- Time
- (Edited)
- Post link
Can you imagine Plagueis in episode 9
Truly exciting times ahead
Deepfake Ideas - Index and Discussion — Page 7
- Time
- Post link
My deepfake face and voice preferences throughout the movies (using Hal9000 edits as my personal Prequel canon movies)
1st - would need more than deepfaking though, but require significant editing of trying to change angles of frames/editing line of sight of other characters (maybe other AI tools for a base filling out beyond the cropping of frames) - Make Anakin older, closer Padme’s age. This would require the character to be taller larger, hence the comment about the serious frame editing. Obviously a voice change to reflect being a teen character. But having the character from the get go being a mess of teenage hormones, desire for identity and independence, emotions, cocky charisma but ‘farmboy’ innocence, coupled with the supernatural abilities would have plugged people into the character arc far more. Anakin and Padme both could have aged up, which would have paid off with the story lines in the following movies, but it was the Anakin age casting that was always the fundamental problem with that relationship in the stories.
2nd - Voice AI, get all references of “padawan” changed simply to “apprentice”.
3rd - Voice AI, get Jar Jar and all Gungans voices altered to an old rural Fen/East Anglian accent:
https://www.youtube.com/watch?v=UTVwdv9Pzo8 (the men 2 minutes in are perfect example)
4th - Voice AI, RotJ, A-Wing pilot voice back to match Poppy Hands
5th - Deepfake and Voice, Solo (Coaxium Heist) - Simply change to Anthony Ingruber
One other, for a laugh - Voice AI, get all Jango and Boba Fett’s lines changed to Ed Begley Junior in A Mighty Wind
- Time
- (Edited)
- Post link
This deepfake of palpatine and clone wars anakin voice doesn’t sound robotic sounds just sounds like film palpatine or is it just me? if not then the lost mission episodes of the clone wars including all of the maul fight could have the audio replaced with this. so no more Tim curry at all.
- Time
- Post link
What’re the best AI voice tools right now, for emulating known character voices? Ideally pre-trained. Fakeyou has some trained voices but they’re not great.
- Time
- Post link
What’re the best AI voice tools right now, for emulating known character voices? Ideally pre-trained. Fakeyou has some trained voices but they’re not great.
Well this one is called ElevenLabs AI.
- Time
- (Edited)
- Post link
Is there a way to adjust the ai audio to get the emotions and tone etc. to sound right?
- Time
- (Edited)
- Post link
Yes, you need to add labels with the appropriate adjectives and I also recommend generating a description of the way the character speaks with ChatGPT. You can put that paragraph into the feature right below the labels.
- Time
- Post link
Yes, you need to add labels with the appropriate adjectives and I also recommend generating a description of the way the character speaks with ChatGPT. You can put that paragraph into the feature right below the labels.
So what happened in this video and this video? https://www.youtube.com/watch?v=YMDp4dyjOxw
- Time
- Post link
I think a big reason why those sound realistic is because the voice was taken right from the movies, likely using AI to isolate it. But yes, the things I mentioned beforehand would help the AI with tone and cadence.
- Time
- (Edited)
- Post link
I think a big reason why those sound realistic is because the voice was taken right from the movies, likely using AI to isolate it. But yes, the things I mentioned beforehand would help the AI with tone and cadence.
No I mean they still sound a bit off is there a way to fix those spots?
- Time
- (Edited)
- Post link
I don’t know what those creators did exactly with the label and voice description features. But in my experience they can’t salvage a voice clone, they just give you better results more often. You’d still have to do a number of generations for each line before getting something that sounds 100% good. Unless you get lucky, of course.
- Time
- Post link
I don’t know what those creators did exactly with the label and voice description features. But in my experience they can’t salvage a voice clone, they just give you better results more often. You’d still have to do a number of generations for each line before getting something that sounds 100% good. Unless you get lucky, of course.
I thought you could listen to the palpatine one and figure out how fix the parts that sound off that’s all.
- Time
- Post link
For a fan edit I’m doing of Attack Of The Clones, I’m experimenting with altering the appearance of Palpatine in the deleted scene where Padme addresses the Senate.
Since the makeup applied to Ian McDiarmid during this scene not only doesn’t match the rest of AOTC, but his appearance doesn’t match how he’s later seen in ROTS either. I’m aiming for an appearance that is more in-line with ROTS, by removing the liver spots and making him less gaunt and overly pale.
This is just a test so far, however if it goes well, I’ll expand it to alter a few other faces in different SW films.
My edits:
https://ifdb.fanedit.org/fanedit-search/tag/faneditorname/phase3/
- Time
- Post link
My goal is to add a line for the Neimodian’s protocol droid, TC-14, and replace C-3PO’s lines with a different voice (if he isn’t totally removed) in Phantom Menace. How easy to make the voices sound appropriate and not too robotic with ElevenLabs or something else?
The blue elephant in the room.
- Time
- (Edited)
- Post link
Bearded Kylo riding on a Tauntaun is the ST’s biggest missed opportunity.
Going off at a bit of a tangent here, but it’s really quite amazing what various tech-heads are coming up with regarding ‘creative AI’ at the moment. And really quite worrying too in certain ways!
The latest development which OpenAI has come up with seems to give impressive video generating results just by interpreting whatever your text instructions are.
While a Tauntaun-riding, bearded Kylo wouldn’t necessarily salvage the sequels… 😉, it sure makes me wonder how all this will evolve over the coming years… - https://mashable.com/article/how-to-try-sora-openai-video-generator
Some further examples of what ‘Sora’ can do - https://openai.com/sora
And another video about it here too - https://www.youtube.com/watch?v=NXpdyAWLDas
- Time
- Post link
Like I said in the sequel idea thread, the problem will be that you won’t be able to describe those things by name in any prompt for a software owned by a big company. Or else they risk inevitable Disney lawsuits.
We’re gonna have to wait for open-source, community-made models that have the Star Wars movies in their training data. So we’re still a number of years out from something like this being usable for our purposes.
- Time
- (Edited)
- Post link
Your points are valid of course J J B, and it will be interesting to see how all this pans out eventually.
I didn’t catch that this development had been mentioned on page 113 of the ‘Unusual Sequel Trilogy Radical Redux Ideas Thread’ yesterday, before I brought it up here - and I’ve now added the video shown by idir_hh in that thread to my previous post above.
- Time
- Post link
Okay, so I (partly) take back what I said. It looks like SORA can also accept video input and apply what is essentially just CGI over top of things based on your text inputs. Granted, you still couldn’t insert things like TIE fighters, but I could see this being extremely useful for other things like changing the environment in shots. I bet it could also remove things from a certain shot, too. This would all be extremely useful, obviously.
- Time
- Post link
What about deepfake Hayden with Sebastian Shaw in the entire prequel trilogy. Between the two, Shaw was the better actor!