Here we are at the end of another year already! Can you believe it?
AI is making so much noise these days – and while people are starting to recognize its potential, it’s becoming more accessible, so more users are generating content with it. And as usual with any new technological advancement in M&E industry – it’s really just a slop generator right now. And that’s exactly why you shouldn’t worry about it, but rather, educate yourself about the tools and what they actually do.
As my regular readers know, I’ve been a huge supporter of the development of the technology for years now. I like to discover new ways of doing something and how I can break it. How can I push it to do what I want – which is usually not what the intended purpose is. But creatives never follow rules or norms, thankfully.
As I’ve pointed out several times in the past few years; we’ve been at this precipice before. As I recently started sorting through boxes from my recent move across the country, I went through several of my career memorabilia boxes and took a little historic portfolio journey since 1980. As I tracked my progress from mechanical engineering and architectural drawings to commercial airbrush artwork for aerospace, national brands and album covers, to starting a small software company in the 90s and writing about a dozen books, and winning a Telly for producing the first ever 2-day drone workshop for CreativeLive.
I only mention this because I’m curious about everything new and shiny and my curiosity usually ends up in obsession – until everyone else jumps on and I’m already on to the next thing. The Generative AI revolution is just a shift in how we do what we do.
My books on animation and green screen production are becoming obsolete now and are just reminders of how far we’ve come in a few short years. And creative minds will always find a way to break the technology to discover new ways to stand out in the midst of all the noise.
Remember how desktop publishing in the late 80s revolutionized print media and empowered consumers to create their own marketing materials and scanning their own photos, to the late 90s where everyone needed to learn how to build a web page, and then on to the digital mobile revolution where content creators and “influencers” iPhone videos – AI is merely another “tool” that is being pushed to the limits every day.
Just like these other technological advancements, we’ll see light after the smoke clears. This is where we are today – and despite what we’ve all been seeing lately, it’s not all just crap. But a lot of it is just… Slop.
MY GOD – WHAT HAVE WE DONE?
We’re seeing huge advancements appear almost daily. And while everyone is jumping on the bandwagon, so many people don’t even know where to start. I gave up trying to update any kind of directory or tools listing a couple years ago, as it’s a full-time job.
I’m still involved with testing and utilizing the tools as much as I can, but obviously I can’t begin to know everything about all of them, so I share what I can and what I’ve tried and used on real world projects, but I too am concerned on many levels what the AI industry is doing to our society and the planet. But that’s for another article I plan to address this next year.
ChatGPT rendered an honest “self-portrait” (see above and the Featured image for this article) based on how epic and awesome it thinks it is. And it’s full of erroneous stereotypes and failed executions – just like the fodder littering the social media feeds.
But there are some tools that are making the options and choices easier to access. ChatGPT and Google Gemini both generate images (no video options yet). I asked Gemini the same question that I asked ChatGPT and got this result.
So basically, AI is narcissistic? Got it.
So where does that put us RIGHT NOW with the state of Gen AI tool development?
By the time you read this article, there will be several “Amazing new updates” to many of the tools we’re currently using. That’s why I gave up trying to keep any kind of AI Tools list current as I stopped in the end of 2023.
But just how far have we come since I published my January 2025 article anticipating what the year in development might look like, and it didn’t disappoint! But I was wrong on many levels…
The tools evolved indeed, but as a result (not surprised) our feeds and streams are now full of senseless crap. The tools have made it easier to generate lots of crazy stuff to the point that we must now question EVERYTHING we see, read or hear – to determine if it’s real or AI. Thankfully, the human brain can still distinguish the soulless characters for the most part, but the dangers of AI abuse are definitely looming.
Corporate clients are still wary of AI use in their marketing productions for several reasons I’ve discovered. Some is fear of getting sued from someone using an LLM that isn’t “safe” or another copyright infringement case. Others think they don’t need marketing writers, artists or animators/producers anymore because they bought an expensive Enterprise program for the year and they can just feed the machine to spit out content. (see my analogy above about desktop publishing in the 90s).
But realistically, as cool as some demos may look and make you think “I can just use (XYZ) to create these for me” – until you actually try. Where ALL Gen AI tools break-down is still the attention to details and laws of physics. (See my article on AI Fails in October) to see some examples I can actually share publicly. Several tests I’ve done for real world clients have shown that we’re not quite “there” yet. Things that seem like they should be simple and garner expected results will ALWAYS mess with your head! Like “why did you add extra people in the scene?” or “why can’t I get them to do/act/speak the way I’m prompting and directing them?” You know you’ve seen some cool stuff online but why doesn’t mine work like that too? I often argue with the tools in one form or another when I don’t get anything close to what I asked for (and credits paid for the failures to boot adds insult to injury!) – but they are rarely amused…
Because we don’t get to see all the failed attempts these producers have gone through to get the results. The best AI video productions have come from individuals who are either embracing the surrealism and artistry of discovery that can emerge from these weird ass dream sequences set to music and storyline – or other experienced video producers/animators who know what the result is they need for their project and use a combination of traditional tools and their skills to carve out an acceptable result.
