Prompting
#1
by Haoming02 - opened
(Compared to Anima), this model really prefers long and complex natural language prompts.
- Pro: Better prompt adherence and details
probably - Con: "Traditional" simple Booru tags produce subpar results; LLM is basically a must...
| \ | Booru (XL Style) | Booru (Anima Style) | NL |
|---|---|---|---|
| Output | ![]() |
|
|
| Positive | masterpiece, best quality, high quality, 1girl, solo, hatsune miku, vocaloid, maid headdress, maid apron, maid outfit, holding plate, serving, looking at viewer, smile, indoors, cafe, anime screenshot, anime coloring, flat color, very aesthetic, aesthetic, absurdres, highres, detailed, beautiful color, amazing quality, safe | masterpiece, best quality, good quality, absurdres, newest. 1girl, solo, hatsune miku, vocaloid. aqua hair, twintails. maid headdress, maid apron, maid outfit. holding plate, serving. looking at viewer, smile. indoors, cafe, restaurant. | Create a bright and highly detailed anime illustration of Hatsune Miku the Vocaloid, shown as a solo character happily serving while looking at the viewer. Keep her canon appearance accurate, with her aqua hair tied into twintails, vivid eyes, and a cheerful, affectionate smile with a soft blush. Dress her in a cute maid outfit, with maid headdress and maid apron, while still keeping her clearly recognizable. She should be holding a plate with food on top, surrounded by a cafe featuring tables and windows. The final image should feel warm, charming, colorful, and very expressive, while preserving softness and energy. |
| Negative | (signature), (simple background), worst quality, bad quality, low quality, name, text, (hdr, cinematic, high contrast), logo, watermark, username, bad anatomy, bad proportions, extra limbs, extra digit, extra legs, extra legs and arms, disfigured, missing arms, too many fingers, fused fingers, missing fingers, unclear eyes, censored, chat, dialogue box | blurry, jpeg artifacts, sepia, watermark, worst quality, low quality, large breasts, muscular, deformed hands, bad anatomy, extra limbs, poorly drawn face, mutated, extra eyes, bad proportions, character doll, chibi, old, early, censored, 3d, high contrast, ai-generated | Avoid blurry, low-quality, noisy, oversmoothed, distorted or messy results. Avoid bad anatomy, deformed hands, extra fingers, fused fingers, missing fingers, broken limbs, duplicate body parts, asymmetrical eyes, distorted face, warped features, wrong proportions, cropped head, cut-off body, bad framing, unreadable text, watermark, logo, signature, background clutter, compression artifacts, muddy colors, and inconsistent details. Keep the image clean, sharp, coherent, expressive, and anatomically believable without removing anime styling. |
- Sampler: Euler, Scheduler: Beta, Steps: 32, Shift: 6, CFG: 4.0
You are absolutely right, the model is completely designed for natural language or a mixture thereof.
This comment has been hidden
Any one know whether artist style is supported? How should artist be prompted?
