Disclaimer: For Pony: add weight to "blue skin" in the 1st Form.
For Illustrious: Nothing it works without issues.
(There is also Additional Info at the bottom on how to switch between forms easier)
Phantylia
Weight: 1.0
Trigger: Phantylia
1st Form
Appearance:
Phantylia, Phan_1st, black hair, blue skin, yellow eyes, blue lips, gold headpiece, black gloves, long hair, jewelry, hair stick, fingernails, bead necklace, forehead ornament,1st Form Outfit:
gold dress, gold bra, black dress, gold trim, green skirt, halter dress, long skirt, detached sleeves, black sleeves, halterneck, navel cutout, cleavage cutout, side cutout, 2nd Form
Appearance:
Phantylia, Phan_2nd, purple hair, purple skin, black hair, blue eyes, purple lips, black headpiece, purple gloves, long hair, jewelry, hair stick, fingernails, bead necklace, forehead ornament,2nd Form Outfit:
purple dress, black dress, purple skirt, halter dress, long skirt, detached sleeves, black sleeves, halterneck, navel cutout, cleavage cutout, side cutout, 3rd Form
Appearance:
Phantylia, Phan_3rd, white hair, pale skin, orange eyes, red lips, orange headpiece, orange gloves, long hair, jewelry, hair stick, fingernails, bead necklace, forehead ornament,3rd Form Outfit:
black dress, black skirt, halter dress, long skirt, detached sleeves, black sleeves, halterneck, navel cutout, cleavage cutout, side cutout, Additional Info: I sorted the tags in a linear way, so it's easier to switch between the forms.
For Appearance switch: Number Trigger Tag / Hair Color / Skin Color / Eye Color / Lip Color / Headpiece Color / Gloves Color (or basically just every tag before "long hair")
For outfit switch: Dress Color / Skirt Color (Or just overall every tag before "halter dress")
Everything else is the same, so you don't have to touch those parts of the Prompt.
The results of your generations are dependent on the additional LoRAs, weights, and models you use, so it may not work or come out as consistent as my sample images.
Description
FAQ
Comments (2)
Wow, this Lora is incredible! I'm amazed at how accurately you've generated her looks, especially considering the limited amount of fanart available. It's truly impressive. I'm curious about the primary sources you used to create her datasets. Could you share some details about that?
I'm glad you like it. Let me share the dataset for you and explain In a nutshell: First of all a lot of screenshots of MMD videos plus Image Re-screenshotting (aka. Crop good images to increase the amount of images). Overall I managed to make 40/20/20 datasets, repeats were 4/8/8, so 160/160/160, so 480 repeats with 4 batchs, so 120 steps per Epoch. As for overall Training settings just Default settings of Google Colab Trainer by Hollowstraberry. I hope this helps. Here is the link to the dataset. https://drive.google.com/drive/folders/1-1VeyJ3RxuHIcJx4tbmP353ud9NYOPDU












