VFX'er, mographer, filmmaker, software'er. My tweets are my own. Unless they are re-tweets, in which case they are yours. I don’t respond to anonymous accounts.
A year ago I had never touched Unreal Engine. Then I took @SchoolOfMotion’s Unreal Engine for 3D Artists class with @JonathanWinbush. That started my Unreal Engine learning journey. 365 days later I’m making stuff like this.
A friend of mine is looking for a studio or team with experience doing Rick & Morty Style animation to help create a pilot and, if greenlit, a full season. If that's you, hit me up and I will connect you. This is a paid gig.
The first AI persona you’ll actually want to meet face-to-face.
Most AI personas stare blankly while you talk. Emma reacts while you’re still speaking.
After 18 months of research at @nuance_ai, this is Emma, built on Nuplex, our full-duplex audiovisual foundation model.
#houdini render. Not ai. This is one of the things that I tried to make with #claude giving it my Houdini file to make a real time web based version and it failed after a few hours of working. Probably because of three.js and glsl limitations.
My point is I don’t think AI has replaced taste, I think it is supremely well positioned to serve a market that has never really cared much about high-quality output.
And if that’s your point, I completely agree with you.
I know someone who cranks out AI generated social media marketing videos at a forest killing rate and she makes a good living at it. It’s all a formula.
But I also think that an artist who has taste can produce amazing things that sit far above that other space we’re discussing - and there is definitely a market for that - and AI alone cannot alone generate what it is these kinds of clients seek.
“Convenience (and speed) will trump taste. Taste isn’t pure.”
I absolutely agree. The way I put it with artists who argue that the journey is the destination and that the creative process is super important to them:
When you were talking about client work, where your creative talent is a commodity - Not everything you’ll make is precious. Not everything is a work of art. Clients want you to create things for them in the most efficient and inexpensive way as possible. They don’t care if you enjoyed making the thing or if it was a creative challenge. They only care about the result.
But even that doesn’t eliminate taste. I think there’s absolutely a taste threshold for different tiers of clients and budgets. There’s a ton of the market that has always thrived on generic output. No arguments there. AI changes absolutely nothing in that respect. I don’t think AI has replaced taste there. I think it didn’t really exist or matter. AI is just another tool at getting to a perfectly usable end result that no one will ever brag about.
But then again, Tiffany, Cartier, Lamborghini, Porsche, NFL, NBA… even if you make things with AI they’re going to want it to be far above average.
I’ve been explaining what I think taste is.
Taste is knowing what’s good. It’s the ability to tell what has real quality, what feels right, and what doesn’t - and to make choices based on that. It helps you decide what to make, what to keep, and what you throw away, beyond just getting the technical stuff correct.
I genuinely don’t mean to be argumentative. You’ve been telling me that taste is being replicated through the technology so I’m asking you to explain how you are defining “taste.”
But don’t take my word for it. I asked the various AI’s what they thought.
I for sure, agree that taste has eroded on the consumer side - but that was happening long before AI. Reality TV is a great example of that. AI microdramas really cash in on short attention, spans and not caring about high production values, which is what TikTok and Instagram have been about for a long time.
On the creator side, I see stuff that could vastly be improved with care, and I’m often left with wondering which of the following three things has caused it to be of a lower quality
1. They know audiences don’t care so it’s not worth making better. (This has long been a successful formula for making money producing hallmark movies and straight to video films.)
2. They don’t know it’s not great so they don’t know they should to improve it.
3. They know it’s not great, but it cost too much to redo it with no ROI for the effort and cost.
I use AI for about 50-75% of my work every day. Visuals (animation/ fx), audio (music, VO), writing for marketing. I don’t feel even remotely threatened by it. is someone who has been a professional artist for more than a quarter of a century, but who has also stayed very much on top of the technology (and even created a bunch, still used in the industry). It’s very clear to me where AI still falls short.
Since I know that you are a very smart person who has a grasp on the tech I feel like we’re using the term taste very differently.
Can you explain what you mean by taste?
We’re still talking about very different things. And I’m honestly surprised. There is literally no evidence that AI has replaced taste. The vast majority of AI content that is generated is garbage and that’s because while AI can make anything, knowing what to make, and how to assemble it meaningfully is still not something that comes easy. Not to AI and not even to people.
Tools are getting better at technique, not at taste. Technique is execution - rendering a coherent anime frame, consistent characters, lighting that roughly matches a style - even making both beautiful and ugly things. Taste is the judgment of what to make and which of the many possible results is actually worth keeping.
Your anime example doesn’t show taste being offloaded. If the work is “clearly from people who don’t watch much anime,” that means the absence of taste is still detectable in the output. Yes it looks kinda like anime, but if it doesn’t feel like anime, that means that they don’t have the taste needed to create whatever it is that anime is.
I look at @PsyopAnime and what I see are people who fully grasp what makes anime anime and make the choices to create something that feels authentic.
Audiences will watch a lot of average stuff; that doesn’t mean the tools have absorbed the ability to choose what’s good or specific. They mostly average the training data. Which means you can absolutely get good looking stuff without trying, but mostly you’ll get meh. Even a lot of the “good stuff”won’t be unique or stand out in a meaningful way.
The selection part, i.e., the rejection of the generic or ugly, the specific references and decisions - that’s still the human part, and it’s what separates the results.
Show me a compelling piece of AI generated media and I will show you someone who has taste. Otherwise, everyone would be Gossip goblin or Robert Gaudette. The very fact that there are filmmakers that stand out above others, proves that taste is still a key ingredient that AI has not yet replicated.