CivArchive
    Priscilla Ricart - v2.0
    NSFW
    Preview undefined

    Priscilla Ricart is a Brazilian model and social media star based in Miami. For her TI, I used 25 images taken from her Insta account. After a failed try (using the same parameters as in Maria Guardiola's TI), I changed to a fixed learning rate (of about 0.003) and interrupted the process at only 1615 steps.

    This TI has been updated. Please check the "About this version" section!

    Description

    I'm uploading an alternate model that feels more accurate, in my opinion. I've altered my working process a little bit in the last few days. First of all, I don't use the "Use PNG alpha channel as loss weight" anymore, as I felt my embeddings were getting overtrained almost from the start. I have a more active approach now too: I begin with a flat learning rate of around 0.004 and patiently wait until I get some decent pic resembling the subject, which usually doesn't take more than 400 steps anyway. Then I interrupt the process, test the result and reduce the learning rate to maybe 0.001. Run the process for a few more hundred steps and usually I already get a decent TI at least. I guess I'll keep improving the process as I run more tests, but for the time being I'm happy with these results. Hope you like it!

    FAQ

    Comments (4)

    7016jayAlexMar 5, 2023· 1 reaction
    CivitAI

    greetings, brother, you are really a genius. You can use such a low number of steps to produce such a high-quality model. Please tell me, the tutorial on youtube says that it takes more than 3000 steps to have a good effect. How do you reduce the number of steps? Is it related to the batch size and the number of pictures? The closer the batch size is to the number of pictures and the richer the pictures are, the lower the number of steps can be set?

    SonetMar 5, 2023· 1 reaction
    CivitAI

    I love your work, bro!! Just awesome! If you have a chance and time, pay attention to this Instagram model - Stephany Gonzalez! She is amazing, I will love to see her among your works

    jycyouyas409Apr 17, 2023
    CivitAI

    hey,bro.Which base model you use to train TI?

    JernauGurgeh
    Author
    Oct 6, 2023
    CivitAI

    Just for the record, this TI hasn't been updated since March. Version 1.0 was unpublished, apparently because there were no example pics left (many were removed after the change in guidelines, months ago). I just uploaded a few pics for version 1.0 and republished it.

    TextualInversion
    SD 1.5

    Details

    Downloads
    197
    Platform
    CivitAI
    Platform Status
    Deleted
    Created
    3/4/2023
    Updated
    5/6/2026
    Deleted
    5/23/2025
    Trigger Words:
    priricart

    Files

    priricart.pt

    Mirrors

    CivitAI (1 mirrors)