| {"ModelId":124189,"ModelName":"Anime Illust Diffusion XL","ModelDescription":"\u003Ch1 id=\u0022heading-6047\u0022\u003E\u6A21\u578B\u4ECB\u7ECD\uFF08Chinese Part\uFF09\u003C/h1\u003E\u003Ch2 id=\u0022heading-6048\u0022\u003EI \u5F15\u8A00\u003C/h2\u003E\u003Cp\u003E\u5728\u672C\u4ECB\u7ECD\u4E2D\uFF0C\u60A8\u5C06\u4E86\u89E3\uFF1A\u003C/p\u003E\u003Col\u003E\u003Cli\u003E\u003Cp\u003E\u6A21\u578B\u4ECB\u7ECD\uFF08\u89C1 II \u90E8\u5206\uFF09\uFF1B\u003C/p\u003E\u003C/li\u003E\u003Cli\u003E\u003Cp\u003E\u4F7F\u7528\u6307\u5357\uFF08\u89C1 III \u90E8\u5206\uFF09\uFF1B\u003C/p\u003E\u003C/li\u003E\u003Cli\u003E\u003Cp\u003E\u8BAD\u7EC3\u53C2\u6570\uFF08\u89C1 IV \u90E8\u5206\uFF09\uFF1B\u003C/p\u003E\u003C/li\u003E\u003Cli\u003E\u003Cp\u003E\u89E6\u53D1\u8BCD\u5217\u8868\uFF08\u89C1\u9644\u5F55 A \u90E8\u5206\uFF09\u003C/p\u003E\u003Cp\u003E\u003C/p\u003E\u003C/li\u003E\u003C/ol\u003E\u003Ch2 id=\u0022heading-6049\u0022\u003EII \u6A21\u578B\u4ECB\u7ECD\u003C/h2\u003E\u003Cp\u003E\u52A8\u6F2B\u63D2\u753B\u8BBE\u8BA1XL\uFF0C\u6216\u79F0 AIDXL \u662F\u4E00\u6B3E\u4E13\u7528\u4E8E\u751F\u6210\u4E8C\u6B21\u5143\u63D2\u56FE\u7684\u6A21\u578B\u3002\u5B83\u5185\u7F6E\u4E86 200 \u79CD\u4EE5\u4E0A\uFF08\u968F\u7740\u66F4\u65B0\u8D8A\u6765\u8D8A\u591A\uFF09\u7684\u63D2\u753B\u98CE\u683C\uFF0C\u4F9D\u9760\u7279\u5B9A\u89E6\u53D1\u8BCD\uFF08\u89C1\u9644\u5F55 A \u90E8\u5206\uFF09\u89E6\u53D1\u3002\u003C/p\u003E\u003Cp\u003E\u4F18\u70B9\uFF1A\u6784\u56FE\u5927\u80C6\uFF0C\u6CA1\u6709\u6446\u62CD\u611F\uFF0C\u4E3B\u4F53\u7A81\u51FA\uFF0C\u6CA1\u6709\u8FC7\u591A\u7E41\u6742\u7684\u7EC6\u8282\uFF0C\u8BA4\u8BC6\u5F88\u591A\u52A8\u6F2B\u4EBA\u7269\uFF08\u4F9D\u9760\u89D2\u8272\u65E5\u6587\u540D\u62FC\u97F3\u89E6\u53D1\uFF0C\u4F8B\u5982\uFF0C\u201Cayanami rei\u201D\u5BF9\u5E94\u89D2\u8272\u201C\u7EEB\u6CE2\u4E3D\u201D\uFF0C\u201Ckamado nezuko\u201D\u5BF9\u5E94\u89D2\u8272\u201C\u7962\u8C46\u5B50\u201D\uFF09\u3002\u003C/p\u003E\u003Cp\u003E\u003Cstrong\u003E\u6A21\u578B\u96BE\u5EA6\u8F83\u5927\uFF0C\u4E0D\u63A8\u8350\u5165\u95E8\u8005\u4F7F\u7528\u3002\u003C/strong\u003E\u003C/p\u003E\u003Cp\u003E\u003C/p\u003E\u003Ch2 id=\u0022heading-6050\u0022\u003EIII \u4F7F\u7528\u6307\u5357\uFF08\u5C06\u4E0E\u65F6\u4FF1\u8FDB\uFF09\u003C/h2\u003E\u003Cp\u003E\u63A8\u8350\u4F7F\u7528 ComfyUI \u751F\u6210\u56FE\u50CF\u3002\u5DE5\u4F5C\u6D41\u53C2\u6570\u5982\u56FE 1 \u6240\u793A\u3002\u003C/p\u003E\u003Cp\u003E\u003Cimg src=\u0022https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/064bf13d-9eb0-42f3-bf6e-4d1b0a64e26a/width=525/064bf13d-9eb0-42f3-bf6e-4d1b0a64e26a.jpeg\u0022 /\u003E\u003C/p\u003E\u003Cp\u003E\u56FE1\uFF1AComfUI \u5DE5\u4F5C\u6D41\u003C/p\u003E\u003Ch3 id=\u0022heading-6051\u0022\u003E1 \u63D0\u793A\u8BCD\u4E66\u5199\u003C/h3\u003E\u003Cp\u003E\u5982\u60A8\u6240\u89C1\uFF0CSDXL \u7684\u6B63\u8D1F\u9762\u63D0\u793A\u8BCD\u90FD\u5206\u4E3A text_g \u548C text_l \u4E24\u4E2A\u90E8\u5206\u3002\u5047\u8BBE\u60A8\u7528\u4E8E\u751F\u6210\u56FE\u50CF\u7684\u63CF\u8FF0\u63D0\u793A\u8BCD\u4E3A {main_prompt}\uFF08\u4E0D\u5305\u542B\u8D28\u91CF\u63CF\u8FF0\u8BED\uFF09\uFF0C\u8D28\u91CF\u63D0\u793A\u8BCD\u4E3A {quality_prompt}\uFF0C\u8D1F\u9762\u63D0\u793A\u6B21\u4E3A {neg_prompt}\u3002\u5219\u5C31\u7ECF\u9A8C\u6765\u8BF4\uFF0C\u6B63\u9762\u63D0\u793A\u8BCD\u7684 text_g \u5904\u586B\u5199 \u201C{main_prompt}\u201D\uFF0Ctext_l \u5904\u586B\u5199 \u201C{quality_prompt}, {main_prompt}\u201D\uFF1B\u8D1F\u9762\u63D0\u793A\u6B21\u7684 text_g \u548C text_l \u90FD\u586B\u5199 \u201C{neg_prompt}\u201D\u3002\u003C/p\u003E\u003Cp\u003E\u53E6\u5916\uFF0C\u4E3A\u4E86\u5F3A\u8C03\u751F\u6210\u56FE\u50CF\u7684\u98CE\u683C\uFF0C\u98CE\u683C\u89E6\u53D1\u8BCD\u7684\u4F4D\u7F6E\u5E94\u5C3D\u91CF\u63D0\u524D\uFF0C\u4F8B\u5982\uFF0C \u003Ccode\u003Eby xxx, 1girl, solo\u003C/code\u003E\u3002\u003C/p\u003E\u003Cp\u003E\u003C/p\u003E\u003Cp\u003E\u003Cstrong\u003E1.1 \u793A\u4F8B\u003C/strong\u003E\u003C/p\u003E\u003Cp\u003E\u8D28\u91CF\u63D0\u793A\u8BCD\u793A\u4F8B\uFF1A \u003Ccode\u003Ea masterpiece with best quality, detailed 4k illustration\u003C/code\u003E\u003C/p\u003E\u003Cp\u003E\u8D1F\u9762\u63D0\u793A\u6B21\u793A\u4F8B\uFF1A \u003Ccode\u003E(low quality:1.4), bad anatomy, worst quality, ugly, disgusting, fused hands and fingers, bad hands, deformed hands and fingers, blurry, lowres, malformed, morbid, overexposure, too colorful and over-saturated, chromatic aberration, jagged\u003C/code\u003E\u003C/p\u003E\u003Cp\u003E\u003C/p\u003E\u003Cp\u003E\u003Cstrong\u003E1.2 \u5DE5\u4F5C\u6D41\u003C/strong\u003E\u003C/p\u003E\u003Cp\u003E\u6211\u5728 Suggested Resources \u5185\u6DFB\u52A0\u4E86 ComfyUI \u7684\u5DE5\u4F5C\u6D41 json \u6587\u4EF6\u3002\u003C/p\u003E\u003Cp\u003E\u003C/p\u003E\u003Ch3 id=\u0022heading-6052\u0022\u003E2 \u751F\u6210\u53C2\u6570\u003C/h3\u003E\u003Cp\u003E\u5982\u679C\u60A8\u65E0\u6CD5\u751F\u6210\u4E0E\u9884\u89C8\u56FE\u76F8\u4F3C\u7684\u56FE\u50CF\uFF0C\u8BF7\u53C2\u7167\u4EE5\u4E0B\u6307\u5357\u3002\u003C/p\u003E\u003Col\u003E\u003Cli\u003E\u003Cp\u003E\u5EFA\u8BAE\u56FE\u50CF\u603B\u5206\u8FA8\u7387\uFF08\u603B\u5206\u8FA8\u7387=\u9AD8\u5EA6x\u5BBD\u5EA6\uFF09\u5927\u4E8E 1024x1024 \u4E14 \u5C0F\u4E8E 1024x1024x1.5\uFF0C\u5426\u5219\u751F\u6210\u7684\u56FE\u50CF\u53EF\u80FD\u8D28\u91CF\u4E0D\u9AD8\u3002\u6B64\u4E3A\u7ECF\u9A8C\u6CD5\u5219\uFF0C\u5373\u751F\u6210\u56FE\u50CF\u7684\u603B\u5206\u8FA8\u7387\u5E94\u9AD8\u4E8E\u8BAD\u7EC3\u96C6\u56FE\u50CF\u7684\u603B\u5206\u8FA8\u7387\uFF0C\u4E14\u540C\u65F6\u4F4E\u4E8E\u8BAD\u7EC3\u96C6\u56FE\u50CF\u603B\u5206\u8FA8\u7387\u7684 1.5 \u500D\uFF0C\u4EE5\u9632\u6B62\u6A21\u7CCA\u548C\u7578\u53D8\u3002\u4F8B\u5982\uFF0C\u672C\u6A21\u578B\u5728 1024x1024 \u603B\u5206\u8FA8\u7387\u4E0A\u8BAD\u7EC3\uFF0C\u56E0\u6B64\u60A8\u6700\u5927\u53EF\u4EE5\u751F\u6210 1024x1536\uFF08\u4EE5 2:3 \u4E3A\u4F8B\uFF09\u5206\u8FA8\u7387\u7684\u56FE\u50CF\u3002\u003C/p\u003E\u003C/li\u003E\u003Cli\u003E\u003Cp\u003E\u4E0D\u8FDB\u884C\u201CClip Skip\u201D\u64CD\u4F5C\u3002\u003C/p\u003E\u003C/li\u003E\u003Cli\u003E\u003Cp\u003E\u91C7\u7528 \u201Cdpmpp_2m\u201D \u91C7\u6837\u5668\uFF08sampler\uFF09\uFF0C\u642D\u914D \u201Ckarras\u201D \u8C03\u5EA6\u5668\uFF08scheduler\uFF09\uFF0C\u8BE5\u7EC4\u5408\u5728 webui \u91CC\u79F0\u4E3A DPM\u002B\u002B 2M Karras\u3002\u5728 8 CFG \u4E0A\u91C7\u6837 50 \u6B65\u3002\u003C/p\u003E\u003C/li\u003E\u003Cli\u003E\u003Cp\u003E\u6240\u6709\u6D89\u53CA\u56FE\u50CF\u5BBD\u5EA6\uFF08width\uFF09\u548C\u9AD8\u5EA6\uFF08height\uFF09\u7684\u53C2\u6570\u5747\u5E94\u4FDD\u6301\u4E00\u81F4\uFF0C\u5373\u6240\u6709\u5BBD\u5EA6\u4E00\u81F4\uFF0C\u6240\u6709\u9AD8\u5EA6\u4E00\u81F4\u3002\u003C/p\u003E\u003C/li\u003E\u003Cli\u003E\u003Cp\u003E\u4EC5\u4F7F\u7528\u6A21\u578B\u672C\u8EAB\uFF0C\u800C\u4E0D\u4F7F\u7528\u7CBE\u70BC\u5668\uFF08Refiner\uFF09\u3002\u003C/p\u003E\u003C/li\u003E\u003Cli\u003E\u003Cp\u003E\u4F7F\u7528\u57FA\u5E95\u6A21\u578B