This model can generate and edit characters to have a nosehook with a pulled-up piggy nose. 🐖🐽🐷💗
Oink, oink! 💗

It is trained on a large varied dataset with 91 images of females, that does not contain artificial data. ❌🤖❌ And is captioned using loose Illustrious/Pony style tokens mixed with strings for triggers. As I find this works best for face edits, as I'm not teaching a replacement function, only a output result.

Important WARNING: ⚠️🚫⛔

As you might later on notice, I did not use any terms related to fetish, bdsm, nosehooks, or anything that might cause the basemodel to refuse the output. But by using my exact trained triggers it will work perfectly. ❤️

To start open with:

"add a vertical strap over her forehead and hair.

change her expression into:

girl with pig nose face hook harness, oral examination dentist style, she looks at me, lick u tongue.

maintain the textures, colors and identity of the person.".

See example images with workflows attached, just drag'n'drop in comfyui with simple-image-saver node installed, otherwise use example prompt.

Triggers: 🔥🔥🔥

1. girl with pig nose face hook harness, (adds strap on nose and hook, but not forehead).

2. oral examination dentist style, (opens the mouth provocatively.)

3. lick u tongue, (sticks the tongue out slightly inside the lowerlip.)

4. open mouth, (open mouth wider.)

5. tongue, (adds more tongue.)

6. raw dslr photography, (helps with achieving better photo quality.)

7. perfect healthy human, (not needed, solved issue with pig traits on previous test-versions.)

8. white teeth, (creates perfect teeth.)

9. ugly teeth, (creates uneven teeth.)

(Do not add the emoji's to your prompts).

Notes: 💌💌💌

1. There is no artificial data in the dataset.

2. All preview images uploaded by me are generated using AI, without real people, some faces come from thispersondoesnotexist.com

3. If during edit the pose changes too much, prompt: keep the background and pose the same.

4. Can do different etnicities, genders, and all that, even though it was only trained on adult adult 18+ women. Don't need to specify or edit the prompt, it works on males too.

Useful Tips about Boogu: 🧠🤓🧠

Boogu works in order of sentences. So for edits, you might first need to remove a item, before adding in a new one within a single prompt. Think of it as the 2-step edit rule.
E.g:

1. remove the blonde hair.

2. add black bangs with a ponytail to the back of the head.

3. remove the collar.

4. add a necklace.

This greatly keeps the model from drawing items on top of each other, which were prompted not to be present, and allows you to transform many things at once, instead of one at a time like in the examples from the Boogu mainpage.

Known limitations: 🤷🤷🤷

1. The lora does not work on pictures of animals,

2. The lora has a hard time with the left angle, flip image vertical in workflow, as the right side does work.

3. The lora has a hard time with nose piercings.

4. The input image often needs to be at a portrait angle.

5. Very large noses need more attempts to work, or don't work at all.

6. When the distance from the character is too large, it might take many attempts, because flux VAE is very bad at small and distant images.

7. With a very large resolution it might not work, depending on aspect ratio. Portraits with 1024x1024 work perfect though.

8. It does not create explicit nudity / R-rated / R-18 type content.

🧑‍⚖️📜

Use this AI model responsibly, and do not upload any prohibited content or prompts. Follow the rules of the TOS and laws. Violations can be reported and can get your account banned. This model was not developed in affiliation with Boogu, and is a purely a hobby project.

Notes: 💌💌💌

The images I uploaded are saved using Simple-Image-Saver, which is a custom node that exports files and passes the civitai metadata filecheck with a much lower filesize, such as jpeg and webp. To use the prompts from the example images either use the node, or the prompt I manually pasted into Civitai. It can be found inside of comfyUI-manager.

I only uploaded to civitai, if you see it on another platform let me know.
I haven't tested extensively, but it's now just sitting on my drive, so I figured let's make this a "us" problem. 😂 It does work though. 😆

For the examples I used Boogu Basemodel: Boogu-Image-Turbo Boogu-Image Turbo - v0.1 | Boogu Checkpoint | Civitai , CLIP: qwen3vl_8b_fp8_scaled, VAE: Flux1\AE, inference steps: 4, sampler: euler, scheduler: sgm_uniform.

Help me power up: ⚡👍⚡

Hit the like button on the model page, and on my images if you enjoy what you see! I'm all out of buzz, and every bit helps so I can train more models! And if you download the lora, posting a image to the page grants me ⚡50 buzz per user, and also saving my images to your gallery helps me out bigtime! So I would appreciate it alot if you could spare a few seconds! Donations are welcome too! <3