手把手教你用stable diffusion绘画ai插件mov2mov生成动画_哔哩哔哩_bilibili手把手教你用stable diffusion绘画ai插件mov2mov生成动画, 视频播放量 14552、弹幕量 3、点赞数 275、投硬币枚数 114、收藏人数 980、转发人数 75, 视频作者 懂你的冷兮, 作者简介 科技改变世界,相关视频:Stable Diffusion 使用mov2mov插件生成动漫视频,【AI动画】使AI动画纵享丝滑~保姆级教程+Stable Diffusion+Mov2mov扩展,轻轻松松一键出AI视频,最强 AI 绘画工具——Stable Diffusion超详细教程,【SD小白变大神系列12】纵享丝滑 手把手教你用mov2mov插件让ai绘画动起来 Stablediffusion webUI绘图软件速成教程,保姆级教程:用stable diffusion制作动画短视频,超简单!,【AI绘画进阶教程】制作跳舞小姐姐 m2m和mov2mov哪家强?Stable DiffusionAI绘图软件保姆级教学,【AIGC】stablediffusion Mov2Mov插件,SD视频生成ai美女跳舞视频教学,【stable diffusion】与AI一起cosplay,玩出摄影级作品!,《txt2video》最新!stable diffusion插件txt2video,文生视频工具,使用教程!,StableDiffusion controlnet插件,让AI画出你想要的人物动作https://www.bilibili.com/video/BV1Yo4y1B7kL/?spm_id_from=333.337.search-card.all.click&vd_source=4aed82e35f26bb600bc5b46e65e25c22Stable diffusion AI视频制作,Controlnet + mov2mov 准确控制动作,画面丝滑,让AI老婆动起来,效果真不错|视频教程|AI跳_哔哩哔哩_bilibili# 本期视频链接:1. Mov2Mov插件链接:https://github.com/Scholar01/sd-webui-mov2mov# 相关内容链接:1. 使用Colab 搭建Stable Diffusion的链接:https://github.com/camenduru/stable-diffusion-webui-colab2. 使用Colab 搭建Stable Diffusion的教程, 视频播放量 71283、弹幕量 29、点赞数 1896、投硬币枚数 856、收藏人数 6378、转发人数 495, 视频作者 闹闹不闹nowsmon, 作者简介 实用型AI教学博主;油管名:闹闹不闹,相关视频:AI绘画视频 stable diffusion 跳舞 网红 美女,【AI动画】使AI动画纵享丝滑~保姆级教程+Stable Diffusion+Mov2mov扩展,轻轻松松一键出AI视频,AI图片生成视频:用AI生成的图片能说话了,还能做表情,手把手教你用stable diffusion创建一个AI老婆,自定义任何你想要的哦,AI生成视频,高清原创!!!躺赚??,Stable Diffusion 使用Controlnet制作动画,近乎完美!,【Stable Diffusion】只用一张照片就可以生成同一个人的不同表情和动作 | ControlNet1.1 reference_only,Blender + AI 让AI动起来/如何将stable diffusion生成的2D图像制作成3D动画,像这样的AI动画视频,学会制作就可以月入2W以上。最近在各大平台的热度都很高,你想知道如何制,照片生成AI,再让它说话StableDiffusionhttps://www.bilibili.com/video/BV1G14y1f7tv/?spm_id_from=333.337.search-card.all.click&vd_source=4aed82e35f26bb600bc5b46e65e25c22
Stable Diffusion进阶教程!超详细的 ControlNet 实用入门指南 - 优设网 - 学设计上优设ControlNet 的作用是通过添加额外控制条件,来引导 Stable Diffusion 按照创作者的创作思路生成图像,从而提升 AI 图像生成的可控性和精度。在使用 ControlNet 前,需要确保已经正确安装 Stable Diffusion 和 ControlNet 插件。如还未安装,可以参考这篇文章...https://www.uisdc.com/controlnet
AI动画,超详细,纯干货,稳如老狗的AI动画制作全过程理论+实战(看完还不会我跳happy猫)_哔哩哔哩_bilibiliLora分层控制bodyh:1,1,1,1,1,1,1,1,0,0,0,1,1,1,1,1,1bodys:1,1,1,1,1,1,0.2,1,0.2,0,0,0.8,1,1,1,1,1faceh:1,0,0,0,0,0,0,0,1,1,1,0,0,0,0,0,0faces:1,0,0,0,0,0,0,0,0.8,1,1,0.2,0,0,0,0,0clothe:1,1,1,1,1,0,0.2,0,0, 视频播放量 18995、弹幕量 7、点赞数 574、投硬币枚数 350、收藏人数 1868、转发人数 92, 视频作者 Octppus, 作者简介 各位观众大家好呀,欢迎来到我的频道,猜猜看,今天要介绍的是什么呢?,相关视频:AI动画制作教学只看我一个就够了,全网最详细AI动画风格化教学(基础篇) | AI划时代 | AI工作流 | AI创作免费攻略下载 | 1v1免费咨询,像这样的AI动画视频,学会制作就可以月入2W以上。