vae\u3002\u003C/p\u003E\u003C/li\u003E\u003Cli\u003E\u003Cp\u003E\u4F7F\u7528\u9644\u5F55\u90E8\u5206\u63D0\u4F9B\u7684\u89E6\u53D1\u8BCD\u4EE5\u6D3B\u7528\u98CE\u683C\u5316\u3002\u5728\u672A\u4F7F\u7528\u98CE\u683C\u5316\u7684\u60C5\u51B5\u4E0B\u751F\u6210\u56FE\u50CF\u7684\u8D28\u91CF\u5C06\u53C2\u5DEE\u4E0D\u9F50\u3002\u003C/p\u003E\u003C/li\u003E\u003Cli\u003E\u003Cp\u003E\u4F7F\u7528 SDXL \u652F\u6301\u7684 VAE \u6A21\u578B\u3001\u6587\u672C\u5D4C\u5165\uFF08embeddings\uFF09\u6A21\u578B\u548C Lora \u6A21\u578B\u3002\u6CE8\u610F\uFF1A\u003Cspan style=\u0022color:rgb(193, 194, 197)\u0022\u003Esd-vae-ft-mse-original \u4E0D\u662F\u652F\u6301 SDXL \u7684 vae\uFF1BEasyNegative\u3001badhandv4 \u7B49\u8D1F\u9762\u6587\u672C\u5D4C\u5165\u4E5F\u4E0D\u662F\u652F\u6301 SDXL \u7684 embeddings\u3002\u003C/span\u003E\u003C/p\u003E\u003C/li\u003E\u003C/ol\u003E\u003Cp\u003E\u003C/p\u003E\u003Cp\u003E\u003C/p\u003E\u003Ch2 id=\u0022heading-6053\u0022\u003EIV \u8BAD\u7EC3\u53C2\u6570\u003C/h2\u003E\u003Cp\u003E\u4EE5 SDXL1.0 \u4E3A\u5E95\u6A21\uFF0C\u4F7F\u7528\u5927\u7EA6 2w \u5F20\u81EA\u5DF1\u6807\u6CE8\u7684\u56FE\u50CF\u5728 5e-6 \u5B66\u4E60\u7387\uFF0C\u603B\u957F\u4E3A 1 \u7684\u4F59\u5F26\u8C03\u5EA6\u5668\u4E0A\u8BAD\u7EC3\u4E86\u7EA6 100 \u671F\u5F97\u5230\u6A21\u578B A\u3002\u4E4B\u540E\u5728 2e-7 \u5B66\u4E60\u7387\uFF0C\u5176\u4F59\u53C2\u6570\u76F8\u540C\u7684\u6761\u4EF6\u4E0B\uFF0C\u8BAD\u7EC3\u5F97\u5230\u6A21\u578B B\u3002\u5C06\u6A21\u578B A \u4E0E B \u6DF7\u5408\u540E\u5F97\u5230 AIDXLv0.1 \u6A21\u578B\u3002\u003C/p\u003E\u003Cp\u003E\u003C/p\u003E\u003Ch2 id=\u0022heading-1368\u0022\u003EV \u5BF9\u6BD4\u57FA\u4E8E SD1.5 \u7684 AID\u003C/h2\u003E\u003Cp\u003E2023/08/08\uFF1AAIDXL \u4F7F\u7528\u4E0E AIDv2.10 \u5B8C\u5168\u76F8\u540C\u7684\u8BAD\u7EC3\u96C6\u8FDB\u884C\u8BAD\u7EC3\uFF0C\u4F46\u8868\u73B0\u4F18\u4E8E AIDv2.10\u3002AIDXL \u66F4\u806A\u660E\uFF0C\u80FD\u505A\u5230\u5F88\u591A\u4EE5 SD1.5 \u4E3A\u5E95\u6A21\u578B\u65E0\u6CD5\u505A\u5230\u7684\u4E8B\u3002\u5B83\u8FD8\u80FD\u5F88\u597D\u5730\u533A\u5206\u4E0D\u540C\u6982\u5FF5\uFF0C\u5B66\u4E60\u56FE\u50CF\u7EC6\u8282\uFF0C\u5904\u7406\u5BF9 SD1.5 \u6765\u8BF4\u96BE\u4E8E\u767B\u5929\u7684\u6784\u56FE\uFF0C\u51E0\u8FD1\u5B8C\u7F8E\u5730\u5B66\u4E60\u65E7\u7248 AID \u65E0\u6CD5\u5B8C\u5168\u638C\u63E1\u7684\u98CE\u683C\u3002\u603B\u7684\u6765\u8BF4\uFF0C\u5B83\u7EDD\u5BF9\u62E5\u6709\u6BD4 SD1.5 \u66F4\u9AD8\u7684\u4E0A\u9650\uFF0C\u6211\u4F1A\u7EE7\u7EED\u66F4\u65B0 AIDXL\u3002\u003C/p\u003E\u003Cp\u003E\u003C/p\u003E\u003Ch1 id=\u0022heading-6054\u0022\u003EModel Introduction (\u82F1\u6587\u90E8\u5206)\u003C/h1\u003E\u003Ch2 id=\u0022heading-6055\u0022\u003EI Introduction\u003C/h2\u003E\u003Cp\u003EIn this introduction, you\u0027ll learn about:\u003C/p\u003E\u003Cp\u003EModel presentation (see Section II);\u003C/p\u003E\u003Cp\u003EInstructions for use (see Section III);\u003C/p\u003E\u003Cp\u003ETraining parameters (see Section IV);\u003C/p\u003E\u003Cp\u003EList of Trigger Words (see Appendix Part A)\u003C/p\u003E\u003Cp\u003E\u003C/p\u003E\u003Ch2 id=\u0022heading-6056\u0022\u003EII AIDXL\u003C/h2\u003E\u003Cp\u003EAnime Illustration Design XL, or AIDXL, is a model dedicated to generating anime illustrations.\u003Cspan style=\u0022color:rgb(60, 64, 67)\u0022\u003E \u003C/span\u003EIt has more than 200 (with more and more updates) built-in illustration styles, which are triggered by specific trigger words (see Appendix A).