AI-Assisted Medical Animation
I’ve written about artificial intelligence resurrecting our dead family members, but even in the short time since we posted “Tell Me a Story Grandpa,” we have found dozens of more practical uses for AI in producing healthcare communications.
The most important thing to know is that the VFX industry still can’t directly use AI to generate medical content. The explanatory graphics and animations that healthcare and pharmaceutical companies need for their educational content must be very accurate and precise, and AI can’t do that yet. Besides that, precious few people on the planet even know about these new mechanisms of disease (MODs) or mechanisms of action (MOAs) before the drug is on the market; an AI wouldn’t be able to illustrate anything about them because these discoveries are not yet in any of its learning models.
That said, AI has become integral in other ways, like producing inspirational imagery for mood boards — one of our pre-production, visual brainstorming tools. AI is also very good at quickly providing many iterations of visual ideas, each slightly different. Our team of humans make the final choice, and then take over to ensure all of the details are accurate. We’ve used this method with great success to create background imagery inside the bloodstream (affectionately known as the “meat wall”).
Or for more detailed elements in a scene, previously we would have had to research and source images, or typically paint custom textures by hand to design the perfect branded (or unbranded) look for each element in a scene. Today using carefully curated text prompts, AI can generate wet, slimy, bumpy, etc. surface textures to be applied to our library of three-dimensional models we’ve developed over our decades of working in this industry. Obviously, the tools for generating these textures have come a long way.
AI has been making inroads into the other senses, too. Voiceover work is increasingly the domain of AI, and there’s a surprising reason why: the fast-talking disclaimers in drug commercials that everyone makes fun of, are actually quite challenging to produce. They typically involve multiple back-and-forth communications with attorneys, and the copy usually changes several times. With the help of AI, we can ensure that there won’t be a change in the narrator’s voice if anything needs to be inserted, sped up, or slowed down.
Creating final animations directly with AI is likely to remain a challenge for the foreseeable future. In addition to the lack of medical accuracy, playing back a sequence of AI generated still images results in flickery motion which is hard on the eyes. Accuracy and flow are two aspects of the creative process that seem to rest squarely in Team Human’s court. For now.