Skip to content

3 Ways Video Creation AI Needs to Improve Before Pros Will Use It

From the boardroom to your Instagram feed and everywhere in between, the world is abuzz with talk of how AI will revolutionize the industry—every industry. It seems that all kinds of businesses are wondering how AI will impact them. 

Video production is no exception, especially now that OpenAI is working on the Sora AI video generator, which creates video from text prompts. 

IdeaRocket Founder and CEO, Will Gadea, has been keeping an eye on video creation AI tools as they develop. Based on his years of video production and animation experience, he’s spotted a few ways AI must improve to make it useful for video production professionals. We asked some other creative professionals and video producers to share their insights as well. 

1. Increase scene and character consistency

We’ve all seen AI generated images with human figures that have too many fingers, or extra arms growing out of their torsos. Issues like those are only the tip of the iceberg. When dealing with moving images, there are many more details that can go wrong from scene to scene. 

Ronny Young, President and Founder of Polymath Motion, said “AI has a very specific look in video right now. Things slide in and out of each other; things are rarely “on model” (to steal a Disney Animation term).”

AI developers will need to build in controls that maintain consistency of character and background from frame-to-frame and scene to scene. Companies like RunwayML are already working on this problem — investigating how AI systems can “understand and simulate the world and its dynamics. 

In June, Luma Labs released the Dream Machine which, “makes high quality, realistic and fantastical videos from text instructions and images.” Their website is sparse so far, but you can try the technology for yourself. 

2. Model for Real-World Physics

The physical world is governed by the laws of physics. We all know that what goes up, must come down, arms bend at the elbow, and we can’t walk through walls. Most AI models don’t seem to be sophisticated enough to match that reality. 

Mike Vannelli, Creative Director at Envy Creative, said, “The results can sometimes be comical or disturbingly off-mark, lacking the subtlety and depth that human touch provides.” 

He points out that video creation AI tools also generally fail to replicate human nuance and emotional resonance. “Professional video production demands seamless transitions and coherent storytelling, which current AI tools struggle to deliver consistently.” 

IdeaRocket Founder William Gadea thinks the first step to improving AI video would be to first create the world out of text, then create a 3d model out of it. “In 3d, we have already solved the physics problem. The physics of liquids, solids and gasses are simulated all the time.” Once the world is modeled, characters can move more realistically through the space. Human creativity will always be needed to guide storytelling and infuse real emotion.

3. Reduce the Potential For Harm

Any new technology comes with an array of ethical and legal questions. Laws need to be written or modified to cover the scope of what the new tech can do. Meanwhile, professionals and the public must come to some agreement on what ethical use looks like. 

Lawsuits challenging how AI tools are built and used have already been filed by YouTube creators, Music Labels, and the New York Times. Meanwhile, Elon Musk has reopened his lawsuit against OpenAI, which he co-founded. The suit includes accusations that the company misrepresented itself to get money from him. Recently, voice actors and Hollywood stars have raised concerns about how their voices may be cloned or otherwise used without their consent. 

Regardless of the legal outcome of these suits, the ethical questions remain. Edward Tian, CEO of GPTZero, a tool that checks for AI in written content, called for stronger limitations on how AI can be used. 

“We have already seen massive harm done by AI video generation, for example generating false but realistic-looking videos of celebrities doing explicit activities,” Tian said. “It should not be so easy for people to create this kind of content. If the limitations for AI in this regard aren’t handled better, the potential for harm caused to any one person could be profound.” 

Some video creators may choose to stay away from generative AI until the legal issues are resolved and the ethics are more clear.

How Creators Are Using Video Creation AI Today

Despite all the areas for improvement, some creators have found ways to use AI video production tools in their workflow. Most say they use it to generate visual ideas, create moodboards, for early drafts of scripts and to handle basic editing tasks. Along the way, creative professionals adjust, edit, and infuse their unique perspective into the work. 

A decade ago, the generative AI tools we have today would have looked like the fever dream of a science fiction writer. AI will almost certainly evolve over the next few years, posing new challenges and opportunities for video creators. 

In short, quality video production still requires the expertise and creativity of real people with unique perspectives. For more on how video creation and video marketing are evolving, join our newsletter. To start your video project today, contact the video production experts at IdeaRocket

Receive our
free book
when you sign
up for our
newsletter!