\u003C/p\u003E\u003Cp\u003EAdvantages: flexible composition, no sense of posing, prominent subject, not too many complicated details, familiar with many anime characters (triggered by the character\u0027s name, for example, \u0022ayanami rei\u0022 corresponds to the character \u0022Ayanami Rei\u0022, \u0022kamado nezuko\u003Cspan style=\u0022color:rgb(60, 64, 67)\u0022\u003E \u003C/span\u003E\u0022 corresponds to the character \u0022Nezuko\u0022, \u0022lucy \\(cyberpunk edgerunners\\) corresponds to the character \u0022Lucy\u0022).\u003C/p\u003E\u003Cp\u003EIt\u0027s a little bit difficult. Not recommended for beginners.\u003C/p\u003E\u003Cp\u003E\u003C/p\u003E\u003Ch2 id=\u0022heading-6057\u0022\u003EIII User Guide\u003C/h2\u003E\u003Cp\u003E(Keep pace with the times)\u003C/p\u003E\u003Cp\u003EIt is recommended to use ComfyUI to generate images rather than WebUI for higher generation speed, flexibility and even image quality.\u003Cspan style=\u0022color:rgb(60, 64, 67)\u0022\u003E \u003C/span\u003EThe workflow parameters are shown in Figure 1.\u003C/p\u003E\u003Cp\u003E\u003Cimg src=\u0022https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/064bf13d-9eb0-42f3-bf6e-4d1b0a64e26a/width=525/064bf13d-9eb0-42f3-bf6e-4d1b0a64e26a.jpeg\u0022 /\u003E\u003C/p\u003E\u003Cp\u003EFigure 1: ComfUI workflow\u003C/p\u003E\u003Cp\u003E\u003C/p\u003E\u003Ch3 id=\u0022heading-6058\u0022\u003E1 Guidance of Prompt Writing\u003C/h3\u003E\u003Cp\u003EAs you can see, SDXL Clip has both positive and negative prompts divided into two parts, text_g and text_l.\u003Cspan style=\u0022color:rgb(60, 64, 67)\u0022\u003E \u003C/span\u003EAssume that the description prompt you use to generate images is {main_prompt} (without the quality prompt), the quality prompt is {quality_prompt}, and the negative prompt is {neg_prompt}.\u003Cspan style=\u0022color:rgb(60, 64, 67)\u0022\u003E \u003C/span\u003EThe rule of thumb is, for positive prompts, filling in \u201C{main_prompt}\u201D in text_g, and \u201C{quality_prompt}, {main_prompt}\u201D in text_l. For negative prompts, filling in \u201C{neg_prompt}\u201D in both text_g and text_l.\u003C/p\u003E\u003Cp\u003EIn addition, in order to emphasize the style of the generated image, the position of the style trigger word should be as early as possible, or even be the first tag. For example, \u003Ccode\u003Eby yoneyama mai, 1girl, solo\u003C/code\u003E.\u003C/p\u003E\u003Cp\u003E\u003C/p\u003E\u003Cp\u003E\u003Cstrong\u003E1.1 Examples\u003C/strong\u003E\u003C/p\u003E\u003Cp\u003EQuality prompts: \u003Ccode\u003Ea masterpiece with best quality, detailed 4k illustration\u003C/code\u003E\u003C/p\u003E\u003Cp\u003ENegative prompts: \u003Ccode\u003E(low quality:1.4), bad anatomy, worst quality, ugly, disgusting, fused hands and fingers, bad hands, deformed hands and fingers, blurry, lowres, malformed, morbid, overexposure, too colorful and over- saturated, chromatic aberration, jagged\u003C/code\u003E\u003C/p\u003E\u003Cp\u003E\u003C/p\u003E\u003Cp\u003E1.2 Workflow\u003C/p\u003E\u003Cp\u003EI added ComfyUI\u0027s workflow json file inside the suggested resources tab.\u003C/p\u003E\u003Cp\u003E\u003C/p\u003E\u003Ch3 id=\u0022heading-6059\u0022\u003E2 Suggested Generation Parameters\u003C/h3\u003E\u003Cp\u003EIf you are unable to generate an image similar to the preview, please follow the guidelines below.\u003C/p\u003E\u003Col\u003E\u003Cli\u003E\u003Cp\u003EIt is recommended that the total image resolution (total resolution = height x width) be greater than 1024x1024 and less than 1024x1024x1.5, otherwise the generated image may not be of high quality.