This LoRA is trained on a synthetic dataset of health bars, one full one on the top, one empty one on the bottom. However, I failed to train it to maintain that symmetry/shape. Thus it's recommended to just use img2img with some rough sketch as input.

Keywords:

• health bar

• left_aligned, centered, right_aligned

• left_tilted, right_tilted

• power_meter

• no_segments, small_segments, large_segments

• pixel_art, minimalist, sleek, ornate, cartoon, photo_realistic, abstract, vintage, surreal, fantasy, stylized

• red, orange, yellow, lime, green, cyan, light_blue, blue, indigo, violet, purple, magenta, pink, brown, gray, black, white, gold, silver, teal