TRAINED ON 97 IMAGES~
ACTIVATION: candyland




















One of my favorite fantasy settings! ٩(ˊᗜˋ*)و ♡
A mix of style/architecture/background~
Generating humans can work (example) but may need help to clean up the face. Add something like dirty face in the negative prompts or inpaint the face without the lora added.
HELPFUL PROMPTS:
Full background
Food focus
Scenery
No humans
Glaze
Drizzle
Drip/Dripping
Icing/Cream/Frosting
Shiny
Reflection
Glitter/Shimmer
Mountain/Tree/Building/Castle/House/Roof/Tower/Architecture (etc.)
Multicolored theme/Various colors/Colorful/Pastel colors
NEGATIVE PROMPTS:
Too many
Simple background
Cropped
Watermark
Lantern/Lamppost (When adding ''glow' or 'glowing' to positive prompts)
Spikes (When mountains and architecture are too sharp)
Border
Messy
Debris
Clutter
Web address
Text
Oversaturated
Dirty
????? face (If generating humans)
???? tree
Mountain/Tree/Building/Architecture/House/Tower (If you want 'above clouds' only)
I'll update the dataset in the near future for more varied results.
-----------------------------------------------------------
* Activation will be changed to either [candyworld] or [sweetscape] -- without the brackets.
* A larger dataset (Thinking around 130-200; For now).
* Tweaking/fixing tags.
+ Possibly removing certain images in the dataset for cleaner/less cluttered results. TRAINED ON 97 IMAGES~
ACTIVATION: candyland
1. 轉载模型僅供學習與交流分享,其版權及最終解释權归原作者。
2. 模型原作者如需認領模型,請通過官方渠道联系海藝AI工作人員進行認證。我們致力於保護每一位創作者的權益。 點擊去認領