\u003Cspan style=\u0022color:rgb(60, 64, 67)\u0022\u003E \u003C/span\u003EThis is a rule of thumb that the total resolution of the generated images should be higher than the total resolution of the training set images and at the same time lower than 1.5 times the total resolution of the training set images to prevent blurring and distortion.\u003Cspan style=\u0022color:rgb(60, 64, 67)\u0022\u003E \u003C/span\u003EFor example, this model is trained on a total resolution of 1024x1024, so you can generate images up to a resolution of 1024x1536 (2:3 for example).\u003C/p\u003E\u003C/li\u003E\u003Cli\u003E\u003Cp\u003EDo not skip clip layers.\u003C/p\u003E\u003C/li\u003E\u003Cli\u003E\u003Cp\u003EUse \u0022dpmpp_2m\u0022 sampler (sampler), with \u0022karras\u0022 scheduler (scheduler), this combination is called DPM\u002B\u002B 2M Karras in webui.\u003Cspan style=\u0022color:rgb(60, 64, 67)\u0022\u003E \u003C/span\u003ESample 50 steps over 8 CFG.\u003C/p\u003E\u003C/li\u003E\u003Cli\u003E\u003Cp\u003EAll parameters related to image width and height should be consistent, that is, all widths are the same, and all heights are the same.\u003C/p\u003E\u003C/li\u003E\u003Cli\u003E\u003Cp\u003EDo not use refiner model.\u003C/p\u003E\u003C/li\u003E\u003Cli\u003E\u003Cp\u003EUse VAE of the model itself.\u003C/p\u003E\u003C/li\u003E\u003Cli\u003E\u003Cp\u003EUse the trigger words provided in the appendix to apply specific style(s). You can merge styles to get amazing results.\u003Cspan style=\u0022color:rgb(60, 64, 67)\u0022\u003E \u003C/span\u003EThe quality of the resulting images without stylization will vary.\u003C/p\u003E\u003C/li\u003E\u003Cli\u003E\u003Cp\u003EUse SDXL supported VAE models, text embeddings (embeddings) models and Lora models.\u003Cspan style=\u0022color:rgb(60, 64, 67)\u0022\u003E \u003C/span\u003ENote: \u003Ccode\u003Esd-vae-ft-mse-original\u003C/code\u003E is not an SDXL-capable vae; negative embeddings like \u003Ccode\u003EEasyNegative\u003C/code\u003E and \u003Ccode\u003Ebadhandv4\u003C/code\u003E are not SDXL-capable embeddings.\u003C/p\u003E\u003C/li\u003E\u003C/ol\u003E\u003Cp\u003E\u003C/p\u003E\u003Ch2 id=\u0022heading-6060\u0022\u003EIV Training Parameters\u003C/h2\u003E\u003Cp\u003EUsing SDXL1.0 as the base model, using about 22k labeled images to train about 100 epochs on a cosine scheduler with a learning rate of 5e-6 and a total length of 1 to obtain model A.\u003Cspan style=\u0022color:rgb(60, 64, 67)\u0022\u003E \u003C/span\u003EThen, using a learning rate of 2e-7 and the same other parameters to obtain model B.\u003Cspan style=\u0022color:rgb(60, 64, 67)\u0022\u003E \u003C/span\u003EThe AIDXLv0.1 model is obtained by merging model A and B.\u003C/p\u003E\u003Cp\u003E\u003C/p\u003E\u003Ch2 id=\u0022heading-1369\u0022\u003EV AIDXL vs SD1.5 based AID\u003C/h2\u003E\u003Cp\u003E2023/08/08. AIDXL is trained on the same training set as AIDv2.10, but outperforms AIDv2.10.\u003Cspan style=\u0022color:rgb(60, 64, 67)\u0022\u003E \u003C/span\u003EAIDXL is smarter and can do many things that SD1.5-based models cannot.\u003Cspan style=\u0022color:rgb(60, 64, 67)\u0022\u003E \u003C/span\u003EIt also does a really good job of distinguishing between concepts, learning image detail, handling compositions that are difficult or even impossible for SD1.5 and AID.\u003Cspan style=\u0022color:rgb(60, 64, 67)\u0022\u003E \u003C/span\u003EOverall, it is absolute potential. I\u0027ll keep updating AIDXL.