Suggestion/建议:
Step 1/第1步: txt2img/文生图:
Negative prompt: (worst quality:2), (low quality:2), (normal quality:2)
Sampling steps/迭代步数: 40-50
Hires. fix/高分辨率修复: On/开
Upscaler/放大算法: R-ESRGAN 4x+ or 4x-UltraSharp
Hires steps/高分迭代步数: 20
Denoising strength/重绘幅度: 0.2-0.4
Step 2/第2步: Send to img2img/>> 图生图:
Resize mode/缩放模式: Just resize/仅调整大小
Denoising strength/重绘幅度: 0.2-0.25
Script/脚本: SD upscale/使用SD放大(Upscaler/放大算法:R-ESRGAN 4x+ or 4x-UltraSharp)
请勿商用
Description
Different background styles/不同的背景风格
Better 2D performance/更好的2D表现
FAQ
Comments (51)
Fantastic work! But how did you do so colored picture? Usually, my work gets very faded colors, even if I write a hint (Bright colors)
Are you using a VAE?
@kusoge I use an automatic VAE, I have not found any clarification on this in the model or posted images. Maybe you can tell me something.
@veseleykaver459 automatic VAE usually is equal to no VAE. Use novelai-vae or lf-f8-anime. You will not find anything about VAE in image description on this site because it's not inside image metadata (https://i.imgur.com/XGrqrHf.png) and most people are using the same VAE.
Use this vae, https://civitai.com/models/22354/clearvae
hello firework image so bad
你好
烟花的形象太差了,会变成红色的树
请问能解决么?
hello
firework image so bad
试试将Denoising strength调到0.3以下,Upscaler不要选择Lantent,这样应该不至于变成树
你的初始出图分辨率很高,你有多大显存?
24G,如果显存不大其实可以多次使用SD upscale来画大图
@gxf15935723312G显存也没出多少细节,除了角色本身,背景特别空。
作者你好,新版是整合了VAE吗?
并没有,只是推荐使用这个VAE来达到更好的2D表现
我下载了clear vae,放到VAE目录中,但是每次 都 报错---NansException: A tensor with all NaNs was produced in VAE. 虽然 我按提示添加了set COMMANDLINE_ARGS= --no-half-vae参数,但并没有用
@lhhsh 试试再加个--disable-nan-check
为什么我使用此模型的时候生成的人物默认都会拿着一个手提包啊?
多试几次吧,或者看看prompt是否和手提包相关
我也是,就算删掉种子号,或者完全原创prompt还是会有大概率拎着各种包,很奇怪,在negative里面怎么加权重,包都去不掉的那种
大佬有合作意向吗
why are my faces twisting
Perhaps because the image size is too small,you can use script(SD upscale) to improve face
作者您好,您没有标注应该使用什么VAE效果比较好,请问有推荐吗?谢谢!!
@gxf159357233 谢谢!
作者你好,我想请问一下,新版文生图的色彩饱和度一直非常低,尝试了许多画面词条,或者搭配lora都没什么效果,也使用了你推荐的VAE,请问您有什么解决办法吗
如果按照我的建议操作和VAE的效果来作图的话应该效果差别不大,有没有可能是迭代步数太低了呢
是不是VAE没有配置?我之前饱和度低就是发现设置里面没配VAE
想请教下大佬是如何做到如此清晰的面部细节和场景的,我也是先小图确定构图再sd放大,但是始终感觉面部细节不够清晰,部分背景模糊
我把图放大到4K左右才把全身横图的脸部细节做好,比较占比太小了,想要小尺寸的图的情况下好效果应该得用一些专门的修复脸部的插件(不是自带的面部修复)
@gxf159357233 万分感谢解答
This model seems cannot draw openning windows. I tried tens of prompt, all windows is cloesd🤣
如果是从别的模型出的图转过来做 img2img 的话,大概怎么来比较好?其它模型有的手太炸了,重回幅度小了/图太高清了,手就修不回来。但是重绘幅度高的话感觉很容易出现噪点
画手的话用controlnet插件不是更好些吗
Does anyone know on which base model (eg. SD 1.5) is this model based on? Since I have the need to incorporate this into a `diffusers` pipeline, it is important for me to know.
这些剪修后体积小这么多的模型,对比没修剪的大体积模型会有什么不同和影响?
出图效果没区别
大体积模型的优势是全面。
我也想问这个问题,既然出图效果没影响,那大体积模型在绘制内容上会更加全面么?
一盘菜,好吃不好吃的都给你了就是full版本,开发者挑选一些他认为好吃的另装一盘就是pruned版本,功能没差,但是出图细节肯定变化会少一些
請問您有Discord或其他群組嗎? 想跟隨您或支付以支持您~
我是个人,没有群组
SD 1.5?
which model did you train your model on?
or which models were merged?
需要什么VAE
这个背景构图非常强大,光影,视觉引导都很强,但是人物主体和背景的协调性好差,感觉像是在两个图层里的东西。
Details
Files
tmndMix_tmndMixIII.safetensors
Mirrors
tmndMix_tmndMixIII.safetensors
tmndMix_tmndMixIII.safetensors
TMND_MixIII.safetensors
tmndMix_tmndMixIII.safetensors
tmndMix_tmndMixIII.safetensors
tmndMix_tmndMixIII.safetensors
tmndMix_tmndMixIII.safetensors
BiBei:TMND-Mix_v1.0.safetensors
必备模型|TMND-Mix_v1.0.safetensors
tmndMix_tmndMixIII.safetensors
tmndMix_tmndMixIII.safetensors
tmndMix_tmndMixIII.safetensors
tmndMix.safetensors
tmndMix_tmndMixIII.safetensors
漫画混合模型tmndMix_tmndMixIII.safetensors
tmIII.safetensors
tmmix.safetensors
tmndMixIII.safetensors
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.





