I saw Part 1 of this and it was pretty good and fun to watch but dead-ended: Q brings a Darth Vader and his Star Destroyer to confront Picard in the Star Trek universe. That video is summarized in the first 30 seconds here.

But this Part 2 is really, REALLY good. Next level. There are all kinds of places the author can take this.

Speaking of whom:

A lot of people have asked how I actually make these videos, so I thought I’d use this pinned comment to explain the process.
There isn’t a button I press that spits out an episode. 😂 A video like this involves hundreds of individual generations, tests, failed attempts and edits before I get something I can actually use.
I start by developing the story and writing the dialogue. From there, I break everything down into individual shots and work out the camera angles, character positions, eyelines, performances and continuity between shots.
I then create the reference images needed for each scene and use ComfyUI and MiniMax H3 to turn those images into video. A single finished shot can take multiple attempts because the movement, facial expression, performance, timing or continuity has to be right. If it isn't, I generate it again.
Dialogue and voices are another part of the process. Depending on the character, I use generated voices, voice references and, in some cases, recordings from real people. Those recordings then need to be edited, cleaned up, EQ'd and mixed with the rest of the scene.
Finally, everything goes into Adobe Premiere Pro, where the episode is assembled shot by shot. That's where I do the editing, timing, colour work, dialogue mixing, sound effects, ambience, music and all the other little details that turn hundreds of separate pieces into the finished video you're watching.
AI is obviously a major part of the toolset, but it's a filmmaking tool here rather than the filmmaker. The story, writing, shot selection, performances, continuity, editing, sound design and the countless decisions about what actually makes the final cut are all part of the production process.
Hopefully that gives you a better idea of what actually goes into making one of these videos.

Somehow I liked the first one better. I think it's way too short. But it has some background story. This one feels more like a tech demo. I mean it's impressive and looks/sounds realistic. But it doesn't really go anywhere?! It opens up places to go, but I don't see any actual idea of where to go. It's just a sequence of characters doing arbitrary stuff, then a crossover pops up and those are confused about where they are. They might say their default sentence, but that's pretty much the entire idea. There are no funny or thoughtful points being made.

(Plus I think we're set on the Strange New Worlds / Doctor Who crossover. TNG kinda just shows they're not aware of what happened in the last 2 years.)

Ultimately I think great achievement. But they didn't make it into my YouTube favourites. There's a few AI videos I think are hilarious, though. For example the AI Severus Snape music video: https://youtube.com/watch?v=SlsEDkavoso

It's not just a demo of how we can fabricate videos now. But they also made a song which has a meaning in the Harry Potter story context. There's all the imaginative stuff about the characters and what they'd be like in a different context. And there's also all these references to real music videos and artists. There's just several layers of meaning going on on top of the in-your-face meaning, you can watch it 3 times to try to get more references and it just feels like someone put thought into the scenes.

Just always need to pay attention to watch them in some private browser video. Seems the YouTube algorithm knows what's AI and if you watch too many of them, it'll make half the next suggestions be some AI stuff.

I'll check the Snape vid

Damn, crosspost to startrek@lemmy.world downvoted to oblivion. I would have thought people deep into ST would appreciate it even if it's just a string of cool vignettes. I think people feel they are defending the human race by hating on anything created with AI in any way -- even if the creator is using it to create a very specific detailed work, not just churning slop.

Hmmh. Yeah, was to be expected. I'd say a bit excessive. I regularly see other ChatGPT written "news" articles, opinion pieces or random other slop get some two digit upvotes on this platform. Because people have no clue what slop is. You make an effort, disclose it in the title, write a post with context and include the making of... And that's all downvotes 😑 You got 2 upvotes here. Just seems that's about the number of people who appreciate discussing these videos.

(And if there's one legitimate use case for AI, it surely has to be spoofs of popular TV shows and movies.)

Wonder if there's more good AI videos out there on things like beloved TV series. I'll occasionally scroll through Youtube and there's definitely a lot of low quality stuff. I'm longing for some really great ideas though. Like the good old human-made dubs, cuts, music videos, skits or a random bluegrass cover of the title music.

one legitimate use case for AI

TBF, curing most diseases via unlocking the proteome, interpreting biological systems like the neural connectome, and quite plausibly autonomously mining asteroids to provide Earth with abundant resources to make O'Neill cylinders.

Those are all better uses that we should be talking about.

AI Severus Snape music video

That was surprisingly good. I particularly liked Grunge Dobby. And the fact that Alan Rickman is dead and that I rather liked him parasocially gives it some legit feels.

Fair enough. I should have been more precise. I'm mainly a bit disappointed with generative AI in specific and with real existing AI versus hypothetical AI.

I mean part of the debate is: we frequently discuss how AI might one day cure cancer, solve our energy crisis or become sentient and start the robot apocalypse etc but then that's more an advertisement campain with hypotheticals. In reality we have a lot of chatbots and those aren't anywhere close. Sadly it's a bit impossible to look up whether we invest more in Alpha Fold or colonizing the moon with AI, than we do invest in gooning pictures like Elon Musk does. And the AI companies aren't working on solving the energy crisis either, quite the opposite. They should, though, IMO.

Not sure, I really think it depends on the use case as well. I think even if we're talking about one and the same video generator, it can be used to cut cost, dial down quality and let go of artists. Trump can use it to generate war propaganda... But it can also be used by random internet people to generate something funny or thought provoking... I think that'd be two different pairs of shoes. It's still slop, but I mean we can turn a moldy sandwich into art, I don't see a problem with that in itself 😂 Issue is just all the circumstances and how very few people can pull it off.