This my first Style LoRA trained with the creations of "Double Deck". It's also provide some characters from his work, Even if this can be a pretty niche style, I'm hoping you may find what you where searching for.
Even with a very small dataset, it do work with realistic and furry style.
The keyword is 'double deck, monster girl' but you can use specific character like :
- 'mushroom girl, plant girl', 'anthro girl, frog girl', 'anthro girl, raccoon girl' for animal/plant characters
- 'robot girl', 'android', 'tesla girl' and 'assaultron' to generate robot
- 'silent hill' plus 'pyramid head' or 'bubble head nurse', which are more chaotic during generation
- 'white and black' for the original style, or 'colored picture' to force painted variant of the style
Some arguments may exist but aren't usefull enough or may break the generation, like 'multiple breasts' or others...
For the Strength of this LoRA, it seem to work OK on 1, but It work best by lowering a bit values if there is artefact during generation.
Description
Minor changed to v3.1
increase iterations per epochs to 4
add yamamura sadako, Exit 8 and demogorgon characters
Reduced total Epoch from 20 to 16
It is more stable than 3.1 at strength 1.0 on Pony, but still prefer 0.8 for this model
FAQ
Comments (7)
he is an extremly special artist for plump and monster sex fantasy
is it possible to add hasshaku-sama from his doujin works?
Hey, there is already hasshaku-sama tag in it, but yeah perhaps it miss some picture that would help get the character even better, I posted some examples to help you. Best trigger words are "hasshaku-sama, mature woman, large woman, tall woman, sagging breasts, huge breast", I may retrain the LoRA if new source get available to increase dataset
@Lazius He recently released a new doujin that is the origins of hasshaku-sama, perhaps you could use those images for more training references? I mainly want to imitate the same faces that the AI doesn't seem to capture that well.
@drpavelitscia I saw it indeed, and after checking the images I do have about the character, it seem like is it too few (3-5 images for a total of 250). I need to rebalance the number of image per character included in it so I can limit (like reduce some other character or remove close shot). I will try to see what I can do, perhaps at the end of this week, I will reorganize my dataset
@Lazius You can also add close-ups of brwasts, hands, faces, and other small details to dataset to teach lora better do this small parts while using adetailer for example
@Rating_Agent its in this version, you can check the patchnote, but the face are messy for Kissing and other stuff, but it do recognize which character you wanted. Internal shot/x-ray was also added (you can see in the 2nd picture but he generated as a part and not the whole picture, I messed with captioning some parameters)