JD's bazaar

[doe.1971]

she was overdressed for the tropics.

Yeah, exactly ... all that oppresive heat, almost 100% humidity makes all the clothes soaked and sticking to the skin, revealing all the naughty parts ... who'd want that?! pervs?!

[LLL]

One nice thing about AI is that makes much easier to reimagine mainstream scenes, something I loved to do ever since.

Freya Allen in the last episode of the 3rd season of Witcher get captured by bounty hunters. AI makes it look like it was supposed to be. Right now only for static frames, soon probably for the whole video flow.

Nice results, but how did you get them? With fooocus?

[doe.1971]

Nice results, but how did you get them? With fooocus?

Ihi. Use Inpaint, mask the clothes, use SDXXXL as model and generate some 10 and choose the one you like most.

[LLL]

Nice results, but how did you get them? With fooocus?

Ihi. Use Inpaint, mask the clothes, use SDXXXL as model and generate some 10 and choose the one you like most.

Would that not lead to blurring of the image around the edges of the clothes? Seems to me that especially the PotC images stay very sharp, even background details previously covered fit in perfectly.

And what about input words? Just "nude"?

[doe.1971]

no, AI will create the edges for you. As for input, for that image was nude, small breasts, skinny, light skin

[doe.1971]

As it should have been done.

[doe.1971]

.

[doe.1971]

*

[bleumune]

no, AI will create the edges for you. As for input, for that image was nude, small breasts, skinny, light skin

Hey, JD! Still struggling with Fooocus. How do you get the results where the pose is virtually the same as your original image?

As an example, I was trying to use Fooocus on the Man's Adventure magazine below, but here's an example of what I got. Is there something I should put in to restrain the image created? The prompts I used were: nude female, large breasts, afraid. I only selected the female, starting just below the wrist ropes and going down to the bottom of the page, trying to stay close to her shape.

I did several repetitions, 10 at a time, and I only got a couple that could loosely be said to be related to the original image. Your images look much more accurate.

[bleumune]

Nice results, but how did you get them? With fooocus?

Ihi. Use Inpaint, mask the clothes, use SDXXXL as model and generate some 10 and choose the one you like most.

Now I'm wondering if Fooocus works better if you start with a photo, where I started with an artist's rendition of a Damsel in Distress . . .