stablediffusion吧 关注:27,536贴子:95,212

SD1.5主模的 介绍概论

取消只看楼主收藏回复

我个人C站SD1.5 主模资源比较全(仅限2次元)
之前就想给大家介绍下,但一直舍不得这点资源的知识。有点小气不好意思啊~
现在SD1.5末期了,藏着掖着也没有什么意义了
再开个坑 慢慢给大家说说吧


IP属地:云南1楼2024-12-21 15:39回复
    我每次会以我仓库的模型目录截图 开始讲
    第一个目录的截图


    IP属地:云南2楼2024-12-21 15:47
    回复
      SunlightMix
      早期光线特化模型,现在看就这样吧



      IP属地:云南4楼2024-12-21 15:57
      回复
        dalcefo_painting模型
        作者早早就离开C站了,我也只到V5左右的印象
        人像效果很不错,我发图你们看,图噶了 我也没办法,字白打了
        看楼层 这是第二次了,发图不易,我都被噶习惯了,还是会翻车。
        V4


        V5



        IP属地:云南5楼2024-12-21 16:06
        回复
          sunshinemix
          一样的强调光线的大模型,当年效果还是很惊艳的


          IP属地:云南6楼2024-12-21 16:12
          收起回复
            先看看是否能正常发提示词
            (masterpiece),(best quality:1.0), (ultra highres:1.0), detailed, 1 young girl, blue sundress, golden hour, detailed beautiful skin, (sitting in a car), detailed eyes
            Negative prompt: 3d, sepia, painting, cartoons, sketch, (worst quality:2), (low quality:2), (normal quality:2), lowres, bad anatomy, bad hands, normal quality, ((monochrome)), ((grayscale)), futanari, full-package_futanari, penis_from_girl, newhalf, nipplepierces, glans penis, collapsed eyeshadow, multiple eyeblows, vaginas in breasts, pink hair, holes on breasts
            Steps: 30, Size: 1536x1024, Seed: 1849473724, Model: sd-v1-5-inpainting, Sampler: Euler a, CFG scale: 7, Mask blur: 4, Model hash: c6bbc15e32, Denoising strength: 0.75, Conditional mask weight: 1.0


            IP属地:云南7楼2024-12-21 16:21
            回复
              BBMIX
              这个主模衍生了很多不同型号 有些还是非常有趣的~
              但是太多了,没用过的人根本分不清。
              图太难发了 介绍几个吧,见谅
              bbMIXALICE

              bbmixLUCI

              BBMixesNUNU

              bbmixVERONICA


              IP属地:云南8楼2024-12-21 16:33
              回复
                sample2-Dokuki
                好像有啥,但是忘了,但好歹有张图
                DiaoDaia_Mix
                当时下架了,连张图都没有 我也没办法
                sample2-Dokuk 图

                masterpiece, best quality, absurdres,high angle
                joy inui sajuna juju, pink hair, pink eyes, long hair, two side up, <lora:chara_SonoBisqueDoll_InuiSajuna_v1:0.7>
                indors frilled swimsuit, thighhighs, detached sleeves <lora:marierose:0.7>
                Negative prompt: EasyNegativeV2, bad-artist-anime, bad-hands-5, lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, jpeg artifacts, signature, watermark, username, blurry, out of focus, Blurry faces, Blank faces, bad face, Ugly, extra ear, amputee, missing hands, missing arms, missing legs, Extra feet, Missing nipples, Extra legs, Extra hands, (painting by bad-artist-anime:0.9), (painting by bad-artist:0.9), text, error, blurry, jpeg artifacts, cropped, normal quality, artist name, (worst quality, low quality:1.4)
                Steps: 38, VAE: vae-ft-mse-840000-ema-pruned.ckpt, Size: 576x768, Seed: 3695485382, Model: sample2-Dokuki, Version: v1.6.0, Sampler: DPM++ 3M SDE Exponential, CFG mode: Half Cosine Up, VAE hash: c6a580b13a, CFG scale: 9, Clip skip: 2, marierose: 5d3aedbc0e5d", Mimic mode: Half Cosine Up, Model hash: 1797e7da21, bad-artist: 2d356134903e", Hires steps: 24, Mimic scale: 7, bad-hands-5: aa7651be154c, Hires upscale: 2, EasyNegativeV2: 339cc9210f70, Hires upscaler: 4x_NMKD-Siax_200k, "EasyNegativeV2: 339cc9210f70, Interpolate Phi: 1, bad-artist-anime: 5f7bea88750c, CFG scale minimum: 4, Denoising strength: 0.57, Scaling Startpoint: MEAN, Mimic scale minimum: 4, Variability Measure: AD, Threshold percentile: 96, Separate Feature Channels: True, Dynamic thresholding enabled: True, "chara_SonoBisqueDoll_InuiSajuna_v1: 60cae3671355


