# ComfyUI

StabilityAI推出的文本到图像生成AI

1 人关注 | 14 篇内容

  • 字节跳动推出Hyper-SD是目前最优秀的加速模型。字节跳动在前期推出 Lightning 模型的基础之上,推出了 Hyper-SD技术作为加速Lora,支持SD 1.5 和SDXL,并明确表示这是目前最先进的图形加速技术之一,并全方位超越自己此前推出的Lightning。在饱和度、对比细节渲染完成度等多方面都要完败其他LCM、Turbo等其他加速技术。 Hyper-SD Lora及Comfyui工作流下载网盘: 链接:网页链接提取码:tYNV Hyper-SD官网Github页面:网页链接 Hyper-SD huggingface页面:网页链接 Hyper-SD演示页面:网页链接 网页链接
  • 在使用 Stable Diffusion Comfy UI 的时候,如果遇到喜欢但不清楚其来源模型的图片,以往往往需要在众多模型网站上逐一查找,就像大海捞针一样困难。现在,有一种免费工具叫作 FindSD,可以解决这个问题: 1.用户只需上传想要查询的图片至 FindSD。 2.FindSD 能够快速搜索并识别出用于生成该图片的对应模型,或者给出相似模型的信息。 3.在检索结果中,每张匹配图片下面都会有一个链接按钮,点击后可以直接跳转至相关模型在C站的下载界面。 4.除此之外,FindSD 还会显示生成该图像所使用的 Stable Diffusion 模型及其具体参数。 因此,通过利用 FindSD,大家在寻找所需模型以及相关信息时,将会更加高效便捷,不再受此问题困扰。
    99次观看
  • #ComfyUI基本的SDXL + Refiner精炼工作流程# Refiner精炼后图片质量提升很大
  • #ComfyUI基本的SDXL工作流程# 一只带着一杯咖啡的人形猫{模型:"juggernautXL_v8Rundiffusion_V8.safetensors",正面提示词:"a humanoide cat with a cup of cofee",负面提示词:"deformed, 3d, worst, low quality, blur,",种子:5,采样器:"dpmpp_2m_sde",调度器:"exponential",步数:30,"cfg":6.5,宽度:768,高度:1024}
  • #ComfyUI基本的SDXL工作流程# 罗马士兵 模型juggernautXL_v8Rundiffusion_V8.safetensors 提示词a Roman soldier in full armor, standing tall with a serious expression on his face. on the background military insignia, ((masterpiece)) , ((detailed))
  • #ComfyUI基本的SDXL工作流程# SDXL模型效果确实要比SD模型好太多了!{模型:"juggernautXL_v8Rundiffusion_V8.safetensors",正面提示词:"anime artwork gypsy 1boy and 1girl, cultural festival, night sky, looking away, . anime style, key visual, vibrant, studio anime, highly detailed",负面提示词:"",种子:2,采样器:"dpmpp_2m_sde",调度器:"exponential",步数:30,"cfg":6.5,宽度:1280,高度:768}
  • #ComfyUI基本的SDXL工作流程# {模型:"juggernautXL_v8Rundiffusion_V8.safetensors",正面提示词:"by Gotohui.com, A lone adventurer facing off against a giant sea monster",负面提示词:"",种子:2,采样器:"dpmpp_2m_sde",调度器:"exponential",步数:25,"cfg":6.5,宽度:1280,高度:768}
  • #ComfyUI基本工作流程SD1.5 局部重绘# {模型_1:"dreamshaper_8.safetensors",种子_1:4,采样器_1:"dpmpp_2m",调度器_1:"karras",步数_1:25,"cfg_1":6.5,种子_2:1,采样器_2:"dpmpp_2m",调度器_2:"karras",步数_2:20,"cfg_2":6.5}
  • #ComfyUI基本工作流程SD1.5 二次通过KSampler采样器# {模型_1:"dreamshaper_8.safetensors",正面提示词_1:"face close-up of her alluring eyes, captures alluring emotions, tantalizing thoughts, playful inhibition. A young appealing woman, 49 years old, reminiscent of Ashley Benson (resemblance rating: 1.3), dynamic pose. ((Her platinum-white hair cascades wildly, framing her alluring eyes)), she is adorned with smokey makeup. Mixed heritage: Norwegian (1.0), Spanish (0.5), German (0.8). (use of Gestalt Principles, Gestalt