Time for experiments!
This time, I've trained two versions of Candy, using the same prompt to create the source material, but two different checkpoints, namely JibMix Pony Realistic and Cyberrealistic v5. Both are awesome models, by the way.
This version was trained on images prompted using CyberRealistic v5.
Please let me know which you like best, and remember, likes are nice, comments are better!
Description
FAQ
Comments (8)
Nice!
Glad you like it! You've made an awesome model, there.
@tvange365 I must unfortunately admit that I always use the default model when creating LoRa and embeddings. I know - crazy... :)
@tvange365 love te results
@Cyberdelia Oh, so do I. But my MXAI series is trained on purely prompted images, and to create those, I used CyberRealistic. You're welcome to see the originals, they're in the training data.
@tvange365 i have seen it and it's a really interesting idea!
@Cyberdelia Yeah, I stopped doing celebs entirely. I had a friend who was the victim of revenge porn, and knowing what I'm able to do with even a couple of pics left a bad taste in my mouth. So I only make fake people these days. Sometimes by mashing a few embeds together, sometimes by prompting, sometimes by training an embed on 100 different people.
Nice girl, I like.
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.