最近在各大平台的热度都很高,你想知道是如何制作的吗?,如何制作趣味动画视频?打开来画动画,万千角色+丰富场景+AI智能配音,各种元素灵活搭配,让动画视频更有趣!Web端、APP端随心选择,随时随地开启高效创作!,时代变了,AI将永久改变动画制作流程,【AI动画】Tiktok这两天爆火的视频原来是这样做的,【AI视频】AI视频要变天了,Stability发布全新text to video技术,像这样的AI动画视频,学会用AI制作,轻松月人五位数,0门槛 亲测有效!!!,Cascadeur-基于AI人工智能辅助的骨骼动画软件(中文讲解教程),适合新手!全程AI人工智能,2分钟学会创作动画视频 | 动画制作 + 配音,有手就能做动画的AI利器Cascadeurhttps://www.bilibili.com/video/BV1Xg4y1j7Wg/?vd_source=4aed82e35f26bb600bc5b46e65e25c22你真的会用LORA吗?超详细讲解LORA分层控制 - 知乎LORA分层控制的教程其实网上已经有一些了,不过我感觉都偏理论,许多像在上数学课, 而且实际运用讨论很少,既然如此,我就自己研究,出一期应用型的心得好了。 既然说是应用型的心得,所以文章也会用大量的实例来…https://zhuanlan.zhihu.com/p/621260240prompt很多都是重复的,这里的第二组prompt,基本就够用了,配合比较不错的底模和embeddings效果还是不错的。
aliga,black hair,short hair,<lora:lora-hanfugirl-v1-5:0.2>,<lora:arknightsTexasThe_v10:1>,1girl,ulzzang-6500-v0.7,(raw photo:1.2),((photorealistic:1.4))best quality,masterpiece,illustration,an extremely delicate and beatutiful,extremely detailed,CG,unity,8k wallpaper,Amazing,finely detail,masterpiece,official,art,extremely detailed CG unity 8k wallpaper,cute,{{student}},{{{school uniform}}},ama usa an uniform,<lora:aliga_v1:0.8>
sketches,(worst quality:2),(low quality:2),(normal quality:2),lowers,normal quality,((monochrome)),((grayscale)),skin spots,acnes,skin blemishes,bad anatomy,(long hair:1.4),DeepNegative,(fat:1.2),facing away,looking away,titled head,lowers,bad anatomy,bad hands,text,error,missing finggers,extra digit,fewer digits,cropped,worstquality,low quality,normal quality,jpegartifacts,signature,watermark,username,blurry,bad feet,cropped,poorly drawn hands,poorly drawn face,mutation,deformed,worst quality,low quality,normal quality,jpeg artifacts,signature,watermark,extra fingers,fewer digits,extra limbs,extra arms,extra legs,malformed limbs,fused fingers,too many fingers,long neck,cross-eyed,mutated hands,polar lowers,bad body,bad proportions,gross proportions,text,error,missing fingers,missing arms,missing digit,extra arms,extra legs,extra foot
图生图+mov2mov+controlnet
model:chilloutmix_NiPrunedFp16Fix.safetensors
embeddings:ulzzang-6500-v1.1.pt.bin,EasyNegative.pt,ng_deepnegative_v1_75t.pt,bad-picture-chill-75v.pt
best quality,ultra high res,(photorealistic:1.4),(detailed beautiful girl:1.4),(medium breasts:0.8),Detailed facial details,beautiful detailed eyes,a girl,solo,teenage,slender,smile,(makeup:0.4),red lips,White skirts,backbeat,(Night:1.2),highly detailed clothes,(ulzzang-6500-v1.1:0.3)