Laws), Panavision DXL2, Full-Frame, 60fps, 40.96 x 21.6mm, 1600 ISO, classic anamorphic feel: linear barrel distortion, oval bokeh, and beautiful blue-streak lens flares, gradual linear focus fall off, gorgeous amber lens flares, flares are blue and purple streaks with orange accents, creamy low contrast, shallow depth of field and stunning bokeh, atmospheric cinematography, ethereal ambiance, dramatic lighting, shallow depth of field, ultra-realistic, beautiful lighting, P3 Theatrical color space, gamma 2.6, 1M:1 Contrast Ratio, 600 nits HDR, 16 stops of dynamic range, 281 trillion hues, 35 million pixels on 8k RAW}, artwork in the style of Michael Turner (Top Cow), Laura Braga, Randy Green, J. Michael Straczynski, Marc Silvestri, Jim Lee, Aron Lusen and Christian Lichtner",负面提示词_1:"blurry, illustration, toy, clay, low quality, flag, nasa, mission patch",种子_1:8,采样器_1:"dpmpp_2m",调度器_1:"karras",步数_1:25,"cfg_1":6.5,宽度_1:512,高度_1:512,种子_2:1001,采样器_2:"dpmpp_2m",调度器_2:"karras",步数_2:20,"cfg_2":6}
  • #ComfyUI基本工作流程SD1.5图像批量生成# {模型:"epicphotogasm_lastUnicorn.safetensors",正面提示词:"superman",负面提示词:"",种子:8,采样器:"dpmpp_2m",调度器:"karras",步数:25,"cfg":6.5,宽度:512,高度:512}
  • #ComfyUI基本工作流程SD1.5图生图# {模型:"dreamshaper_8.safetensors",正面提示词:"artificial intelligence",负面提示词:"",种子:8,采样器:"dpmpp_2m_sde_gpu",调度器:"karras",步数:25,"cfg":6.5}
  • #ComfyUI基本工作流程SD1.5模型 + LoRA# {模型:"epicphotogasm_lastUnicorn.safetensors",正面提示词:"Gentleman in a suit, Full body portrait, Office setting, Confident posture, Professional attire, Realism, Close-up shot,Tables, windows,computer",负面提示词:"blurry, illustration, toy, clay, low quality, flag, nasa, mission patch",种子:8,采样器:"dpmpp_2m",调度器:"karras",步数:30,"cfg":7,宽度:800,高度:1200}
  • #ComfyUI最基本的SD1.5模型 + VAE工作流程# {模型:"epicphotogasm_lastUnicorn.safetensors",正面提示词:"building, car, city, cloud, cloudy sky, day, grass, house, lamppost, mountain, no humans, outdoors, river, road, ruins, scenery, sky, street, tree",负面提示词:"blurry, illustration, toy, clay, low quality, flag, nasa, mission patch",种子:8,采样器:"dpmpp_2m",调度器:"karras",步数:30,"cfg":7,宽度:1200,高度:904}
  • #ComfyUI 最基本的SD1.5模型工作流程# {模型:"dreamshaper_8.safetensors",正面提示词:"(step-by-step comic panel, 1 frames:1)\nClose-up face of a weathered beautiful female cowgirl face, etched with sun and grit. Eyes squint against the harsh sunlight, reflecting a glint of determination.\nstandard rectangles",负面提示词:"bad anatomy, bad hands, text",种子:795818837290732,采样器:"euler",调度器:"normal",步数:20,"cfg":8,宽度:600,高度:904}
人家是有底线的~

用户排行榜