• Welcome! The TrekBBS is the number one place to chat about Star Trek with like-minded fans.
    If you are not already a member then please register an account and join in the discussion!

Yikes! Did season 1 episode 6 use AI-generated art?

Ghost writers usually get paid and are humab at least. A.I. doesn't get paid and of course isnt human. The danger of these people that can't write is that theyll publish something 100% from a.i. and the a.i. used some things from other books unknowingly of course and the non writer find his or herself in a lawsuit.
But ghosts are not real so they can't be real writers :rimshot
 
I find this attitude fascinating.
So who wrote something is more important than if it’s good or not?

I can ask an a.i. to make a book in seconds as well. What's the point of buying one from someone that just used a.i. if I could donit myself. I just prefer stories that came from someone's own feelings and experiences. Could an a.i. book fool me? Maybe. Maybe not. But I think ill never find out. Im not planning on buying any if I can help it. I mostly follow well established authors and read a lot of older books so I am insulated pretty decently. But yeah im not shelling out money to someone that just asked a.i. to write a book and put no effort into it. I could do that on my own easily and in seconds on my phone.
 
Last edited:
by the way, this is what Gemini Pro comes up with if told to ink and color that sketch, without any other specific instructions.
Fantastic approach, thank you! And this is likely exactly what happened.

b946e83dda32.jpg

The big "tell" for me is that in each case, the AI introduced clearly destinct facial features of the actors (Shatner, Takei, Pine, Quinto) into each of these images, due to the reference images inherited in the training data of the models - whereas the original sketch has a pretty "generic" hero face.

 
The big "tell" for me is that in each case, the AI introduced clearly destinct facial features of the actors (Shatner, Takei, Quinto) into each of these images, due to the reference images inherited in the training data of the models - whereas the original sketch has a pretty "generic" hero face.
This simply makes no sense, because obviously there was a stage where it was decided that Captain Chi was to be an Asian character. So there either was a second pass on the sketches that refined the details and gave him Asian features or they were introduced with the inking. None of the characters look anything like Shatner, Quinto or Takei.
 
This simply makes no sense, because obviously there was a stage where it was decided that Captain Chi was to be an Asian character. So there either was a second pass on the sketches that refined the details and gave him Asian features or they were introduced with the inking. None of the characters look anything like Shatner, Quinto or Takei.
"Captain Chi" is clearly just Sulu - on bottom left the Takei version, bottom right a beefed-up John Cho.
They probably came up with "Capitan Chi" AFTER the image generation - in the artist' sketch it's clearly not an asian guy.
Also please don't tell me you don't recognise William Shatner in the top & bottom image in the blue shirt??
 
Fantastic approach, thank you! And this is likely exactly what happened.





The big "tell" for me is that in each case, the AI introduced clearly destinct facial features of the actors (Shatner, Takei, Pine, Quinto) into each of these images, due to the reference images inherited in the training data of the models - whereas the original sketch has a pretty "generic" hero face.
Actually what I notice here is that the AI versions are much closer to the sketch than the CBS one. Which makes sense if the latter was made by a human using the sketch as a reference instead of drawing “over” them.
 
Actually what I notice here is that the AI versions are much closer to the sketch than the CBS one. Which makes sense if the latter was made by a human using the sketch as a reference instead of drawing “over” them.
Also note that the CBS one had to be created at least more than a year ago, before/during filming of season 1.
The AI models have improved quite a bit just the last year.

(There's also I think more work by the artist being done - could have fed in the images separately & then rearranged, created backgrounds separately etc. - it's not just one page, but several that need to be consistent. There's quite some prompt engineering & Photoshop magic going on)
 
Also note that the CBS one had to be created at least more than a year ago, before/during filming of season 1.
The AI models have improved quite a bit just the last year.
Uh, very good point actually! I remember the kind of artworks dall-e used to generate one year ago and it wasn’t by far as consistent as it is now. Also, I don’t think that before say this summer you could expect ai to generate an image based on one that already existed (first time I thought of trying was in September)
 
I am curious why there's a persistent drive to find what is borderline a conspiracy about covered up AI usage in the artwork instead of facing the quite real possibility that the artist was just not very good and could only make derivative characters.

Some artists are just bad and it's okay to call them out on their work.
 
I am curious why there's a persistent drive to find what is borderline a conspiracy about covered up AI usage in the artwork instead of facing the quite real possibility that the artist was just not very good and could only make derivative characters.

Some artists are just bad and it's okay to call them out on their work.
Don't blame the artist. If you check his sketch, the guy is quite good actually. His job was to deliver a fully functional comic prop, likely on a very short deadline with limited resources.

This is not a "conspiracy" - it's the exact same company that tells us they didn't use cgi in the car chases of Fast & Furious or the dogfights of Top Gun Maverick
It's just about Hollywood pretending vfx people don't exist, aren't "real" artists and don't deserve proper wages or recognition.
 
Don't blame the artist.
As a general statement sometimes we should. Coddling and toxic positivity towards artists is not a good thing.

In regards to this maybe it was time crunch, maybe it was skill issue, maybe it was even mandated but the comic is very bland and generic even before anachronistic uniforms are added to it. I wasn't there so I don't know.
 
I think its really interesting that both AI in our experiment figured out that the subject of the comic was Star Trek based solely on the single visible Starfleet Delta. that's impressive.
I wonder what would happen if someone tried to deliberately throw the AI for a loop by including, say, the Millennium Falcon in the background?
 
I wonder what would happen if someone tried to deliberately throw the AI for a loop by including, say, the Millennium Falcon in the background?
That could be interesting but I think it wouldn’t be distinctive enough.

Perhaps a combined character, like a stormtrooper in Starfleet uniform?
 
If you are not already a member then please register an account and join in the discussion!

Sign up / Register


Back
Top