Recent content by Casshern2

I see your liking that outfit and model!

Is this in general or in response to a post?

I got stuck trying to make these cuties bounce around, BUT, there is more to sexy ladies than bouncing boobs, right! I should experiment with other things...but boob bouncing is nice :cool: These are the cuties from the other thread...

Did a few cuties today.

I'm still experimenting with Wan2.2, anyone else working with this or other models?

Your checkpoint looks to be coming along just fine! Great work! :D

OUTSTANDING! Thanks for sharing this! She came out beautiful!

I could NOT get a guy in a prison gray pajama outfit like I see in those opening scenes. But, the general idea was fun getting at least something almost similar... https://mega.nz/file/aNFQibxA#UCWe9IULH3gcQ5y_8zdLjey9ZB0lxPcvKoHdjOhSl8o

Very nice!!! How are handling consistent characters?

They came out pretty good! At some point I want to try for a clip that resembles what kharo88 was looking at doing. But after more experiments :cool:

Wan is so much better at prompt adherence. Again, using the NSFW model, these were using the simple prompt of "A sexy Japanese woman stands bent over facing away, shaking her ass up and down" produced this. Of course, it doesn't follow the prompts with 100% accuracy, but close enough to make...

Ran another batch, here were some standouts (for me) :D For those interested... Checkpoint: cyberrealisticPony_v125 Prompt: score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up, source_cartoon, BREAK, (older:0.6), (Japanese:1.8) aunt, Rio Hamasaki, bending over for her neighbor...

This one came out very nice, I think. If you have a vertical monitor, enjoy this one,I know I am!

Also, to clarify, the process would be to train a LoRA to produce the kinds of images you want first. Then you would use those images+prompts to try to create 5sec videos. Entire scenes are difficult even for the very large Graphics Cards. There are workflows to extend videos beyond 5sec but...

Here is another example of the Wan all-in-one Image-to-Video (I2V) model with NSFW literally in its name. This one can adhere to prompts a bit better for different actions, but you'll see one that doesn't exactly follow the prompt. It happens, and this was against some images using a pretty...