Mainly trained this to see how well the 3D style of the movie would translate into a character LoRA, the results were pretty decent! Just like the movie. Decent!
Main trigger:
Aaron MitchellYou might also want to include 'child' into your prompt for it to be more accurate to the character.
Outfit:
blue t-shirt, shoes, grey pants, white socksOr whatever else you want.
To get the actual movie's artstyle you've gotta include '3d' in the prompt.
Should work well with other style LoRAs, though if you're going for a more 2D style you should probably put '3d' in the negative prompt.
"Wait, you already have a Pony version of this, why release it separately instead of adding it to the existent submission?"
You can't really filter the additional versions through searching. Let's say you try looking for "Aaron Mitchell" and set the filter to Illustrious only. That new version won't pop up since Pony is the default for that one. I search through submissions like that, and I assume others do too.
Description
FAQ
Comments (1)
Your experience with 3D training data mirrors my own with a bunch of LoRAs (I've got 6+ where 3D is a major chunk of the training data): If you tell the trainer the 3D training data is 3D, it works fine (even if you don't include 3D as a negative at prompt). It's when you don't that things get strange. Same thing for a bunch of other styles (monochrome, low poly, realistic)
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.