基本上是我之前融合模型的融合 basically a mix of my previous mix
majicMIX realistic - v4 | Stable Diffusion Checkpoint | Civitai
majicMIX fantasy - v2.0 | Stable Diffusion Checkpoint | Civitai
这个模型风格炸裂,远距离脸部需要inpaint以达成最好效果,使用adetailer.
This model is strongly stylized in creativity, but long-range facial detail require inpainting to achieve the best results. Use Adetailer.
请参考我的prompt来使用
recommended positive prompts: official art, unity 8k wallpaper, ultra detailed, beautiful and aesthetic, beautiful, masterpiece, best quality, (zentangle, mandala, tangle, entangle)
use ng_deepnegative_v1_75t and badhandv4 in negative
脸部修复方法 to inpaint the face:inpaint-->only masked-->set to 512x512-->Denoising strength:0.2~0.5
我的TG频道可以看到更多例图 https://t.me/majic_NSFW
Description
FAQ
Comments (42)
你这些词都是怎么想出来的?把脑壳寄过来借我用一下
生成出来的东西都不一样,和演示图片有好大区别哦
是的,我和你情况一样,生成出来都不一样的。
Only characters?
You can try with landscape or other object
yes, I'm already trying
爱了爱了
太漂亮了
炸裂
泰裤辣!
这个模型融合了,之前你做的别二个:realistic和 fantasy?那二个就可以不用,还是有所区别呢,请教下,你做模型还是很不错的。
是的,这个模型是比fantasy更写实一点,比real更魔幻一点,各有千秋吧
make loras of your models too pls
hmmm I am not sure the lora will be distinguished,but I will upload it later
great, thank you! :) of course a lora wont have the same possibilities as the full model but it offers flexibility for combining stuff without the need to swap models and inpaint back and forth depending on the project, so imo adding a lora of a model always makes sense if the main model has unique style or any other unique elements or concepts and isnt just general purpose, and yours all have those.
绝了,类型如CHILLOUT和REV的融合,加油!
could someone make a diffusers model from this? :)
请教为什么把prompt拷贝过来以后生成的图和原图一点都不像,为什么呢?
有的图二次inpaint过 或者图生图放大过的,prompt有可能有改动,不一样,还有的图用了不一样的插件,细节可能不一样,至少我自己的图是这样,不过大佬作者自己传的例图应该是能复现的
有很多影响因素,有的人为了保证元素准确第一次出图选用了较小的分辨率同时勾选了高分辨率修复,由于炼丹师大多是用较为小画幅的图片来练制模型所以缩小画幅可以有效提高词条识别的准确性。另外种子也很重要,AI绘画具有相当随机性种子就是一些随机性的体现相当于抽卡抽到的编号,所以善用种子可以降低随机性
@2625787627151 请问种子是随机生成的然后随到好的就保存一下号码下次继续用吗
大佬,用了你的真实的那个模型,现在看你还有幻想的模型,但是有好几个幻想的模型的样子。幻想的推荐哪个呢?
(zentangle, mandala, tangle, entangle)这串提示词是什么意思啊
请问一下,如何才能使高清修复后的图与直出图保持一致?我用了Lora,但高清修复后,即使重绘幅度设为0.1也总是会变回大模型的脸,我应该怎么操作?
conditioning-highres-fix,搜这个插件,高清修复原图调节程度,
@gushi2276206 谢谢!
@eisdxlagmailcom308 。。昨天没仔细看,如果说脸变了的话不知道你用没用作者大佬推荐的插件Adetailer,专门重绘你想要的lora的脸,一起用好一些吧
Details
Files
majicmixLux_v1.safetensors
Mirrors
majicmixLux_v10.safetensors
majicmixLux_v10.safetensors
majicmixLux_v10.safetensors
majicmixLux_v1.safetensors
majicmixLux_v10.safetensors
majicmixLux_v10.safetensors
majicmixLux_v10.safetensors
majicmixLux_v10.safetensors
majicmixLux_v10.safetensors
majicmixLux_v10.safetensors
majicmixLux_v1.safetensors
majicMIX_HuaLi:lux_v1.0.safetensors
majicmixLux_v10.safetensors
majicmixLux_v10.safetensors
majicmixLux_v10.safetensors
majicmixLux_v10.safetensors
majicMIXaestheticv1-fp16-no-ema.safetensors
majicmixLux_v10.safetensors
majicmixLux_v10.safetensors
majicMIX lux_v1.0.safetensors
majicmixLux_v1.safetensors





