How Gen AI is transforming video, motion design, and the visual arts

Thank you for reading my latest article.? I regularly write about technology and Digital Health trends. Follow me for future articles

AI’s impact will be felt quicker in creative industries than almost any other.? Journalism, copywriting, photography, design, motion graphics and video production, are all beginning to feel the effects, as the technology rapidly evolves.

With the democratisation and automation of the creative process however, comes with pitfalls.? In many cases, AI will simply allow humans to achieve quicker outputs at worse quality.??

The fundamentals of great design, copy writing and production quality will never be more important, as the world becomes deluged with motion graphics, avatars, animations and video over the coming years.

Highly skilled designers and editors will always be in demand at the top end of the market but as some of these tools replace time consuming tasks, billable hours will start falling.? Low level jobs will increasingly be done by marketers and business owners, skilled at manipulating AR interfaces, rather than hiring agencies.

With all that said, the evolution of these tools is happening at an incredible pace, and in my personal work (architectural photography, video) many are already saving hours of time.? As a creative marketer, I’m fascinated in evaluating the advancements and what they will mean for the future creative marketer.

Here’s a round-up of the platforms and features I am finding most useful at the current time (and to get more of a flavour of what they are capable of, take a look at the videos below).? I have absolutely no doubt this list will be out of date in a couple of months.

5?real world examples of AI driven video and image creation

Image generation with DALL-E 3 and Microsoft Bing.

Along with Chat GPT, image creation was one of the first Gen AI use cases that got people excited.? Now it seems everyone can be an image creator, with the unfortunate side effect of poor quality images flooding social media.? MidJourney for a long time has been the one to beat, but is limited by continuing to be hosted within Discord.? The recent (Dec 2023) MidJourney 6 brings new prompt inputs, and a better handling of text outputs). But I’ve recently switched to using Microsoft’s Bing Image Creator, which only requires access to Bing.com, no need for an OpenAI account or signing up and installing Discord.? Powered by DALL-E 3 , it's easy to use, and fast.? In my view images are of an equivalent quality or better than Midjourney 5.? See the video for more.

Great for: Any image requirements, Linked In posts, thumbnails and much more. Alternatives: Midjourney, Stable Diffusion

Image to video animation - Runway Gen-2

You may never have heard of Runway, but they’ve been hitting it out of the park recently, and none more so than in the area of video animation.? In lock down, I spent two years trying to make headway in 3D animation (using Cinema 4D), and while I made progress, it was a tortuous and time consuming learning curve, with rendering even short sequences taking days.? Now much of what I was trying to achieve can be done using Runway Gen 2, with its ability to render 3D type images, either from images or raw text prompts. A significant advantage of driving video from an image, rather than raw text, is the creative control it affords.? Add in the new Motion Brush feature, which allows even more precision, results in a very compelling toolset. Watch the vid to see it in action.

Great for: Short, 4 second, video animations, Linked In posts, opening sequences. Alternatives: Pika, Stable Video

Adobe Photoshop Generative Fill

Like any large incumbent that dominates its segment, Adobe comes in for a fair amount of flak.? But last year, the company showed us all that it's no slouch when it comes to integrating AI into its feature set.? Generative Fill absolutely wowed image creators in 2023.? Its ability to generate image extensions, remove distractions (I use it extensively in my architectural photography to clone out people, cranes, traffic and clutter) is genuinely astounding. It can do in minutes what used to take hours of work with the clone tool.? Check out the video above for a glimpse of what it can do.

Great for: Image extension, up-scaling, and anything that calls for photography. Alternatives: Luminar, Canva (but neither is as powerful)

Creating digital clones and avatars for video with HeyGen.

Remember the deep fake Tom Cruise from a few years back?? Well the same kind of technology is now within reach of all of us.? HeyGen allows you to upload a two minute video of yourself, and then generate a very realistic video and speech clone of yourself.? You can also choose from a range of ‘off the shelf’ avatars.? Upgrading to the ‘Fine Tune’ options, allows greater fidelity and resolution? (though I didn't notice enough of a bump to justify the price).? Once you’ve set up the avatar, you can upload your own scripts and have your avatar of choice present them, saving hours of time.? There is also wide language support, meaning you can have your avatar speak in Mandarin or any other language.? (I found results varied with this, my Chinese speaking friend could understand my Mandarin Avatar, though none of my Thai relatives could understand a word of my Thai version).?

Great for: Explainer videos, product demos, training, internal comms and more. Alternatives: Sythesia.

Text to video creation with Pika

Gen AI video creation, whether from raw text prompts, or a starting static image, is a complex challenge 10 times the complexity of static image generation.? With recent advancements in GPU hardware, engineers and researchers are making rapid progress on providing platforms that promise new levels of video and animation creativity.? Leading amongst the pack in Pika, whose text to video platform went live recently. While still in its infancy, these early generation products are already seriously impressive, even if they are restricted currently to very short 3-4 second videos.? One other limitation, is that the level of user inputs and control points are fairly limited, and a text prompt is a pretty blunt instrument in this regard.? In time we will see more interface controls, in a similar way to Runway, which provides a great deal of fidelity with its camera controls.? Discover more about Pika here.

Great for: short videos, animated thumbnails and intros. Alternatives: Runway, Leonardo.ai

Let me know in the comments your own experience of using these tools, including ones I may have missed, and where you think we will head next with this exciting technology.