引言
使用今天分享的提示词在Refly.ai进行文生图需要这三个步骤:
1.在Refly.ai中上传自己或喜欢的图片
2.使用Refly.ai生成图片
3、下载图片

使用的AI工具如下
Google Gemini Enterprise:
https://business.gemini.google/home/
Refly.ai: https://refly.ai/app/wfa-o804y67b6c7hyu7rayuksnns
一、今日分享的提示词
裸眼3D万能的提示词模板,
使用建议:优先适合用Nano Banana Pro 执行
若用gemini中的图片生成,在pro版本实测会比免费版好
/IMAGINE PROMPT:REFERENCE ANCHOR (HIGHEST PRIORITY):Use the uploaded reference image as the SINGLE and ONLY source of truth.ALL generated appearances MUST be derived from this reference.The reference defines identity, structure, material, and biological reality.--------------------------------------------------AUTOMATIC SUBJECT TYPE IDENTIFICATION (MANDATORY):First, analyze the reference image and determine the SUBJECT TYPE:- Human (person, character, portrait)- Animal (real or stylized creature)- Plant (flowers, trees, leaves, organic vegetation)- Object (product, sculpture, artifact, architecture)Based on the identified SUBJECT TYPE, AUTOMATICALLY APPLY the correct consistency rules below.This decision MUST happen before image generation.--------------------------------------------------GLOBAL IDENTITY CONSISTENCY RULE (ABSOLUTE):Regardless of subject type:- The SAME subject appears in ALL positions- Identity, structure, proportions, and core features NEVER change- Pose, angle, composition, lighting MAY change- Identity, form, material, and surface logic MUST remain constant--------------------------------------------------SUBJECT-SPECIFIC CONSISTENCY LOGIC (AUTO-APPLIED):IF SUBJECT TYPE = HUMAN:- Same person, same face, same bone structure- Same age, ethnicity, expression baseline- Same hairstyle and grooming logic- Same outfit and accessories across ALL appearances- No facial drift, no outfit changeIF SUBJECT TYPE = ANIMAL:- Same species and same individual animal- Same fur/skin/scale pattern and markings- Same eye color and body proportions- No mutation, no hybridizationIF SUBJECT TYPE = PLANT:- Same plant specimen- Same leaf shape, vein structure, growth pattern- Same color distribution and organic form- No species mixing, no structural mutationIF SUBJECT TYPE = OBJECT:- Same object identity- Same geometry, materials, textures- Same design details and proportions- No redesign or material substitution--------------------------------------------------COMPOSITION STRUCTURE:Canvas:- Aspect ratio: 2:3 portrait- High-resolution editorial qualityBACKGROUND LAYER (Z=0):- Clean 3×3 grid layout (9 cells total)- Thick white grid lines separating cells (3–4px)- Bright minimalist studio environment- Deep depth of field (f/16 or higher)- NO blur, NO bokeh, NO shallow focusGRID VISIBILITY RULE:- Center cell [2,2] is COMPLETELY HIDDEN (100% occluded)- Only 8 background cells are visible--------------------------------------------------8 BACKGROUND STYLE INTERPRETATIONS:Each visible cell presents the SAME SUBJECT in a DIFFERENT editorial / visual interpretation.Identity and form remain unchanged.[1,1] High-fashion editorial (Vogue-like):- Confident, iconic composition- Sophisticated lighting[1,2] Elegant editorial (Harper’s Bazaar-like):- Refined mood, graceful angle[1,3] Natural / lifestyle editorial (Elle-like):- Casual, environmental realism[2,1] Graphic experimental (i-D-like):- Bold composition, strong shapes[2,3] Avant-garde editorial (Dazed-like):- Dynamic motion or abstract framing[3,1] Clean professional editorial (Marie Claire-like):- Structured, composed, polished[3,2] Minimalist architectural (GQ-like):- Strong geometry, restrained palette[3,3] Fashion-forward experimental (W Magazine-like):- Expressive, artistic, conceptualALL background cells:- Same subject- Sharp focus- Clear details- Even studio lighting--------------------------------------------------FOREGROUND LAYER (Z=5–10cm FORWARD):- SAME SUBJECT as reference- SAME identity and form- Hyper-realistic, dominant, full-body or full-form version- Positioned EXACTLY at canvas center- Head/top reaches canvas top edge- Base/feet reach canvas bottom edge- Occupies maximum vertical spacePOSE / ACTION:- Dynamic forward-facing presence- Strong visual authority- Direct engagement with viewerVISUAL SEPARATION:- +20% saturation vs background- Slightly sharper than background- Ultra-detailed surface texture--------------------------------------------------OCCLUSION & DEPTH RULES:- Center cell [2,2] fully hidden- Partial overlaps:* Top cell: 10–15%* Left cell: 15–20%* Right cell: 15–20%* Bottom cell: 10–15%- Overlaps must break grid lines naturally- Soft organic edges, NO hard cutout--------------------------------------------------SHADOW & DEPTH EFFECTS:Drop shadow:- Blur: 12px- Color: rgba(0,0,0,0.25)- Offset: X=6px, Y=10pxContact shadow:- Blur: 8px- Color: rgba(0,0,0,0.35)--------------------------------------------------FORBIDDEN ELEMENTS (STRICT):- Multiple identities- Identity drift- Inconsistent structure or materials- Outfit or form changes- Blur, bokeh, shallow depth of field- Missing or thin grid lines- Flat, depthless composition--------------------------------------------------QUALITY TARGET:Editorial-grade, surreal but physically believable,clear Z-axis hierarchy, strong 3D pop-out illusion,a single subject interpreted across multiple visual languages.

二、在Refly AI中生成图片
打开Refly的链接,
https://refly.ai/app/wfa-o804y67b6c7hyu7rayuksnns
上传图片,运行即可。


仅供学习和研究使用,请在下载后24小时内删除
内容来自于网络 如有不妥联系站长删除
706资源网欢迎前来投稿文章
© 版权声明
文章版权归作者所有,未经允许请勿转载。
THE END

























暂无评论内容