欢迎您的到来! 您还不是本站会员,请:注册登录
人气 335

[ANYTHING任何事物] 【CHECKPOINT】SDXL Unstable Diffusers ヤメールの帝国 YamerMIX[DDA8C0514C] [复制链接]

种下一粒种子 ~ 2024-2-7 19:28
SDXL Unstable Diffusers ヤメールの帝国 YamerMIX万物混合
文件名称:
Unstable Diffusers V11 + RunDiffusion
资源大小:
程序版本:
SDXL
资源类型:
风格 真实感 游戏角色 女性 
资源来源:
公共

马上注册,结交更多好友,享用更多功能,让你轻松玩转社区。

您需要 登录 才可以下载或查看,没有账号?立即注册

×

资源详情:
LoRA: XL YAMER'S STYLE ♠️ Princeps Omnia LoRA
Restore Face: It is not needed to generate high quality images (Not used in the current showcase.)
Clip skip: 1-2 (The model works well with clip skip set to both.)
SDXL Refiner: Not used with my models!

P͓̽e͓̽r͓̽s͓̽o͓̽n͓̽a͓̽l͓̽ ͓̽R͓̽a͓̽t͓̽i͓̽n͓̽g͓̽:͓̽
Quality: 9/10 (for version 1 to 9 - 11) | 7.5/10 (Turbo)
Stability (Variety): 6/10, the lower = the more variety with one prompt
Flexibility: 8.5/10
CFG resistance: 30/30 (for version 1 to 9 - 11) | 3-7 (Turbo)
Style strength: 7.5/10
Style Variety: 9/10 (for version 1 to 9 - 11) | 8/10 (Turbo)
LoRA: XL YAMER 的风格 ♠️ Princeps Omnia LoRA
恢复人脸:不需要生成高质量的图像(当前展示中未使用)。
剪辑跳过:1-2(该模型在将剪辑跳过设置为两者时效果很好。
SDXL 精简程序:不与我的模型一起使用!
P͓̽e͓̽r͓̽s͓̽o͓̽n͓̽a͓̽l͓̽ ͓̽R͓̽a͓̽t͓̽i͓̽n͓̽g͓̽:͓̽
质量: 9/10 (适用于版本 1 至 9 - 11) |7.5/10 (涡轮增压)
稳定性(品种):6/10,越低=一个提示的品种越多
灵活性: 8.5/10
CFG 电阻:30/30(适用于版本 1 至 9 - 11) |3-7 (涡轮增压)
风格强度:7.5/10
款式多样性:9/10(适用于版本 1 至 9 - 11)|8/10 (涡轮增压)
图像质量:1024x1024(SDXL 标准)、16:9、4:3、6:13
步长:35-150步(在30步以下,可能会出现一些伪影和/或奇怪的饱和度,例如:图像可能看起来更粗糙,色彩更少)。
雇用升频器:4x_foolhardy_Remacri 或 4xUltraSharp
租用高档:唯一的限制是你的 GPU(我放大了基础图像的 2,5 倍,576x1024)
VAE:SDXL VAE

Hash
parameter
SHA256
DDA8C0514CA0B28BB4285385CC8696871F6FA1FA31AA49CA5FD83FA17C35E87B
CRC32
22AD2163
BLAKE3
A7656BB5114035B907C48F5F50CD5D006E1384F94FBD4ED55398B57B52727BAC
AUTOV1
530D0453
AUTOV2
DDA8C0514C

作品展示:
00008-906573843.jpeg
作品参数:
  1. amazing watercolor painting  award winning, ((masterpiece)), ((best quality)), ((detailed)), Unparalleled, ((Enchanting)), ((Impressive)), a black cat sitting on top of a reflective surface, cyberpunk art, inspired by Dan Mumford, furry art, details and vivid colors, electric wallpaper, beeple and jeremiah ketner, multicolored vector art, intricate, in a painting of a cat, full res psychedelic therapy, trending on ArtStation, ink splatters, pen lines, incredible detail, creative, positive energy, happy, unique, negative space, pure imagination painted by artgerm ,Surreal Harmony, SLPASHTER_3-7000, Gorgeous splash of vibrant paint, colorful paint, Colorsplash
  2. Negative prompt: bad quality, bad anatomy, worst quality, low quality, low resolution, extra fingers, blur, blurry, ugly, wrong proportions, watermark, image artifacts, lowres, ugly,  jpeg artifacts, deformed, noisy image, deformation, skin moles
  3. Steps: 50, Size: 576x1024, Seed: 906573843, Model: Unstable V11 + RunDiffusion, Version: 1.7.0, Sampler: DPM++ 3M SDE Exponential, CFG scale: 5, Model hash: dda8c0514c, Hires steps: 45, Hires upscale: 2.5, Hires upscaler: 4x_foolhardy_Remacri, Denoising strength: 0.25
复制代码

EB48067E37E843C646286F5A537B5309366A88C554F31762709A0B1F50F23F98.jpeg
作品参数:
  1. ink illustration, ink stains, ink splatters, ink runs, ink spots, faded ink, dramatic bonsai tree, plant on black reflective surface, sunrise, detailed, amazing natural lighting, earthy pastel colors, weeping for the lost children, symmetrical fractal pattern, linquivera
  2. Negative prompt: Closed eyes, text, watermark, deformed, mutated
  3. Steps: 36, Size: 1216x832, Seed: 2044874001, Sampler: DPM++ 2M Karras, CFG scale: 5, Clip skip: 2
复制代码


