You don't know what a Refmod is? Check FourBunny's post https://civarchive.com/models/2993593/bunny-h3-character-refmod-collection?modelVersionId=3394914
⚠ New 11k Tokens Max - compact version :
After trying a compact 11k tokens versions with 18-20 latent images at lower resolutions (half base resolution), I'll release them as principal versions. This version grants you more references than 25k versions with same stability and less memory usage.
New compact version are labelled "VX".
Previous versions (5-25k) are all joined in "All - HighRes" version label, depending on the usage, I will not add new HighRes Refmods.
I sincerely apologize for the multiple updates and changes that have required redownloading the Refmods over the past few days. 🙏🏻
- Extract files into \models\refmods
- Use a WF with Refmods like DaSiWa Mythic Alchemy
VX Refmods:
Model's thumbnails give you all the info (except) the text description:
HOLO FRAMES FOR VX's
Total images present in latent (bottom left)
Voice or No voice (bottom right)
Miikki:
Miikki speaks Japanese and uses a Japanese accent when speaking English.
Alice:
Alice speaks French and uses a French accent when speaking English.
Iris:
Iris speaks English and uses an Irish accent when speaking English.
Meva:
Meva speaks Spanish and uses a Spanish accent when speaking English.
Olivia :
Olivia speaks Italian and uses an Italian accent when speaking English.
Nala :
Nala speaks English with a strong African accent.
HighRes Refmods:
Model's thumbnails give you all the info (except) the text description:
Total tokens (top left)
Total images present in latent (bottom left)
Voice or No voice (bottom right)
Miikki:
Miikki speaks Japanese and uses a Japanese accent when speaking English.
Alice:
Alice speaks French and uses a French accent when speaking English.
Iris:
Iris speaks English and uses an Irish accent when speaking English.
Meva:
Meva speaks Spanish and uses a Spanish accent when speaking English.
Olivia:
⚠️No 5K version
Olivia speaks Italian and uses an Italian accent when speaking English.
Credits
RefMod nodes: Luisacaotica / ComfyUI-MiniMaxH3Mod.
Description
REFMOD — TECHNICAL SHEET
Type: Bundle
Video clips: 0
Voice included: Yes
Audio: 1 sample, 10–12 seconds
Language: Italian, with an Italian accent when speaking English
FAQ
Comments (3)
Hello!🪵🐸🪵
Do you use the head as a reference apart from the body? Just so you know, several of those characters have huge head/body ratio. Especially the last two
@FrogOnStilts Hi! All the girls were created using Krea2, with the same settings.
They are all full-body shots.
However, for the last two, since I insisted on a specific bust size, some LoRAs might have enlarged the heads a bit. Their image set was also less varied. I admit it doesn't bother me, but since I spend hours looking at images of the same models, I'm likely biased.
@FrogOnStilts Thank you for the feedback 👍🏻. I'll try to correct this default during the week.