands-on aspects of creativity express concern that reliance on such fashionable tech undermines genuine artistic integrity in favor of quick and inexpensive solutions. How can this be if Sora creates realistic videos? Well, even OpenAI notes that Sora is a work in progress – but this won’t stop some businesses from adopting a “Let’s just be fast and quick because we can be” mentality. In other words, businesses that don’t have a high bar for creativity will either crank out first drafts of video content and call it “final” to minimalize costs or ask their agency partners to do so. A recent Ad Age article describes the potential problem with businesses settling for video content that lowers the bar for what is considered effective:
When OpenAI revealed the first videos of Sora just over a week ago, Phil Russell, a video editor at production agency The Production Factory, also felt bowled over. But 20 minutes later, that feeling became something more like hollowness as he realized that Sora could accelerate demand for unimaginative and stock-like ad creative.
“We’re just filling more empty content, as opposed to [bridging] any real gap in the industry,” he said.
In addition, because Sora can create realistic videos, questions about transparency have arisen. If consumers cannot tell the difference between an AI-generated video and a human-generated one, will AI-generated video require a disclaimer, such as “This ad was generated entirely with AI?”
What Happens to Video Talent?
And, of course, Sora has sparked anxiety over job loss. And that’s understandable. Video production can be an expensive process. Sora can cut costs and reduce the time required to create videos for ad campaigns.
But AI could create new roles, too. For instance, agencies may use Sora and tools like it to create ideas and development them with clients faster. “What if” scenario planning can become more creative because Sora makes it possible to do better A/B testing for videos for different campaigns, audiences, and publishing formats. As an advertising executive told Adweek, “Instead of asking for one ad, clients could ask for four ads. You may not need as many producers on staff, but you’re going to need people who can work with the new technology.”
We just don’t know yet. But anyone involved in advertising can start adapting.
Ways to Use Gen AI Tools such as Sora
So, take a deep breath. As I noted, we don’t yet know where all of this is headed. But if you are in the advertising industry, I would recommend you use these tools and figure out what your business case is for making your job as a human even more valuable with AI. For instance, consider AI helping you with:
Idea Generation and Concept Exploration
Use these tools to quickly generate a wide range of visual concepts based on loose creative briefs or campaign themes. This overcomes initial creative blocks and lets you explore unexpected directions. In addition, experiment with different styles just by changing your text prompts (e.g., “retro cartoon,” “hyperrealistic product demo,” “noir cinematic trailer”). This can help define a campaign’s visual language. And as noted above, consider AI for those “what if” scenarios of testing new ideas against a client’s brand.
Prototyping and Storyboarding
Create quick-and-dirty video storyboards for client presentations or to test audience reactions before investing in costly production. And use AI tools to “block out” complex camera movements or scenes that would be difficult to explain with words alone. This gets everyone on the same page visually.
Efficiency and Iteration
Generate multiple variations on a video with slightly different prompts to quickly provide options for client feedback or A/B testing. And use these tools to adapt to feedback. Don’t love the opening shot the AI generated? Use text inputs to change it, experimenting with different angles, lighting, or even the product’s color. Think iteration as you create variations of a video tailored to different audience segments based on their data profiles or preferences.
Where Your Human Expertise Shines
AI is rapidly changing, but today it does not replace these important attributes that people bring to creativity:
- Strategic vision: AI won’t replace the big-picture thinking needed to connect a video campaign to overall brand goals and messaging.
- Emotional resonance: no AI yet understands human emotion as you do. Ensure videos have that something special that cuts through the noise.
- Understanding nuance: AI can struggle with subtleties of language or humor. Your role will be crucial in refining the text prompts and ensuring the output perfectly aligns with the desired tone.
- Technical finesse and polish: while AI video generation is impressive, it’ll often need the final touch of a skilled editor for seamlessness or adding audio elements.
- Ethical oversight: be the final arbiter of whether the AI output is appropriate, free of bias, or could cause unintended harm.
Think of these tools as additions to your creative toolbox, not replacements for your talent. By using them strategically, you can unlock new ideas, improve your output’s efficiency, and ultimately deliver even more successful campaigns. Explore our website to learn how we help our clients with their paid media campaigns, including the use of AI.