作品参数:
  1. in top of city a beautiful sunset with stars and planets in the sky, concept art, inspired by Cyril Rolando, fractal thunder dan mumford, water everwhere fantasy, ultra hd anime wallpaper, very beautiful photo, purplish space in background, by Agnes Cecile, by Andrà Masson, fine art, cinematic lighting, 8k concept art, trending on artstation, deep color, ink drawing, In Futuristic Ultra Detailed, fear Background, Highly Detailed Galaxy, Scale Like Skin, Daz 3d, Unity, Unreal Engine 5, Uhd Resolution, Sharp Focus, 8k, 12MP
  2. Negative prompt: bad quality, bad anatomy, worst quality, low quality, low resolution, extra fingers, blur, blurry, ugly, wrong proportions, watermark, image artifacts, lowres, ugly,  jpeg artifacts, deformed, noisy image, deformation, skin moles
  3. Steps: 50, Size: 768x1368, Seed: 2687909475, Model: Unstable V11 + RunDiffusion, Version: 1.7.0, Sampler: DPM++ 3M SDE Exponential, CFG scale: 5.5, Model hash: dda8c0514c, Hires steps: 45, Hires upscale: 2.5, Hires upscaler: 4x_foolhardy_Remacri, Denoising strength: 0.25
复制代码

1BC876BC0640E0AC73F77A65F38DDD689F8A2493D23678D267B7BBE4A15AAEEF.jpeg
作品参数:
  1. ink illustration, realistic, wide angle, female  wearing white robes (dissolving in the wind:1.5), fierce, movie pose, abandoned city street, wild wind, heavy driving rain, storms, obscured view, channeling a raven,seducing the viewer, ink stains, ink splatters, ink runs, ink spots, faded ink, linquivera, mentixis
  2. Negative prompt: Closed eyes, text, watermark, deformed, mutated
  3. Steps: 34, Size: 832x1216, Seed: 1980631941, Sampler: DPM++ 2M Karras, CFG scale: 4, Clip skip: 2
复制代码

CD088AE60ABEB61292BD369604D01C5EA66D51901624B086D71AB8E55F2526AC.jpeg
作品参数:
  1. ink illustration, realistic, wide angle, female eleven feather mage wearing white robes, fierce, movie pose, abandoned city street, wind, rain, storms, channeling a raven, ink stains, ink splatters, ink runs, ink spots, faded ink, linquivera
  2. Negative prompt: Closed eyes, text, watermark, deformed, mutated
  3. Steps: 34, Size: 832x1216, Seed: 331359891, Sampler: DPM++ 2M Karras, CFG scale: 4, Clip skip: 2
复制代码

下载地址:
游客,如果您要查看本帖隐藏内容请回复
等待就是结果 ~ 2024-2-7 20:11
我尝试了两个涡轮增压检查点模型和多个 SD1.5 Lora,似乎 SDXL 涡轮增压器可以与 SD1.5 Lora 一起运行,至少在 ComfyUI 中是这样。不是 100% 确定 lora 是否真的做了他们训练过的事情, 图像有点偏离并且不同, 组合 SD1.5+SDXL 不会像以前那样抛出错误.此外,似乎可以在没有 ecode/encode 的情况下将 SD1.5 潜伏扩展到涡轮增压潜伏。奇怪。

点评

嗯,有趣的是,SD 1.5 LoRA 不应该与 SDXL 型号一起使用,如果您发现任何有趣的东西,请告诉我。 如果可以的话,请上传一些图片,看看有没有 LoRA 有什么区别, 谢谢你通知我这件事!   发表于 2024-2-7 20:12
素以为绚 ~ 2024-2-7 20:15
哇,这个模型超级棒!可能是因为其中有 30% 的运行扩散模型

点评

感谢您使用我的检查点生成图像  发表于 2024-2-7 20:15
您需要登录后才可以回帖 登录 | 立即注册

本版积分规则

站长推荐

“三人行,必有我师焉。择其善者而从之,其

子曰:“述而不作,信而好古,窃比于我老彭。” 子曰:“默而识之,学而不厌,诲人不 ...

【LORA】Bejba-Clouds云的基础元素Clouds[418FECC517]

资源详情: A LoRa trained on the very good greenscreen-transition-clouds from the polish ESC perfor ...

【CHECKPOINT】AgAnimeAG基本动漫模型aganime_v11[53901460FB]

资源详情: 替代另一种动漫组合,在这种情况下,该模型可以提供更详细的结果、更柔和的线条和更多种类的伽 ...

【CHECKPOINT】Anime bulldozer_BETA贝塔动漫SDXLAnimeBulldozer_v20[44785DD22C]

资源详情: 该模型能够根据艺术家标签生成相应的风格,为了模型训练使用了艺术家的风格,对此表示歉意。这 ...

【CHECKPOINT】Animagine XL V3动漫工程师animagineXLV3_v30[1449E5B0B9]

资源详情: Animagine XL 3.0 是复杂的开源动漫文本到图像模型的最新版本,基于其前身 Animagine XL 2.0 ...

1 2 3 4 5

Archiver|小黑屋|思学Ai ( 苏ICP备2020049347号-3| 苏ICP备2020049347号-4 )

GMT+8, 2024-11-21 20:58

Powered by Discuz! X3.5

© 2001-2024 Discuz! Team.

快速回复 返回顶部 返回列表