                IP属地:云南10楼2024-12-21 16:45
                回复
                  betterMix_Anime
                  这个是纯纯的混合,但是效果看着还行
                  hiten的底子方便融合其它模型
                  样图看着相当不错,用着手脚都是灾难




                  IP属地:云南11楼2024-12-21 16:54
                  回复
                    camelliamix
                    这个系列的主模真的是很棒,我个人认为SD1.5,最优秀的那一批
                    哪怕现在来用也能出很不错的图片,而且作者更新力度强,一直在进步。


                    camelliamix 2.5D



                    IP属地:云南12楼2024-12-21 17:10
                    回复
                      NeverEnding Dream (NED)
                      Lykon 大佬2.5d模型 SD1.5天花板之一
                      超强泛用性,能出各种你想象到的图片,效果惊艳特,特别是脸部。
                      这张样图,我能记很多年

                      (masterpiece:1.3), (8k, photorealistic, RAW photo, best quality: 1.4), (1girl), beautiful face, (realistic face), (black hair, short hair:1.3), beautiful hairstyle, realistic eyes, beautiful detailed eyes, (realistic skin), beautiful skin, (navy blue sweater), absurdres, attractive, ultra high res, ultra realistic, highly detailed, golden ratio, dark, rim lighting
                      Negative prompt: easynegative, bad-hands-5, ng_deepnegative_v1_75t, bad-picture-chill-75v, bad-image-v2-39000, verybadimagenegative_v1.3, painting, sketches, text, cropped, jpeg artifacts, ((monochrome)), ((grayscale)), (worst quality, low quality, normal quality, blurry:1.4), (ugly, morbid, mutation, disfigured:1.3), out of frame, long neck, age spot, skin blemishes, mole, acnes, cross-eyed, unbalanced eyes, different eyes color, extra fingers, mutated hands, bad anatomy, long nails, bad proportions, extra limbs, malformed limbs, missing arms, missing legs, extra arms, extra legs, fused fingers, too many fingers, twisted legs, fused legs, long neck, username, watermark, poor facial details, bad composition, deformed body features
                      Steps: 25, Size: 512x768, Seed: 1616570648, Model: neverendingDreamNED_v122BakedVae, Sampler: DPM++ 2M Karras, CFG scale: 7, Clip skip: 2, Model hash: ecefb796ff, AddNet Enabled: True, AddNet Model 1: LowRA(0dfc93870ba3), AddNet Model 2: fte_futuanerV1(dde6f5dd5ffe), AddNet Module 1: LoRA, AddNet Module 2: LoRA, AddNet Weight A 1: 0.75, AddNet Weight A 2: 0.5, AddNet Weight B 1: 0.75, AddNet Weight B 2: 0.5, Postprocess upscaler: 4x-UltraSharp, Postprocess upscale by: 2


                      IP属地:云南15楼2024-12-21 21:34
                      收起回复
                        ReV Animated
                        基本上SD1.5里的顶点,万能主模 C站美学和下载量的巅峰。
                        这个主模实在是太好用,唯一的缺点,就是用的人太多了,
                        略微的审美疲劳。



                        IP属地:云南16楼2024-12-21 21:42
                        回复
                          Based64、Based65,
                          常用二次元主模的不会陌生,优秀的动漫风格
                          特别是画动漫角色和游戏角色,作者有多个模型,我个人主要用过这两个。
                          Based64


                          Based65



                          IP属地:云南17楼2024-12-22 13:29
                          回复
                            detailedproject
                            C站下架模型,不好找,有张图
                            侧重动漫细节的主模


                            IP属地:云南18楼2024-12-22 13:32
                            回复
                              gottaMIX
                              这个模型是复古动漫风格,曾经风光无限
                              C站不少大佬都用它出过图,我入坑的时候已经下架了


                              IP属地:云南19楼2024-12-22 13:38
                              回复