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

[njboywonder]

I just discovered it and have the benefit of being able to use my own photos suited up as Robin as references.

It's pretty amazing what it can do, especially at this early stage.
Brett

[njboywonder]

If I sold a Riddler costume to anyone here and you'd like to see something created, please PM.

[BiffPow]

Care to post some of your outcomes?

[njboywonder]

I'll see if I can figure out how. Been a very long time and I no longer have Photobucket that I used back then.

[njboywonder]

Can anyone recommend a free image hosting site?

Some random thoughts as I work on more images...

You have to learn the idiosyncrasies of the directions and capabilities/restrictions. It's also interesting that it prohibits certain requests but then offers to create the same images using different words. And you have to create a lot--one at a time--to get a very few good ones. Working with a successfully completed image as a base can help. Starting a new project can eliminate repeated glitches/problems. Some look more cartoonish than others. Using the term photorealistic can help. It refuses to get Robin's boots right even after I posted several images of just the boots as references. If it gets something wrong, asking it to exactly recreate that image but with the correction doesn't usually work. It knows enough about the show to ask questions for clarification--such as whether I want a campy or dramatic tone. Happy to answer any questions.

[BiffPow]

Can anyone recommend a free image hosting site?

Can you not just upload pictures from your device using the “add files” under “attachments” below? If the image is too large, take a screenshot and crop out any unnecessary portions. That will usually bring the image file size down to where it can be attached.

Thanks for the tips. I have not used ChatGPT. I don’t intend to use it, but who knows about the future? I still do most everything by hand.

[njboywonder]

Ah, didn't know that option existed, thanks.

This is one created from adding a photo of me as Robin and asking it to create an image of Robin entering Riddler's lair. I didn't provide details that could complicate it so it worked fairly well from the start. It's a very good likeness and fairly realistic. But it still took many tries to get one I liked a lot. It's cropped to fit here.

I'll add more that show that it can follow more detailed directions fairly well, if not always to my satisfaction.

[njboywonder]

This was a one of my first requests for a more active shot. It worked extremely well but for the one henchman's arm.

It gave me the idea that all would turn out this well, which definitely wasn't the case. I tried to build upon it to create a story board, but when you start adding multiple characters, elaborate props, and directed action, it has a much more difficult time producing really nice, viable scenes.

Cropped as well.

I'll do one more with a trap.

[BiffPow]

These are AWESOME! A little bit of Photoshop can fix whatever small details it misses.

[njboywonder]

Thanks, BP. They are but it takes a lot of time and thought/rewording to get them to look this good. Good thing it's raining here...

[njboywonder]

It can take dozens of attempts to get one simple single person image.

It had a heck of a time with Robin on the Batpole the way I wanted it...

Settling for this for now so I can move on to trying something else but will come back to it (For all the time I put into taking real photos recreating scenes from the series, I never got a Batpole shot. Even tried some firehouses but none had one.). Also cropped.

[njboywonder]

Next I'm going to see if I can insert myself into an actual scene.

[njboywonder]

I hit my limit for today. Some things I learned...

In some images I look older, younger, more handsome, very not handsome, sometimes more like Nathan Fillion or Pete Buttiegieg than myself. I can see Pete having the costume but not posting images on line so I'm calling that a coincidence... ; )

Sometimes it can't read Robin's tights. Proportions, likenesses, perspective, etc., get more problematic with more complicated requests, even if providing complete scene reference photos so I stopped wasting opportunities on those for now.

One image very clearly used a still from a fan film as its reference.

Trying to insert myself into a still of an actual trap scene from the series but for the reasons noted, it's very difficult. First tried the sub-zero temperature vaporizing cabinet right after henchmen pushed Batman into it but no good results. Then tried just Batman and Robin in the cabinet to simplify. Still nothing usable. That'll have to come later after I get a few really great more simple/portrait type images to print and hang.

More rain tomorrow so will likely pick up again then.
Brett

[BiffPow]

Very informative. You’re getting good results on the pics posted here. I’m not sure if this would work, but perhaps you can instruct ChatGPT to create the figure(s) only, with no background to confuse the program. Then ask ChatGPT to create the backdrop only, then use photo editing software to copy and paste/incorporate the image into the background.

[njboywonder]

I'm actually a (wildlife) photographer but don't have those skills. But a good idea.

It would be great if the program could maintain an image, so you could just put me in Robin's place in a still, but it has to recreate whatever you submit.

Some great images, but it took probably close to 100 to get maybe 10 I love. If it wasn't so much fun it would be too much work/time. As is, was kinda exciting and rewarding.