Loading
Vous avez un workflow ComfyUI ? Obtenez une documentation lisible directement dans votre navigateur.Rendre un workflow lisible.Essayer maintenant ↗
Ferham from Mega Man X Command Mission, a RPG spin-off game for PS2 & Nintendo Game Cube. Saw AI art of her on the net, but no lora on civitai, so I decided to try and make one. Command Mission is a fun game and...
Source du modèle
Description de la source
Ferham from Mega Man X Command Mission, a RPG spin-off game for PS2 & Nintendo Game Cube. Saw AI art of her on the net, but no lora on civitai, so I decided to try and make one. Command Mission is a fun game and contains a weird bit of lore, X speaks about tactical reploids and combat reploids at one point, which made me think that if reploids have such classifications, what in the world was Ferham designed to be? A whip wielding babe with a sadistic streak wearing heeled boots? An official dommy mommy if there ever was one...
Usage notes for different versions:
Anima: Works with strength 1. The beach outfit has the least amount of representation in the dataset so it is the least consistent. If her wings generate with multiple colors be sure to add "red wings" to the prompt to fix that with her default outfit. Natural language helps when tags alone do not generate the outfit perfectly, in an example image I've provided I have described her unworn helmet in natural language so that it generated properly in her hands (See provided sample images for the metadata).
Sources
1 sourceArtefacts du modèle
1 artefactFerham_(Mega_Man_X_Command_Mission)Ani.safetensors
safetensors · BF16 · 33,1 MB · SHA-256 caf2ff2620d2…59eb · Civitai
TéléchargerExtraits de sources
1 extraitSource context: 489 downloads · Type LORA · Base model Anima
Illustrious: Works with strength 1. Training data contains some images of the christmas and beach outfits. So the lora can generate her in those to some degree. All in all, due to Ferham's outfit having a lot of detail the lora isn't perfect. You can try to prompt for more specific details, but don't be surprised if the armor isn't 100% identical to the official one. Adding "red armor" in addition to "pink armor" can also help. If you want to generate her with no helmet, add "helmetless" to positive prompts and "helmet" to negatives. Adding "robot ears" to negatives can also help.
PonyXL: Works with strength 1. Does not function perfectly in the PonyDiffusionV6 checkpoint, results may vary a lot in quality with that checkpoint. I recommend trying PrefectPony checkpoint for this lora.
SD 1.5: The lora does a lot better on checkpoints that lean more into 3d, as the dataset did contain a bunch of 3d model imagery, but works fine on 2d checkpoints as well. Use strengths between 0.7 - 1 if you care a lot about details. Lower strengths for me made her outfit look generic and medieval in some cases (must've been the "armor" prompt I was using). Her wings can be prompted with "mechanical wings", but they tend to be a bit all over the place, so be vary of that. Also if prompted with "mecha", "armor", "android" etc. it can be hard to try to remove her helmet. Test around for yourself what works best for your purposes. Also, no imagery of her XDiVE variant outfits was used, aside few pictures of her with her helmet off. So the outfit with the visor hat for example is not part of the lora's dataset.
Hope you like the lora!