\u003C/p\u003E\u003Cp\u003E\u003C/p\u003E\u003Ch2 id=\u0022heading-6061\u0022\u003EAppendix / \u9644\u5F55\u003C/h2\u003E\u003Ch3 id=\u0022heading-6062\u0022\u003EA. Trigger Words List / \u89E6\u53D1\u8BCD\u5217\u8868\u003C/h3\u003E\u003Cp\u003E\u003Cstrong\u003Ev0.1 \u0026amp; v0.2\u003C/strong\u003E: \u003Cspan style=\u0022color:rgb(193, 194, 197)\u0022\u003Eby 35s00, by 3meiji, by 5eyo, by 7nu, by 7thknights, by adenim, by agm, by ajimita, by akizero, by ame929, by anmi, by anteiru, by arutera, by ask, by atelier irrlicht, by bunbun, by caaaaarrot, by camu, by canking, by ccroquette, by chi4, by chicken utk, by chon, by cola, by cutesexyrobutts, by darumakarei, by dino, by dora, by dsmile9, by ei maestrl, by ekita kuro, by ekita xuan, by eku uekura, by fadingz, by fajyobore, by foomidori, by freng, by fuzichoco, by gesoking, by gomzi, by hachisan, by hakuhiru oeoe, by hamukukka, by haru, by hata, by hidulme, by hikinito0902, by hinaki, by hitoimim, by hitomio16, by hizumi, by homutan, by hotatenshi, by houk1se1, by hyatsu, by icecenya, by ichigo ame, by inoriac, by iromishiro, by iwzry, by jnthed, by joezunzun, by junsui0906, by karohroka, by kaya7hara, by kazari tayu, by killow, by kin, by kinta, by kishiyo, by kitada mo, by kkuni, by konya karasue, by kooork55, by kot rou020, by krenz, by kurige horse, by kuroume, by lalalalack, by lemoneco, by lm7, by lovelymelm, by lpmya, by mar takagi, by matcha, by matsukenmanga, by melowh, by menou, by midori xu, by mika pikazo, by misumigumi, by miv4t, by mochizukikei, by mogumo, by momoco, by momoku, by morikuraen, by mqkyrie, by muina, by munashichi, by muryou tada, by myaru, by myc0t0xin, by myung yi, by nack, by naji yanagida, by nanmo, by nardack, by narue, by nekojira, by netural, by nezukonezu32, by nico tine, by nikuzume, by nine, by nineo, by ninev, by niwa uxx, by nixeu, by noco, by noodle4cool, by nounoknown, by noyu, by oda non, by omutatsu, by onineko, by palow, by panp, by pikuson, by poharo, by poire, by potg, by pro-p, by qooo003, by rai hito, by rattan, by reiko, by rella, by rhtkd, by rin7914, by roitz, by ryuseilan, by saberiii, by sais, by sakiika, by samip, by sanosomeha, by say hana, by scottie0073, by senryoko, by serie niai, by seuhyo99, by shal-e, by shimanun, by shirabii, by shiraishi kanoya, by shiren, by shirentutu, by sho, by sia, by siki, by silver, by solipsist, by some1else45, by sonomura00, by sooon, by star furu, by starshadowmagic, by starzin07, by sui 0z0, by sul, by sushi0831, by suzukasuraimu, by taiki, by takumi bis, by teffish, by tidsean, by tira27, by tsukiho tsukioka, by tsvbvra, by ttosom, by tukumi bis, by uiiv, by ukiatsuya, by umaiyo puyoman, by void, by wait ar, by walzrj, by wanke, by whoisshe, by wlop, by xilmo, by yejji, by yogisya, by yohan, by yomu, by yoneyama mai, by yosk6000, by yumenouchi, by yun216, by yunikon147, by yunsang, by ziyun, by zumoti4\u003C/span\u003E\u003C/p\u003E\u003Cp\u003E\u003C/p\u003E\u003Cp\u003E\u003Cspan style=\u0022color:rgb(193, 194, 197)\u0022\u003Ev0.3 adds: by akita hika, by asaikeu, by atdan, by bannou, by bison, by bodhi, by bonnie23, by cell, by chela77, by coco1758, by ebkim, by eichi, by electrophorus, by fungi, by gekidan, by glutton, by hews, by hirohorn, by hle, by hlymoriia, by icomochi, by iumu, by jeone0, by kana dfy, by kikinoki, by kumori ufo, by kurige, by lam, by liaowen, by limuli ceey, by lirseven, by maenoo, by magotsuki, by marusin, by mechari, by minncn, by modare, by r1zen, by rag ragko, by rannou, by rolua, by rurudo, by saclia, by sai izumi, by takunomi, by tedineon, by torino, by tororoshanyao, by tsubonari, by uuuzan, by yamanokami eaka, by zumizumi\u003C/span\u003E\u003C/p\u003E\u003Cp\u003E\u003C/p\u003E\u003Ch3 