Canceled by Civitai new ToS [2025]
When new Version is on i write it here - I dont Upload any RL LoRA here on Civitai more!
Free Version 1.2
1.2 Changelog:
Fixed broken face for many Checkpoints
Better stable generation for AnimateDiff
Enchant Output
All Preview has no Upscaled or ADetailer on!
LoRA Weight: 0.6-0.8
Sample Promts: (Promt List see below)
medium breast, pink nipples, normal_nipples, normal areola
big breast, pink nipples, saggy breast, big areola
Try many Times to Gen your Breasts.
Nipples and Forms are in eary stage!
Triggers:
Breast Size (Cup trigger - WiP)
a-cup breast
b-cup breast
c-cup breast
d-cup breast
Breast Size (Size trigger)
flat breast
small breast
medium breast
big breast
fake breast
bimbo breast
Breast form
torpedo_breast
saggy breast
Nipple Color
pink nipples
brown nipples
Nipple Style
normal_nipples
big nipples
inverted nipples
puffy nipples
Areola Size
small areola
normal areola
big areola
!OLD! Different Breast Types (triggerwords) (Beta 1&2)
flat_tits
small_tits
medium_tits
big_tits
fake_tits
bimbo_tits
Beta 2 Recommended strenght; 0.5-0.6!
Beta 1 Recommended strength: 0.6-0.8!
small boobs note: small boobs pictures were trained by shemales and femboys as well as 21+ year old models
Description
FAQ
Comments (12)
Can we get visual comparison with Patreon version?
https://civitai.com/images/3783021
1.5B is harder to use work on 1.6 with full rework - when you dont like 1.5B so pleace wait for 1.6
With version 1.2 all faces look the same. Back to the previous version and everything is ok. I'm disappointed :(
https://i.imgur.com/JSk93fP.jpg
strange my train images has no faces ^^" i think is the training checkpoint i try to fix it.
@SysDeep Thanks for the answer. The checkpoint is Realistic Vision V5.1
@Coconut07 Can you try to see if your Prompt, has some word, that could be a Embedding?
I have had the same problem, where an Embedding were named F, so when ever i used f1.2 the face were all the same :D (and other embeddings, so check those, there are many that uses short names, or girl, woman... so when ever you use those words, it uses the embedding)
not compatible with negative embaddings
It works, but only if you focused on breast and nothing else. It changes image drastically: change face, refocus all from full shot to waist shot and closer, so, perhaps, it is better to use it only for inpainting.
The funny is: my train data has no faces ^^ i dont know why he change faces
@SysDeep i suppose, that type of breast is strongly bounded (inside SD) with body type, ethnicity, face type
@SysDeep I was just thinking the same. I did a plotXY with 12 checkpoints with a blond woman (no embedding) on a black background then I load this LoRa and with the same seed and settings + the trigger words it produces something completely out of the ordinary, including the faces.
And it messes up with the breasts totally.
I looking at your images it looks like changing booba style creates different woman as well - with the face and makeup.
So the training data is only breasts? No faces and bodies?
trigger word comes up as <lora:add_detail:1> when adding in SD... is this right?
Details
Files
Breast Helper - [Concept] - Version 1.2.safetensors
Mirrors
Breast_Helper_12.safetensors
Breast Helper - [Concept] - Version 1.2.safetensors
Breast_Helper_12.safetensors
Breast Helper - [Concept] - Version 1.2.safetensors
Breast Helper - [Concept] - Version 1.2.safetensors
Breast Helper - [Concept] - Version 1.2.safetensors
Breast Helper - [Concept] - Version 1.2.safetensors
Available On (2 platforms)
Same model published on other platforms. May have additional downloads or version variants.



















