Sign In

Kome (VM500) Style - Pony / Illustrious

258

Updated: Aug 9, 2026

styleanimegirls

Download

1 variant available

SafeTensor

217.88 MB

Verified:

Type
LoRA
Stats

441

117

26

Reviews
Published

Jan 11, 2025

Base Model

Illustrious

Usage Tips
Strength: 0.9
Hash
AutoV2
338C51294A
Trigger Words
vm500,
default creator card background decoration
Likes - 17452

17.5K

Downloads - 111917

111.9K

Generations - 979520

979.5K

00082-205328040.jpg

Kome (VM500) Style.

I made this lora to try and reproduce the style of Kome a.k.a VM500.

Trigger: vm500,

Advised strength: 0.7-0.9

Version 2: I made this one, because I found the first version I made for Illustrious lacking. Next to that I also did some research in finding the optimal training settings (for Civitai lora trainer) together with Claude (I send Claude out to find the best data sources) and the final result is spot on.

I used a slightly different bunch of quality tags: masterpiece, best quality, traditional media, sketch, jaggy lines, I also found that flat color helps to get a more manga/doujin feel.

I also use reForge with Adetailer (only the face model), CD Turner, FreeU and SelfAttentionGuidance (only for Illustrious), but they are not a must.

Prompting Illustrious

For all versions I recommend you use: masterpiece, best quality, amazing quality, very aesthetic, high resolution, in your prompt and I myself am using the following negative prompt: text, abstract, glitch, deformed, mutated, ugly, disfigured, long body, lowres, bad anatomy, bad hands, missing fingers, extra digit, fewer digits, cropped, lowres, (worst quality, bad quality:1.2), bad anatomy, jpeg artifacts, signature, watermark, username, signature, simple background, conjoined,

The above showcase images were made with Addetailer extension enabled and using a face and eyes model.

This LoRa is trained on Illustrious 0.1, but I made the above showcase images in different checkpoints, because their style was more to my liking.

Prompting Pony

For all versions I recommend you use: score_9, score_8_up, score_7_up, score_6_up, in your prompt and using the following negative prompt: score_5_up, score_4, score_3, score_2, score_1, censored, mosaic censoring

The above showcase images were made with Addetailer extension enabled and using a face and eyes model.

This LoRa is trained on Pony V6, but I made the above showcase images in different checkpoints, because their style was more to my liking.

If you need a place to start take a look at the showcase images above.