Fan Art AI Something Unlimited AI Fanart Thread - Show Us Your Skills!

[Jimwalrus]

So is there another model which can make something unlimited like art other then pony? I know about lora but not working with pony. I was able to make lora scarlet from ff7 in pony I think.

There's loads of SD1.5 Checkpoints that work well with the SD1.5

You must be registered to see the links

LoRA -

You must be registered to see the links

, 4.5 or 5 seem to be best, but most cartoon-focused Checkpoints give decent results.

In case you don't know, Pony is just a (very good, porn-focused) Checkpoint for SDXL. Many people have used it as the training base for their SDXL LoRAs. Pony-trained LoRAs work in other SDXL Checkpoints, and non-Pony SDXL LoRAs will work in Pony. But neither is ideal.

However, SD1.5 LoRAs are completely incompatible with SDXL Checkpoints and vice versa.

 

[fiendkingg]

anyone can make one from Mala?

 

[CaptainBlt]

 

[gozer123]

Can anyone tell me how to create a line of people receding in perspective using sdxl?

 

[RiderofDeath]

View attachment 3748939

How about some Wonder Woman and Black Manta action?

 

[Jimwalrus]

"Hey Rou, I could really use some ready cash right now - maybe I could be a stripper?"

"Jinx, honey, I get my strippers for free! And I keep their tips."
"Well, OK. How much do the waitresses get paid, then?"
[Laughs]"Aw, that's cute! They pay me a grand a night each to work here."
"?"
"Well, for a girl with the right... shall we say 'attitude towards her customers', it can be extremely lucrative."
"Oooh! How much we talking?"
"It depends, but the top girls are taking home twenty big ones a night. And with you being a 'Special Guest Attraction' I'd put you in the VIP Room, so the sky's the limit..."


"Welcome to the Glamor Slam VIP Room, Sir."
"I'm Jinx, I'm here to service you. I'm sorry - 'I'm your server for this evening'."



"We don't actually have a cocktail called a 'Blowjob', Sir - but I'm never one to disappoint..."



"Welcome back, Sir. The usual?"



"You up for a little more fun, Sir?"



"For a tip like that, Sir, you can finish wherever you want..."

 

[CaptainBlt]

 

[Sdrs141]

View attachment 3755907

The other Artemis I love

 

[Racheakt]

Wow, really liked these:

View attachment 3686240

MFCG PDXL proceeding some good stuff on Yodayo

View attachment 3671521 View attachment 3671522 View attachment 3671524

View attachment 3740362

Since we already have Starfire and Blackfire in Star Wars slave outfits, would anyone be willing to put either of them (or both) in Oola's fishnet sling bikini?

Love me a slavegirl on a leash. Especially with the boob slip.

 

[Jimwalrus]

Wow, really liked these:





Since we already have Starfire and Blackfire in Star Wars slave outfits, would anyone be willing to put either of them (or both) in Oola's fishnet sling bikini?

Love me a slavegirl on a leash. Especially with the boob slip.

Unfortunately the only trained models I can find for Oola's outfit are for Oola herself, not just her outfit.
Best I can manage for Starfire is:


As you can see, it's right on the limit of 'falling apart' in terms of rendering a decent image.


As for generating Oola herself in an SU style, no problems there (although the outfit has been "SR7-ified" and made a lot simpler):

 

[Jimwalrus]

Lois Lane has been poached from the Daily Planet by a local TV news station, owned by LexCorp...

"Seriously, you actually expect me to not wear a bra? And this skirt is ridiculous!"


"I don't care how many viewers you think it'll bring in, I think it's degrading!"


"OK, fine. I'm a professional. But I would like it noted that I'm not happy. Especially at how this skirt rides up..."


"This is Lois Lane for K-LEX Evening News reporting from Chinatown..."


"Our ratings are how high? Wow! And my pay is directly linked to the ratings? Maybe I should take it a little further."




"And now on K-LEX, we go live to our very own 'Luscious Lois Lane' for a special topless report."
"Thanks Mike. You join me here in the suburbs, on an unseasonably warm evening, where I am about to interview Bizarro. Which is likely to go about as well as you might expect."

 

[mypossiblepasts]

Spoiler

You don't have permission to view the spoiler content. Log in or register now.

 

[bayoassman]

I know beggars can't be choosers, but could you guys try to make more widescreen gens?

 

[Jimwalrus]

I know beggars can't be choosers, but could you guys try to make more widescreen gens?

Sure, although it's a little more complex than just setting it to a Landscape* aspect ratio and getting a full length woman standing in a wide shot. SD in particular will try to fill the field of view with the subject. So she could well be lying down, or super close-up or whatever. A bit annoying, but there are tools available that can be used to force it.

Spoiler: *Dad joke

You don't have permission to view the spoiler content. Log in or register now.

 

[Boetharch]

 

[CaptainBlt]

 

[CaptainBlt]

 

[Racheakt]

Unfortunately the only trained models I can find for Oola's outfit are for Oola herself, not just her outfit.
Best I can manage for Starfire is:
View attachment 3757677
View attachment 3757676
As you can see, it's right on the limit of 'falling apart' in terms of rendering a decent image.


As for generating Oola herself in an SU style, no problems there (although the outfit has been "SR7-ified" and made a lot simpler):
View attachment 3757681
View attachment 3757679 View attachment 3757680 View attachment 3757682

Really good!

Yeah, it's not got a ton of good samples to work with... I wonder if you'd get better results using the sling outfit... uh... "Classic" I think it is in SU?

Uh, there we go. Wonder if some of her other poses might work better too...

Anyway, looks good. Thanks for trying it does look good.

 

[Jimwalrus]

Really good!

Yeah, it's not got a ton of good samples to work with... I wonder if you'd get better results using the sling outfit... uh... "Classic" I think it is in SU?

Uh, there we go. Wonder if some of her other poses might work better too...

Anyway, looks good. Thanks for trying it does look good.

Actually, that's a good idea - I wasn't using img2img, I was just using straight text2img which uses only the trained models and text prompts rather than a reference 'starter' image. Using those images and converting them in img2img is much more likely to work. I'll give it a go later!

 

[deadshots2842]

View attachment 3761897 View attachment 3761911

Which model used?