Sign In

Flammie/フラミー (Mana Series) [PonyXL/Illustrious]

53
224
2.7k
20
Verified:
SafeTensor
Type
LoRA
Stats
72
376
124
Reviews
Published
May 11, 2025
Base Model
Illustrious
Training
Steps: 758
Epochs: 10
Usage Tips
Clip Skip: 1
Trigger Words
flammie_mana
dragon, furred dragon, feathered wings, (four wings:1.5), color tipped wings, tail, claws
white fur, orange hair, orange underbelly, blue tipped wings, blue eyes
(see model description for customization)
Hash
AutoV2
D90AFDDA18

Introduction

Flammie from the Mana series. First appearing in Secret of Mana then became a recurring character in subsequent game.

As such, this model is trained on all versions of Flammie from all the games with a mix of SFW and NSFW pictures that were used for this model. Had one of the most intensive tagging session with this model, so expect a lot of customization (SEE HOW TO CUSTOMIZE BELOW)

Using This LoRA and want to support me? Use the 'add post' in the gallery below - This gives me a lot of blue buzz to train more LoRAs!

How to Customize

Pick one prompt from each category and it'll generate accordingly:

  1. Fur: white fur, yellow fur

  2. Hair: brown hair, blonde hair, orange hair

  3. Underbelly: orange underbelly, brown underbelly, yellow underbelly

  4. Wing color: blue tipped wings, green tipped wings

  5. Eye color: purple eyes, blue eyes, green eyes, red eyes, orange eyes

For example: yellow fur, orange hair, brown underbelly, green tipped wings, red eyes


Illustrious (V1)

It should be able to generate:

  • With weightage about 0.8 to 1.

  • Anthro and Feral (Put "Feral" and "Anthro" trigger words accordingly)

  • SFW and NSFW

  • Male and Female (Put "1boy" or "1girl" trigger words accordingly)

Prompts

Trigger Words: flammie_mana

Additional prompts: dragon, furred dragon, feathered wings, (four wings:1.5), color tipped wings, tail, claws

Optional prompts for generic Flammie: white fur, orange hair, orange underbelly, blue tipped wings, blue eyes

Make sure to see at the top of the page on how to customize!

Notes

  • Model is trained on the base Illustrious model.

  • Generation quality should be about 80 - 85%+ accurate. This means you should generate mostly without much problem with proper prompting, but might still face that occasional error.

  • Faced quite a bit of issues in getting the most 'canon' look, and the hair was the hardest to get right. See my picture prompts if you need some help. Here's what I did:

    • For hair, I put in 'short hair' in positives with the hair color.

    • For negatives, I put "((Hair tufts, framing strands, medium hair, long hair, side bangs haircut, side flares, bob cut haircut:1.6))"

    • Increase weightage of prompts and LoRA as needed.

  • For the wings, include "(six wings, two wings)" in negatives if you keep generating less or extra wings.

  • LoRA is customizable with different hair color, underbelly color, etc. Make sure to check the top of the page on how to do it. Or if you're lazy, use the default prompt already included for a quick start!


Pony (V1)

It should be able to generate:

  • With weightage about 0.8 to 1.

  • Anthro and Feral (Put "Feral" and "Anthro" trigger words accordingly)

  • SFW and NSFW

  • Male and Female (Put "1boy, male" or "1girl, female" trigger words accordingly)

Prompts

Trigger Words: flammie_mana

Additional prompts: dragon, furred dragon, feathered wings, (multiple wings, four wings:1.5), color tipped wings, tail, claws

Optional prompts for generic Flammie: white fur, orange hair, orange underbelly, blue tipped wings, blue eyes

Make sure to see at the top of the page on how to customize!

Notes

  • Model is trained on the base Pony Diffusion model.

  • Model accuracy is about 80%, with lower points mainly because of the wings issue which AI has a lot of troubles with. That's because anything out of the standard 'dragon' causes a lot more consistent generation. However besides the wing issue, model quality generates very well from my testing.

  • For the wings, I put "(multiple wings, four wings:1.5)" in the prompts as this worked well for me to generate the four wings much more consistently which was around 85-90% of the time. Without the weightage, the model just seem to generate the normal two wings. I also put (two wings, six wings) in negatives as well as the multiple wings prompt can cause AI to generate unlimited wings lol. Adjust the weightage yourself higher or lower if needed.

  • For Anthro model, increase weightage for the underbelly prompts if it does not generate at all. I faced this issue and increase my underbelly prompt strength to 1.2 to generate it more consistently.

  • Certain poses may cause issues with the wings. One of the pose that I use often which is presenting back and butt causes a lot of issues with the wings (AI goes crazy with either two wings or six to seven wings) Adjust your weightage here if needed.

  • With the training, you can customize Flammie with white or yellow fur. Also with many other prompts like eye color, wing tip color, etc. See my SFW pictures for example of how I mix and match the different prompts.

  • For anthro models, use "hands" and "fingers" in prompts for more proper human anatomy (standard 4 finger 1 thumb) or "claws" if you want a slightly more feral look (typically generates 4 fingers)

  • For feral models, I tend to include these in negatives: "abs, fingers, belly button, hands, pov, human".

Want to support me without Tipping? Here's how:

  1. Liking my posts, collecting my images or collecting my models all generate buzz for me.

  2. Generate more pictures of my models. I get a base commission from anyone who generates using my model, even without the tip.

  3. Leave a like on my model. This helps me with the algorithm in reaching more people

  4. Leave a review. I take all feedback very seriously, good or bad.

  5. Consider using the 'add post' under gallery... if you are posting pictures of just my LoRA. The images will also still appear on your profile but this will give me A LOT of buzz, while you can also complete your daily post quest.

    Thank you very much!