Signalis Style LoRA

This is a Signalis LoRA i trained recently. I've been meaning too for months, but I had it on the backburner until recently. If you havent played Signalis, go give it a purchase! It's easily one of the best games i've ever played.

It's trained on the Cutscene images from Signalis using a variety of characters, though the characters themselves are untagged since this is a general style LoRA. Many of the images used were hand edited and color corrected in GIMP for more consistent visuals, or stitched from scrolling images to create completed character portraits. It should work great for both SFW and purposes. It was also trained on my test mix model, though, it should work on others relatively well. They wont get this exact look obviously. But some decent prompting can get you close, or maybe even get better resukts,

This was designed for txt2img generation. I have yet to test its Img2img properties, but they should work in theory? The recommended weight for the tag is somewhere around .7 if you arent trying to generate a Signalis themed character specifically.

For tags:

• black : Will get you the standard Signalis look. It should work in other colors?

• Short Black hair: Most of the characters in Signalis have it, so this will get you the closest to the Replika hair look. Though, other hair styles and colors should work fine/great.

• face augements and eye augements: Will get you their face lines, nose squares, and ear pieces. Yes, I know its spelled "augments". I purposely spelled it differently so as to separate it from any previous concept of "augment" a model might have in it's training data.

• Military Uniform, White Armor, black and orange armor: Will get you the minor outfits in in the data. They wont be super accurate, but they are additions that look quite nice.

• blue eyes: Will try to get you the "Replika Eye look" to some degree.

• bandage on arm, bandage on face: Should get you the injuries Isa Itou suffered. Though its consistency is questionable.

• Pixilated: Goes without saying.