Future generations might believe some of these Seedance 2.0 videos – unearthed from a long-buried flash drive – represent the state of the art in mid-20th- and early-21st-century cinema. Perhaps they won’t notice the odd movements, lack of blinking, stilted dialogue, and fixation on hand-to-hand combat.
I hope that’s not the case, but if you consider how actual film is decaying at an alarming rate and that digital content is nothing more than stored, everlasting bits and bytes, this scenario isn’t quite so far-fetched. That, of course, would be a calamity.
A silly little AI 1960s Comedy short that I created with Nano Banana and Seedance 2. pic.twitter.com/haKQuQVYqSFebruary 22, 2026
For its similarity to mid-20th-century Douglas Sirk films and their 1950s-style saturated blues, it’s almost enchanting. That is, if you can look beyond AI artifacts like a 12-piece band populated by a dozen duplicate musicians. Or, if you don’t mind, numerous restaurant customers who look oddly similar.
There are other anomalies, and many of them are hallmarks of AI Video slop produced on Seedance 2.0 and other platforms. Still, the abundance of Seedance 2.0 content is almost unprecedented. As I write this, social media is flooded with short videos featuring countless characters, usually engaged in some sort of battle or an impossible cross-brand intersection.
I’ve seen at least two Matrix-rip-off videos featuring rematches between Neo and Mr. Smith. There’s a video of Marvel’s Doctor Strange battling DC’s Superman, and another featuring the cast of The Office meeting Iron Man.
One battle after another
A Matrix-level action scene used to cost $10M+ in Hollywood industry.Now it’s done in 2 minutes.Seedance 2.0 on MartiniArt_ 🔥 pic.twitter.com/uL1Y3Wf1IuFebruary 23, 2026
Time after AI Time
In every instance, those creating the prompt and generating the videos pay no mind to intellectual property rules or the concerns voiced by those representing actors. What’s worse is that the person you know best for playing [fill in the blank] character is forced to play it again (without their consent) in these renegade videos.
It’s a big problem, to be sure, but I found myself gravitating to two of the more original videos, the ones trying to tell fresh stories without tapping into someone else’s IP.
I started wondering how they were created and considering how, even when the ideas and characters are fresh, there are the inescapable oddities of the Seedance 2.0 videos.
In Hashe Al-Ghaili’s Time Traveler, a character who unsurprisingly looks a bit like Al-Ghaili enters an HG Wells-style time machine and travels through time to stop a Vienna art school from rejecting Adolph Hitler and thereby averting World War II. It doesn’t work, though; it only postpones the inevitable.
Time Traveler (Made with Seedance 2.0)I created this time travel short scene using Seedance 2.0 in just one day for under $200. pic.twitter.com/ImeoTh0vLeFebruary 22, 2026
It’s an attractive 5:30-minute clip, but the AI oddities just keep piling up. For some reason, it’s all “shot” Wes Anderson style, with every character framed dead center.
Nobody blinks, and emotions are either missing or delivered in odd ticks, like one of the characters seeming to sniff his pen in panic.
Like much of the Seedance 2.0 content I’ve consumed, I noticed how the skin on most characters is, at times, a bit plasticky. The effects can be good, but they tend to be repetitive. My guess is that Al-Ghaili generated them once and then reused the sequences.
My favorite part might be the robot. Though it, like so many things in this and other AI-generated videos, is derivative.
All of the shots were created with this single image created in Nano Banana on @freepik – for a couple shots, I took screengrabs from videos and brought them back into Nano Banana to create variations, or to edits slightly. pic.twitter.com/F0EXvwKmbBFebruary 23, 2026
Another AI time, another AI place
For as much as I might dislike these videos and the dismay, anxiety, and consternation they’re generating across multiple industries, I am fascinated by how they’re made.
Many creators like to claim that they created the work with a “single prompt,” but I suspect they’re being somewhat disingenuous.
I noticed in Christopher Gwinn’s post that he credited Nano Banana with some of the work in his “Silly little AI 1960s comedy short.” I had to learn more, so I peppered him on social media with questions:
- Was in a single prompt or multiple?
- Who wrote the dialogue?
- How much description did he have to deliver to Seedance 2.0 for the desired outcome?
- Did he tell it to “use the same “actors” across multiple scenes and within the same scene?”
More than just a prompt
Gwinn, who works in Hollywood as a digital creator, told me on X that he started with a single Nano Banana AI-generated image (above) that he built in Freepik. That image, which was inspired by the films of French filmmaker Jaques Tati (famous for the Monsieur Hulot comdedies (he directed and played the lead), was used to flesh out the entire Seedance 2.0 sequence.
While Gwinn usually writes his own dialogue, he went a different route with this short comedy: “I told Seedance what was happening in the shot. I only wrote a couple lines myself – sometimes after it generated some original dialog, I would tweak that slightly and re-run the prompt,” he shared with me on Threads.
Gwinn also reused some characters across multiple shots. Once he had all the pieces, including the same couple dancing in multiple scenes, he cut and edited in traditional video editing software – he switches between Adobe Premiere and CapCut.
What Gwinn described to me was a process, and ultimately not one too different from what a traditional filmmaker might do. There are notable exceptions, like the use of AI-generated people instead of actors. Plus, for all Gwinn’s work, he can’t quite remove the funhouse mirror feel of the whole enterprise.
Something’s off
Sure, it might remind you of comedies from the 1950s, 1960s, or even the 1970s, but it also feels just off. The slapstick comedy makes little sense since there’s almost no setup for each gag. We enter almost in the middle of each comedic moment. It kind of made me think I was watching a trailer for a mid-brow period comedy that was trying too hard for laughs.
The other anomalies, like physics not quite working and bodies sometimes moving as if they didn’t have bones, are evident across virtually all Seedance 2.0 clips. With the rapid pace of AI advancement, however, they’ll be solved in a few months.
I like understanding how these videos got made. It makes me feel a little bit better about the rapid progression of this “art,” knowing that the digital creators behind it are probably using far more than just a single prompt to achieve the desired result.
I hope, though, that in their quest to generate ever-more bizarre scenarios for Neo, Iron Man, Superman, Brad Pitt, and Tom Cruise, they stop and think about how they can use these tools to create something new, and art that can finally stand on its own.
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds. Make sure to click the Follow button!
And of course, you can also follow TechRadar on YouTube and TikTok for news, reviews, unboxings in video form, and get regular updates from us on WhatsApp too.
https://cdn.mos.cms.futurecdn.net/Yaoy2guUrJ8bPSDvikBbXD-1920-80.jpg
Source link
lance.ulanoff@futurenet.com (Lance Ulanoff)
