I made a short AI homage to action-packed 80s TV show “The A Team” in support of all the unjustly neglected older creatives in British TV.
These battle-hardened veterans of set, location and cutting room may be mavericks, but they will fight you out of any corner – and they are the same age as the people who actually watch television. Just because they won’t see 45 again, does not mean they are ready for the nursing home.
Here is the video:
AI-wise, this was a mission for The G Team – Google’s quirky family of Gemini, Whisk and Veo3. And like most missions, it had its share of hitches.
Gemini helped me brainstorm possible methodologies, and I was able to assemble a document resembling a Gulf War Action Plan.
But Whisk did not want to play ball with the cunning schemes I had concocted with its Google stablemate. There were a fair few “Oops, something went wrong” moments, and some stonkingly high degress of self-censorship: No, prompting for a character to be Asian does not mean that the content is harmful. No, a character probably isn’t a prominent public figure if we just made them up together.
Sometimes, it felt a bit like one of those comedy clubs where imagined sensitivities de-platform even the most harmless attempts at humour.
But Whisk did turn out to be better than expected at text-to-image. This is not meant to be its strength. And it did create some good group shots in specific locations – with a bit of a fiddle.
There are three different categories in Whisk: Subject, Scene and Style. I plonked a publicity still from an 80s action movie in Style every time, which was hardly sophisticated but sort of worked. I only ever put a single image in the Scene box – warehouse, desert or whatever – not exactly astrophysics.
But the Subject boxes were trickier. Ideally, I would have put all five team members in the Subject box, and it teased me with that possibility. Only, once I had them all nicely lined up in there, did it decide not to work.
I tried some advanced mathematics: what if I created one team of three and one team of two? That way, I would only be combining two Subject images to make my team of five. Stephen Hawking would have been proud. I love it when a plan comes together.
The rest was all good old Veo3 – some Frames to Video, some Ingredients to Video, with some Scenebuilder shenanigans delivering reasonable consistency.
AI performances are sometimes a tad clunky, but I like to think there is sometimes a bit of a Leslie Nielsen deadpannery to them.
Anyway, there is no A Team agency as yet – but the Gmail address is real. If anyone gets in touch, asking to hire maverick TV veterans, I will be sure to let you know.
