POV

こばと🕊️_ダヴ(DOVE)·3 months ago

I don't think that's it. It’s true that doing things that resemble voyeurism in POV shots is a no-go, so it might be "out" in that sense, but fundamentally, V3 is an image-based model while WAI is an SDL-based model. I think the meaning isn't getting through because it's a model that uses so-called "tagging" like (masterpiece), (smile:1.2). Conversely, if you use (score_7_up) in V3, you just get random text 😂✌🕊️

まれん2OP·3 months ago

I see.
However, I don't think it would say "inappropriate" if it just didn't understand the meaning.
Since I think that's a message that appears when there is inappropriate content, I assumed that such an image must have been generated.

But you're right, the two models do have different characteristics, don't they?🤔
By the way, in Animji it became a girl, and in Anim4gine it became something holding a mysterious object.

66·3 months ago

こばと🕊️_ダヴ(DOVE)·3 months ago

I don't understand 🤔 A mystery? ✨ But it's funny 😂👍🕊️

😄1

たぬー·3 months ago

It makes you want to try it out, the way one mystery leads to another.

👍1

ね

ねむ·3 months ago

I tried it too!
It didn't result in a first-person perspective, so I don't think it has "POV" = "first-person perspective" ingrained in it 🤔
For example, when I entered "POV, looking up from under a tree with right hand reaching upwards" into WAI, it produced this image. It's cute, though...

❤️1

まれん2OP·3 months ago

Thank you very much.
It's very lovely, but it’s not POV, is it? 🌳
It didn't quite get across yet... I'll look forward to next time!

❤️1

こばと🕊️_ダヴ(DOVE)·3 months ago

@まれん2

こばと🕊️_ダヴ(DOVE)·3 months ago

@こばと🕊️_ダヴ(DOVE)

In WAI, POV shots come out normally for me, though…🤔🕊️

😕1 👀1

たぬー·3 months ago

Hmm, this is just my personal opinion, but it doesn't seem like it's reacting. I think it's generating images, but it looks like it's carrying over data like an afterimage of the pictures I wrote about previously, and I suspect that's why different images are appearing here. I'm not sure if it's the seed or the cookie, but perhaps if there's any information being referenced that shows both men and women, or swimsuits, it might be more likely to become inappropriate (my own theory).

👀1

まれん2OP·3 months ago

Huh.
I was wondering why only V3 seemed to be carried over so much, but I wonder if the others are like that too.

たぬー·3 months ago

The previous creation of male feet and a sunset landscape was inappropriate.
Without any people or relevant subject matter, the image turned out like this.
For both, the previous creation used only POV as the prompt.

まれん2OP·3 months ago

Maybe it got caught up in the composition as well.
It’s invisible, so everything remains a mystery, doesn't it~🤔
I guess it really comes down to entering the details properly.

たぬー·3 months ago

@まれん2

That's true. It remains a mystery, but I suppose it means to make sure to configure it properly.

👍1

さ

さな_全人類ケモナー化計画·3 months ago

It’s just a gut feeling, but I get the impression that the "inappropriate" error also pops up when the composition is physically impossible. When you try to generate a POV on its own, if there's a girl or someone else who could also be the POV, the AI tries to make it her POV. But if you've specified parts of that girl—like her eyes or hair—that would never be visible from a first-person perspective, it seems like the AI flags it as an impossible composition and throws an "inappropriate" error. So, it's likely that you'll get errors unless you explicitly specify whose POV it is, like "male POV."

👍3