Model made for making female first person perspective / point of view images
Trained on mostly Illustrious and Pony, results may vary with animagine and sdxl.
Was a pretty tough concept to get down consistently (especially sfw) without loras and heavy prompting. Use trial and error to get the pose and perspective you want with prompting, but once you get it, its pretty consistent.
Trigger words: (female pov shot, first person point of view, female pov)
I tested on my existing clothing loras with pretty good results.
For viewer only images, try not to include features on the viewer that arent visible to the self because most models will just make another person with those traits (ex: Eye color, hair color).
Most models also tend to go for a lying down pose to make sense of looking at the viewers body. Use (laying down) in the negative prompts to reduce this effect.
For NSFW, Yuri tags were used.
Can use tags like "insertive partner, top" or "receptive partner, bottom" to control viewer's actions (hit or miss on working)
Description
FAQ
Comments (7)
This is a great lora! However, it is not stable enough at present, and I look forward to your update!
这是一个非常棒的lora！只是目前还不够稳定，我非常期待你的更新！
Can you let me know what issues you're having so I have an idea on any updates?
I find that you have to tweak the prompting to get a desired pose, but its pretty stable with the pose in remixes once you do get what you're looking for. You also have to keep in mind that most models will trend towards a lying down pose because it makes the most sense to see the viewers body, whereas standing upright doesnt naturally give that view.
This is a large gallery 900+ images of fpov I collected. Maybe it could help training fpov lora. (most non ai are in the middle pages I mean those for training) https://e-hentai.org/g/3551302/8522cc3e81/
I have a couple more clothing loras I want to work on first, but I definitely want to improve the fpov one before the end of the year. Thanks for the link, I'll look into it
@RKKTZ About four months ago, you mentioned plans to improve this LORA model. May I ask about the current progress? Do you still plan to update and maintain this LORA model?
[AI translation was used]
@NanaaaNya I still plan to make a better version, but maybe later on. I need a lot of yellow buzz to generate images for this type of lora and the original was made when I could still use blue for it. Logic wise, a person wont really see their body from their point of view while standing up, so most of the poses that will generate are going to be lying down, sitting and looking down at their lap, or use really large breasts to appear at the bottom of the frame; and it'll usually require yellow for nsfw. Local generation on my computer is slow too.
This one is also for illustrious, so I want to wait a bit to see if another big checkpoint overtakes it in popularity & effectiveness like il did to sdxl.
@RKKTZ Thank you for your response. I'll look forward to the update.
[AI translation was used]
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.