AI is starting to be scary of realism and even better results than blockbusters …
Remember, nothing here is real, just prompts …
Feeling like a caveman fighting with a big stick against primitives and converters
ZEPHYR - AI Short Film Intro (Seedance 2.0)
Zephyr - Special Episode | Higgsfield Original Series (2026)
We Made a K-Pop Action Series Using Seedance 2.0
See also Oneiric on the same channel (not age restricted but a bit NSFW), which is also a lot impressive (about 20 min)
I don’t like any of this generative AI stuff. No matter how ‘realistic’ it may appear to be at times, to me it’s all slop, and devoid of creative value and expression.
Indeed, but it worths to note that with only a few thousand dollars, you can make productions that looks better than a full first season of LOTR: The Rings of Power (which costed almost the GDP of Dominica if you count the budget + rights).
And without the woke crap!
Yes, but I am unimpressed by gen-AI videos. They’re derivative, and these projects can only exist because the costly and labor-intensive creations from which they were derived, were created first, and were available to be scraped & copied. If the production costs of the inputs (training data) are factored into the costs of the videos, they wouldn’t seem inexpensive.
Moreover, gen-AI just can’t produce original content. Even if the source material is highly obfuscated and difficult to trace, the result is still just a Frankenstein patchwork of other people’s creations. In addition to unenforced intellectual property rights, gen-AI is also currently heavily subsidized by venture capital. This cannot last much longer. As the big players in the industry burn up their cash reserves, the pricing for consumers will gradually increase. Notwithstanding the likely inevitable intellectual property rights enforcement and licensing fees, if consumers only had to pay the actual compute cost, gen-AI-videos would be significantly more expensive to make; prohibitively expensive in many cases.
Aside from the derivative nature of gen-AI and the cost of compute, in my opinion, creation should be labor-intensive. Some things should take considerable time and effort. There’s nothing wrong with creation being a slow and painstaking process. It was time-consuming creative processes that led to all of the fantastic human creations that gen-AI consumes, trains-on, and is utterly dependent upon.
With few thousand dollars you can make small budget film that’s actually good. Generally the most important thing is writing and sure, having pretty effects are nice but I’ve seen stickman animations that are interesting
Yeah, I get what you mean, I also had that angsty feeling a while back and thought of needing to pivot my video business. But that feeling is long gone since I had to make a full genAI video for a client. It’s just not worth it. Even from a business standpoint. It’s simply too generic and the only feature is superficial easy eye-candy, which lacks personality. You don’t get the quality needed to reach a real audience.
GenAi production is super frustrating and I don’t get how anyone, who really cares about his art (no matter how good that art is), would choose to use genAI over making it themself. The way these things work makes it impossible to create something truly original.
I mean, Higgsfield is the best example. Everything is just a copy of a copy. Who cares about their K-Pop Demon Hunters from Temu when I have the original?
Yes, that little dopamine hit when you get your first usable video clip is nice, but the longer you work with this stuff, the more dull and repulsive it becomes.
I’m sometimes using AI for making humoristic pictures (Gemini only, free version).
The result is not that bad.
I also tried Gemini Flow (free, limited to 3 videos of 10s per day) from the same image to make a more realistic “drama” version and the result is quite impressive.
It also makes the voice with tonality, flow and expressivity, like local people would do, even the faces are typically greek.
Here is the original concept in Greek, the target language and culture, that I use as thumb for the video on Youtube, and English version for everyone to understand the point.
The video output was made with only a few scripting description.
It’s impressive but you give up most of your creative control and ideas you might have had when shooting it on location with real people. Again, we all know it’s impressive, but it is still what it is, no matter how “good”.
I tried to create a full movie with AI, and I had the same thought process you have right now. But the longer you work on it, and try to make it work to your needs, the more frustrating it becomes. Nowadays, when I have to work in ComfyUI it kinda makes me sick.
I made the whole journey from not wanting to try it, trying it for a client, to becoming quite amazed by it and now I am back to hating it again, lol.
That’s also why I love Synfig and other ACTUAL creative tools so much. Oneself is the one and only to decide what goes onto the canvas and how to bring stuff to live. It’s much more rewarding, and people can feel it is made by a human. I have to make hundreds decision in each scene for an animated feature. All these things don’t come from a place of pleasing mediocrity, but from my own talent, hard work and also flaws.
That’s the reason why I don’t like AI, because it decides for me what to do. When I want piece of image or animation I have very specific image in my mind and I would need to give AI 2 paragraphs of description to get it right but I know it would ignore most of it anyway.
Especially with fantasy while making creatures there is no 1 way to depict them so if you want something like goblin AI will create generic little green dude