id=\u0022heading-488\u0022\u003EB. Examples / \u6848\u4F8B\u003C/h3\u003E\u003Cp\u003E(Keep pace with the times) / \u4E0E\u65F6\u4FF1\u8FDB\u003C/p\u003E\u003Cp\u003E\u003C/p\u003E\u003Cp\u003E*Translation / \u4E2D\u82F1\u5BF9\u7167\u8BCD\u5178:\u003C/p\u003E\u003Cul\u003E\u003Cli\u003E\u003Cp\u003EQuality prompts = \u8D28\u91CF\u63D0\u793A\u8BCD\u003C/p\u003E\u003C/li\u003E\u003Cli\u003E\u003Cp\u003EPrompts = \u4E00\u822C\u63D0\u793A\u8BCD\u003C/p\u003E\u003C/li\u003E\u003Cli\u003E\u003Cp\u003ENegative prompts = \u8D1F\u9762\u63D0\u793A\u8BCD\u003C/p\u003E\u003C/li\u003E\u003Cli\u003E\u003Cp\u003EUsed model version = \u6240\u4F7F\u7528\u7684\u6A21\u578B\u7248\u672C\u003C/p\u003E\u003Cp\u003E\u003C/p\u003E\u003C/li\u003E\u003C/ul\u003E\u003Cp\u003EExample 1.\u003C/p\u003E\u003Cp\u003E\u003Cem\u003EUsed model version: v0.2\u003C/em\u003E\u003C/p\u003E\u003Cp\u003E\u003Cem\u003EQuality prompts: best quality, masterpiece, detailed, 8k, 4k, highres\u003C/em\u003E\u003C/p\u003E\u003Cp\u003E\u003Cem\u003EPrompts: by chi4, exusiai (arknights), 1girl, halo, solo, red hair, gloves, smile, one eye closed, white gloves, short hair, upper body, looking at viewer, short sleeves, wings, bowtie, bow, gradient background, see-through, orange eyes, black bow, black bowtie, gradient, detached wings, energy wings, grey background, shirt, white shirt, grin, official alternate costume, bangs, alternate costume\u003C/em\u003E\u003C/p\u003E\u003Cp\u003E\u003Cem\u003ENegative prompts: low quality, bad anatomy, monochrome, worst quality, noise, blurry, fuzzy, lowres, ugly, disgusting, fused fingers, bad hands, deformed hands and fingers, blurry, lowres, malformed, morbid, overexposure, too colorful and over-saturated, chromatic aberration, jagged edge\u003C/em\u003E\u003C/p\u003E\u003Cp\u003E\u003Cimg src=\u0022https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/c4fec206-1fa7-443c-82d9-1a6039387e58/width=525/c4fec206-1fa7-443c-82d9-1a6039387e58.jpeg\u0022 /\u003E\u003C/p\u003E\u003Cp\u003E\u003C/p\u003E\u003Cp\u003EExample 2.\u003C/p\u003E\u003Cp\u003E\u003Cem\u003EUsed model version: v0.2\u003C/em\u003E\u003C/p\u003E\u003Cp\u003E\u003Cem\u003EQuality prompts: best quality, masterpiece, detailed, 8k, 4k, highres\u003C/em\u003E\u003C/p\u003E\u003Cp\u003E\u003Cem\u003EPrompts: by chon, 1girl, japanese clothes, kimono, solo, hand fan, bow, lemon, holding, jewelry, obi, lemon slice, upper body, white background, long sleeves, sash, drinking straw, wide sleeves, food, hair ornament, looking at viewer, fruit, earrings, simple background, white bow, bangs, yellow nails, blue eyes, cropped torso, cup, striped, nail polish, hair bow, closed mouth, hand on own face, blonde hair, multicolored hair, hand up, signature, mask, piercing, folding fan, holding fan, braid, checkered clothes, short hair, hand on own cheek, brown hair, bracelet, ear piercing, drink, ice cube, hairclip\u003C/em\u003E\u003C/p\u003E\u003Cp\u003E\u003Cem\u003ENegative prompts: low quality, bad anatomy, monochrome, worst quality, noise, blurry, fuzzy, lowres, ugly, disgusting, fused fingers, bad hands, deformed hands and fingers, blurry, lowres, malformed, morbid, overexposure, too colorful and over-saturated, chromatic aberration, jagged edge\u003C/em\u003E\u003C/p\u003E\u003Cp\u003E\u003Cimg src=\u0022https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/c774861b-778e-4729-9f8f-c53ee7f511bf/width=525/c774861b-778e-4729-9f8f-c53ee7f511bf.jpeg\u0022 /\u003E\u003C/p\u003E\u003Cp\u003E\u003C/p\u003E\u003Cp\u003EExample 3.