Has anyone used ChatGPT to create images of themselves as Batman, Robin, etc.?

[antman67]

I think it depend WHICH AI you use and what words you use , but my results are pretty good? You either Love AI or HATE it. It's not Purrrrrrfect but it's pretty darn close!

[BiffPow]

Nice one!

[dell]

As I stated earlier, I dropped my $10 a month Hailuo account last month. My free daily credits (enough for three videos) disappeared shortly after that with no explanation, so no more Hailuo account for me. HOWEVER, I did find a way to create up to three videos a day using Hailuo without an account.

You should be able to use the link just below. It stores your 10 latest attempts. No watermarks. It resets at 8PM eastern time for me. I avoid 8PM as generation time can be very long at that time. After 10PM or the next morning work better. I find I need to leave the site and come back to see the video. Video creation times are all over the place - 10 minutes or maybe 2 hours. Sometimes it rejects your request for no reason and that can count as one of your three requests for the day. Remember, it is free.

https://minimaxai.video/playground

Key camera commands:
[Truck Left/Truck Right]
[Pan Left/Pan Right]
[Push In/Pull Out]
[Pedestal Up/Pedestal Down]
[Tilt Up/Tilt Down]
[Zoom In/Zoom Out]
[Shake]
[Tracking Shot]

[BiffPow]

Dell, you made that sound so easy that I was tempted to use AI.

I’m sure a lot of fans will appreciate this.

[Ben Bentley]

For all the (rightful) talk and critiques of likenesses in merchandising throughout this fan base and others, I can't begin to properly express my confusion and distaste for the seemingly open-arms that some folks are embracing so much of this underwhelming, generative AI slop.

With Antman's photo in the initial post as the example, (and this is a genuine question) - what is being achieved or communicated here by this image? If a human had done this, it would fall into the category of artistic expression, but it's not... this is generative from the ground up after a wack-a-mole-word-search to assign some descriptors for a large language model.

As far the intention (or lack of it), this image is very similar in framing to an actual still from the show, except with the added bonus of distinctly lacking any real resemblance to either Sharyn or Julie or the physics of photography. I can understand incorporating generative material as a tool or a workflow into a larger project with human intention and creative direction, but to be endlessly spewing the amount of miscellaneous, generative '66 still-images out there online is beyond me. The real irony here is that eventually so much of this wonky and comparably primitive AI TV-Bat-Material will be swallowed into the machine learning that generates this stuff in the first place and potentially dilute future endeavors.

[dell]

You bring up some great points Ben. You can ask the AI to make a video of Julie as Catwoman, even using a still from the show, but the AI gets a lot of things wrong. Subtle things that you don't really think about until you see how wrong the AI actually gets it wrong. For instance, the AI has no idea on how Julie would act or look given certain emotions. The facial expressions and mannerisms are very important to making something like this look correct and the AI misses it by a huge amount. Looking at Sharyn's face in the second screenshot is a perfect example of how the AI has to extrapolate based on its limited understanding of what it "sees" in the first photo. Sharyn doesn't look like that, but the AI thinks she does.

As a result of the errors, I now stay away from trying to create videos of characters from Batman or other TV shows. I play with it, but I keep things generic and stay away from specific characters.

[njboywonder]

I'm just about done with all I wanted to accomplish (based upon what I was willing to expend).

I'd mention more that I've learned about using the two sites but I probably already provided too much information for anyone but those wanting to try them.

I have several dozen videos with which I'm happy.

I had to sacrifice a few scenes as either Chat GPT refused to create the photo or Hailuo couldn't portray the action correctly. I really wanted to do a giant prop type trap but GPT does not like having people confined. Often I found ways to reword to override its rules but so far not in this case.

I want to post more on YouTube but the site won't let me without verifying my account, which it's not letting me do. Even after twenty four hour waits--the alternative to verification, I can't post. So I'm going to have to figure out something there.

Should I have any leftover credit, I'd be happy to create something upon request, if the system will oblige to my satisfaction.

PM if interested.

Brett

[njboywonder]

On last thing...

It occurred to me that while I can't post video here, I can post 'photos.'

These are a few of my favorite source images I (had) made. All are cropped to fit.

Some of what I made are based upon the series and some upon real experiences I've had.

Despite being so simple, the first one is my favorite (I did many video versions of it as, if a source image works well the first time, it tends to work well repeatedly.) both because of the aesthetic (the atmosphere and lighting) as well as it reminding me of a number of fun exchanges I had with police officers (who show up in the videos...).

The second defied the rules, though technically the video program somehow knew there was no door--just open space with the words floating in air--since Chat GPT refused to depict it, as Robin would sometimes reach outside the box (as it were...).

Finally, one totally series-based, I was wary about adding more characters but eventually got some great results.

A final note is that videos were, perhaps obviously, often vary tonally in line with the source images. That created a nice variety of styles in addition to subject.

Brett

[antman67]

you can use either an existing image and it will almost duplicate it or create you own with AI. People either like it or hate it. This will be the last image I post here since so many are against it.

[AndyFish]

Yeah I'm with Ben.
I find the whole idea distasteful and the end result lacks something every time. I can get why people would want to play with it, and no doubt AI is going to eventually put a lot of people out of work-- can't say I favor that-- but its going to need to improve dramatically. It's like early CGI which was truly terrible, now that has killed my interest in seeing movies.
I remember when Michael Keaton was on Letterman or something and he showed a clip from whatever Spider-Man film he was in, and the response from commentators was "where there even any people in that clip?"-- and Keaton seemed embarrassed to be showing it.

[njboywonder]

(I went back to the programs so...) I'll give one more tip and one more observation...

I'd suggest not indicating how someone is feeling or what his or her facial expression should be. Aside from the fact that that can result in exaggerated/unnatural results, the programs know how to make the characters respond to the situations.

*** Admin Edit***
Policy Violation : Long standing "No more Robin in peril" imagery, artwork, short stories or manipulated images from the show.
Its been a while since I've had to remind people of this. AI or not. ***

[njboywonder]

My apologies, Admin. Wasn't thinking in those terms and should have.

[njboywonder]

Barring further discussion, in all likelihood my final post here as I finished the project unless/until I come up with something creative enough that I have to add to it.

But, pretty much unrelated to the AI and what I did with it, had these been available when I was younger, this would have been a goal, a blast. I have to think others here have had the same thought when seeing them...


Brett

[BiffPow]

I am digging the getaway go-cart-Batmobile. Like a bigger Robin side car.