I personally have tried to work on several projects thinking that AI might help streamline my process and make subsequent edits and revisions a snap – but just the opposite has been true. I can spend hundreds of hours just trying to create some Broll clips from still images to fill in where actual footage is lacking (often unnoticed by clients) and even more time trying to get realistic representations of their products which still end up requiring a lot of work in Photoshop to “fix” all the artifacts and missing data. I have actually spent MORE time trying to get an AI result to work than if I had approached it traditionally. So while AI Tools can be helpful in one area, they fall apart in others – and trying to blend the worlds of realities to make a cohesive deliverable production is a real juggling act. And just why the “button pushers” will keep churning out “slop” – like already unbelievable UGC that floods TikTok and YouTube streams these days. It still takes a human brain and soul to make a compelling story with video and animations.
Ok, I’m getting off my soapbox for now… But let me just add this nugget of wisdom regarding the utilization of Gen AI as the sole (soulless?) source of your productions.
This brilliant satirical video from Cale Frombach pretty much sums up what we’re all experiencing in our social feeds these days! “Generate, Hallucinate, Litigate, Repeat.”
Learn more about Cale and his work on LinkedIn: https://www.linkedin.com/in/calefrombach/
AI Tools: It’s now all about the workflow
Several AI tool developers have expanded their offerings to include additional tools, like ElevenLabs, which has recently added lots of options to their amazing voice tools, such as music, sound effects and now text/image/video options that can access multiple render models through their portal.
These “combination tools” are emerging that offer seamless workflows for accessing other LLM models and processors. Other AI Tools like OpenArt, Veed.io, Kaiber and even Adobe Firefly allow you to create images and video clips from your input (text or images) and select what model you want to use to process your requests without jumping from one tool to another.
But there’s a new node-based workflow tool that I’ve discovered recently, thanks to a tip from my friend/colleague Matt Silverman with Swordfish (who will also to be featured here on PVC in February, so stay tuned!)
Weavy.ai is a node-based workflow tool they call “Artistic Intelligence” which pulls you away from the fray in button pushing mentality of most tool interfaces. It accesses most of the popular models for processing each step of your workflow so they can be combined in line with other tools – or an extension of your workflow process and ideation of the results to spark new ideas and routes to new processes you may not have considered in your initial approach.
For those used to node-based editing systems, you’ll feel at home with this sensible workflow. It’s straight-forward and logical. And you don’t have to leave your workspace to build/test your renders. And you can modify and build on one shot to another in the same workspace.
In this example, I imported my first frame model image and my pre-recorded VO audio file and a single prompt that I connected all three windows to two different render models, which garnered two very different results. I also imported a motion-only video combined with the other two windows running to three different render models for lip-sync.
Another attempt to get a decent rendered result with all the available render models (which can all render in their prospective windows simultaneously) to see what models rendered most accurately. Note that some models allow you to enter negative prompts as well, which sometimes can correct for repeated errors on your result.
Unfortunately, none of the renders produced anything with detail correctly for this test. So none of them are ready for primetime just yet!
The AI Tool slot machine – watch those credits!!
Holy crap, this gets costly just playing around with these tools! It can feel like you’re just feeding a row of slot machines, watching your credits go and go. Some tools are way more expensive than others, so you really need to keep an eye on how fast you’re consuming them!
NOTE: Beware of these “combination” tools that resell credits from all the other render models. Simply wildly clicking “Generate” without keeping track of your credits can cost you. A LOT! Using the “wrong” model in your process can vary in cost from as little as 10 up to 1300 or more, depending on the model.
For instance, Veed.io presents you with a selection of models to apply to your project prompts – which render different results, but not always successful ones. You can clearly see the vast difference between the models costs – which may be different from what you may be paying for directly in the tool of your choice.
So if you’re in the middle of a production and you bump up against your monthly credit limit in one tool, and you jump over to another to see if it can produce the results you are looking for, and it burns up all your credits there and still not giving you the results you need, you STILL get charged for all the failed renders. Very few tools credit you back for fails – only unsuccessful renders (machine/code errors) – not unsuccessful results!
Just be mindful as you proceed – especially if you’re on budget with a project. It helps to plan ahead and be sure you have enough credits in a tool before you start a project and plan/budget accordingly, so you can actually complete said project.
Carry on… but carry a big wallet.
It’s the future, baby – everything new is old again!
So let’s remember that Gen AI is still simply a tool right now. It’s not going to do the dishes or clean your toilet for you (yet). It’s also not going to produce high quality content or messaging without YOU!
Otherwise, you ‘ll just get more “slop”!
I’ll leave you with this brilliant work from artist Zahir Kahn – a satirical video about producing AI video content.
Damn, I feel this video… (written by a human)
Click image to go to the LinkedIn post and video:
