Updated: Aug 21, 2026
styleDownload
1 variant available
bf16 SafeTensor
mage-flow-LoRA-nder72Run2EPOCH100.safetensors
BF16, good balance • 225.04 MB
Verified: a month ago
140 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9
(2)
Aug 21, 2026
MageFlow

1700 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9
6950 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 90 1 2 3 4 5 6 7 8 9
I may be going crazy
Please support the original artist: N_der72
Trigger word is: @nder72
This is a highly experimental, highly unstable, lowkey no highkey sucky lora trained on suboptimal config on the sucky Mageflow 4B base model by Microsoft. But even despite the low quality of the base model itself, through this lora (and the other one), I've been able to conclude that there is potential to the model. It managed to learn the major aspect of the style fairly well, as well as adapting completely to tag based prompting and shifting from realism to anime within just 100 epoch on a LoRA, which is quite unprecedented. The model is also 25% faster than Anima while being 2 billion parameters bigger.
If you want to use Mageflow in its raw state and this lora for some reason and want to know how to install the model to try, visit Comfy-Org/Mage-Flow
This is the precursor to a bigger project.
Disclaimer
I respect the artists who honed their crafts for years to achieve these styles among all else, if any artists whom LoRA i've made from shown displeasure/objection to their arts being used as AI training material or if they request a takedown notice on Danbooru , I'll delete the LoRA with no further notice.

