Experiment - Chibi LoRA test

There isn't much to this one: Find an old work (linked in the comments) and put a chibi LoRA on its prompt and see what happens. This one was one of the better outcomes. I was using a low-cost model, so maybe you'll get better results with other models/adjusting Bespoke's settings. I also experimented with the LoRA strength slider...and honestly, the default 0.7 strength gave the most appealing outcomes with less hallucinations/errors. That's the technical side. I also wanted to share this because I thought it came out really cute. I hope you enjoy it as much as I do.

Veröffentlicht am Aug 22, 2026

orcland2048

Mehr zu sehen

simple background, 1girl, profile, child, chinese clothes, red dress, black shorts, double bun, high visibility vest, black hair, brown eyes, sneakers, bandaid on knee, looking up, v shaped eyebrows, shouting, holding megaphone, pointing up, chibi, solid oval eyes, masterpiece

Parameter anzeigen

Ursprüngliche Eingabeaufforderung: simple background, 1girl, profile, child, chinese clothes, red dress, black shorts, double bun, high-visibility vest, black hair, brown eyes, sneakers, bandaid on knee, looking up, v-shaped eyebrows, shouting, holding megaphone, pointing up, ,chibi, solid oval eyes
Größe: 768 x 1280
Sampling-Schritte: 8
Sampling-Methode: DDIM
CFG scale: 3
Negativ: bad quality, worst quality, bad anatomy, sketch, jpeg artifacts, ugly, poorly drawn, blurry, censored, artist name, signature, watermark

Prompt-Assistent

Verwendetes Modell und LoRA