Type | |
Stats | 1,098 152,994 |
Reviews | (250) |
Published | Jun 5, 2024 |
Base Model | |
Training | Steps: 1,300 Epochs: 10 |
Usage Tips | Clip Skip: 1 |
Hash | AutoV2 3FC37F41E0 |
July 16, 2024: I've launched a new version of this LoRA. Check it out!
KFT Fur / Scale / Scene Enhancer [PDXL]
Built up from 650 images using the Bing/Dall-e3 generator - mostly focused on creating fantasy furry/scaly characters with detailed backgrounds. Should help enhance general character prompts to make more dynamic individuals and environments, and serve as a basis for starting new prompts.
Show me your best examples using this model for potential buzz tips!
LORA Strength
Recommend a general LORA strength of 0.5-0.9, beyond that starts getting a little crazy (but fun) - click Show More for examples.
The higher strength given, the more detailed the character, environment, and post-processing effects will be, but it will generally force a closer character view as well - per examples. A lower strength (e.g. .25-.5) should help enhance most general prompts to provide good enhancement.
Example 1: Dragon-crocodile-wolf woman sun bathing with pink hat/bikini and tropical drink
Example 2: Skinny lynx girl picking flowers in garden
(Side note: It also tends to add pools and starts wanting to put the models in water at really high strength - lots of training photos included those to bypass the bing censors)
Prompt Tips
Doesn't require any specific prompting, but it has been trained on certain concepts more than others. Examples:
Dragon-wolf anthro hybrid female
Mature rugged curvy crocodile
In tropical resort, wearing bikini, holding tropical drink
Skeletal wolf witch
On pirate ship, wearing pirate hat
In dark cave, wearing witch hat, casting spells
Smoking cigarette in swamp
Have fun with it!