SDXL LoRA trained with Civit-AI trainer for Pony V6/Pony merges, trained on the style of Blithedragon. A furry artist known for both pokemon, MLP and occasional franchise and dragons. Typically female or futa.
Trigger words 'blithedragon'
Sample Starter:
<lora:Blithedragon_Style_PonyXL:0.8> blithedragon, score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up, source_furry,
Description
New workflow to correct a problem during training. Captioning has been improved. LoRA retrained and much more responsive and style accurate now
FAQ
Comments (2)
Very interesting lora. I've been working with styles myself and I wonder about the needed settings to make an effective style lora like this.
I assume it's just a matter of trail and error since from experience, not every artist's style will be translated well through CivitAI's trainer, even if you got the "right parameters" set up. Though, I feel that's more on my end.
Yeah it can be hit or miss, and tagging is important too. Having too many unrelated words or such can clutter and mess with concepts. I usually try to aim for 30ish epochs if I can though usually that hits the civitAI trainer's cap of 10,000 steps
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.