\u003C/p\u003E\u003Cp\u003E\u003Cem\u003EUsed model version: v0.2\u003C/em\u003E\u003C/p\u003E\u003Cp\u003E\u003Cem\u003EQuality prompts: best quality, masterpiece, detailed, 8k, 4k, highres\u003C/em\u003E\u003C/p\u003E\u003Cp\u003E\u003Cem\u003EPrompts: by dino, by chi4, by nixeu, 1girl, breasts, eyewear on head, long hair, cleavage, solo, skirt, choker, smile, looking at viewer, thigh strap, brown hair, large breasts, red eyes, o-ring, off shoulder, school uniform, two side up, neckerchief, sunglasses, pleated skirt, shirt, red neckerchief, serafuku, bangs, grey skirt, hair between eyes, blush, o-ring choker, black choker, collarbone, sailor collar\u003C/em\u003E\u003C/p\u003E\u003Cp\u003E\u003Cem\u003ENegative prompts: text, watermark, signature, logo, low quality, bad anatomy, monochrome, worst quality, noise, blurry, fuzzy, lowres, ugly, disgusting, fused fingers, bad hands, deformed hands and fingers, blurry, lowres, malformed, morbid, overexposure, too colorful and over-saturated, chromatic aberration, jagged edge\u003C/em\u003E\u003C/p\u003E\u003Cp\u003E\u003Cimg src=\u0022https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/46687472-760a-4aa2-85d2-3c44137d8157/width=525/46687472-760a-4aa2-85d2-3c44137d8157.jpeg\u0022 /\u003E\u003C/p\u003E\u003Cp\u003E\u003C/p\u003E\u003Cp\u003EExample 4.\u003C/p\u003E\u003Cp\u003E\u003Cem\u003EUsed model version: v0.2\u003C/em\u003E\u003C/p\u003E\u003Cp\u003E\u003Cem\u003EQuality prompts: a masterpiece of highest quality, detailed 4k illustration\u003C/em\u003E\u003C/p\u003E\u003Cp\u003E\u003Cem\u003EPrompts: by yoneyama mai, by chi4, 1girl, solo, looking at viewer, long hair, black background, blue eyes, upper body, white hair\u003C/em\u003E\u003C/p\u003E\u003Cp\u003E\u003Cem\u003ENegative prompts: noise, grit, dull, washed out, low contrast, blurry, deep-fried, hazy, malformed, warped, deformed, (low quality:1.4), bad anatomy, worst quality, ugly, disgusting, fused hands and fingers, bad hands, deformed hands and fingers, blurry, lowres, malformed, morbid, overexposure, too colorful and over-saturated, chromatic aberration, jagged edge\u003C/em\u003E\u003C/p\u003E\u003Cp\u003E\u003Cimg src=\u0022https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/11568b4e-140d-4cd3-8d4b-1ebff98e1679/width=525/11568b4e-140d-4cd3-8d4b-1ebff98e1679.jpeg\u0022 /\u003E\u003C/p\u003E","Nsfw":false,"Tags":["base model"],"ModelType":"Checkpoint","VersionId":164729,"VersionName":"v0.41","VersionDescription":"\u003Cp\u003Ev0.41 \u7248\u672C\u660E\u663E\u4F18\u4E8E v0.4\u3002\u003C/p\u003E\u003Cp\u003E\u6211\u624B\u52A8\u79FB\u690D\u4E86 AIDv2.10 \uFF08\u4E00\u4E2A SD1.5 \u6A21\u578B\uFF09\u7684\u6587\u672C\u7F16\u7801\u5668\u5230 SDXL \u4E0A\uFF0C\u5E76\u878D\u5408\u4E86\u5176\u4ED6\u4E00\u4E9B\u975E\u5E38\u597D\u7684\u52A8\u6F2B\u6A21\u578B\uFF0C\u57FA\u4E8E v0.4 \u5236\u4F5C\u4E86 v0.41\u3002\u003C/p\u003E\u003Cp\u003Ev0.41 is much better than v0.4.\u003C/p\u003E\u003Cp\u003EI manually transplanted the text encoder from AIDv2.10 (an SD1.5 model) to SDXL and merged some other great anime models to make v0.41 based on v0.4.\u003C/p\u003E","BaseModel":"SDXL 1.0","FileMetadata":{"fp":"fp16","size":"pruned","format":"SafeTensor"},"ImportedAt":"2023-10-11T08:36:02.9102996+00:00","Hashes":{"SHA256":"9DA16D6EB46A6CEA2E5A3C6D6712F39E47FAA9193D247879BEBEE7D8B9C27906","CRC32":"77861012","BLAKE3":"55E0AB0D2247D976773DA9A4CB4EB2C669924690C3E3E89A0BE9A3A4B3F60271"},"UserTitle":null,"ThumbnailImageUrl":null} |