EasyNegative,paintings,sketches,ugly,3d,(worst quality:2),(low quality:2),(normal quality:2),lowers,normal quality,((monochrome)),((grayscale)),skin spots,acnes,skin blemishes,age spot,manboobs,backlight,(ugly:1.3),(duplicate:1.3),(morbid:1.2),(mutilated:1.2),(tranny:1.3),mutated hands,(poorly drawn hands:1.3),blurry,(bad anatomy:1.2),(bad proportions:1.3),extra limbs,(disfigured:1.3),(more than 2 nipples:1.3),(more than 1 navel:1.3),(missing arms:1.3),(extra legs:1.3),(fused fingers:1.6),(too many fingers:1.6),(unclear eyes:1.3),bad hands,missing fingers,extra digit,(futa:1.1),bad body,double navel,mutad arms,hused arms,(puffy nipples,dark areolae,dark nipples,rei no himo,inverted nipples,long nipples),NG_DeepNegative_V1_75t,pubic hair,fat rolls,obese,bad-picture-chill-75v
ulzzang-6500-v1.1是一个embadding,
是所谓的KoreanDollLikeness这个lora的前身,是很古老很古老的一个插件,比chilloutmix初版还要早。
至于你要说这个插件是什么内容,简短中国字就是——“脸”
“ulzzang”是南朝鲜地区对网红脸女生的一种称呼,所以用这个插件会长出一股网红味的棒子脸来,也是chilloutmix等模型范图经常是网红棒子脸的原因之一。
6500指训练这个embadding跑了6500步,但据说训练数据集多达200多张。
v1.1是指tag中的版本是1.1版本,是继续训练后的补丁版本,最早发布的ulzzang是没有版本号的,只有6500。
cfg_scale:10,denoising sreength:0.3,control weight:0.25
model:anything-v3-fp16-pruned.safetensors
embeddings:ulzzang-6500-v1.1.pt.bin,EasyNegative.pt,ng_deepnegative_v1_75t.pt,bad-picture-chill-75v.pt
img2img选图时:添加了controlnet,openpose_full,control weight:1,starting control step:0,ending control step:1
model:
(ultra high res,photorealistic,realistic,best quality,photo-realistic),(high detailed skin,visible pores),(real person,photograph),(8k,raw photo,best quality,masterpiece),purple,blue,Solid color,((extremely detailed CG unity 8k wallpaper)),((world masterpiece theatre)),(best illumination),(best shadow),dynamic angle,depth of field,((masterpiece)),((best quality)),highres,((extremely detailed CG wallpaper)),((cinematic lighting)),((extremely detailed illustration)),finely detail,depth of field,original,extremely detailed CG,extremely detailed wallpaper,(solo),highres,((extremely detailed illustration)),(dynamic angle),best illumination,((1girl)),(loli:1.4),(small breast),beautiful detailed eyes,beatuiful detailed girl,extremely delicate and beautiful girls,(red eyes:1.2),((upper body)),looking at viewer,(red hair:1.4),(medium hair:1.4),(red hooded sweatshirt:1.4),(earphone:1.4),(backpack:1.3),detailed background,(streetcar:1.2),Tram stop,(blue sky:1.25),(sunlight),streets,cars,(crowd),
nsfw,((Multiple people)),lowers,bad anatomy,bad hands,text,error,missing fingers,extra digit,fewer digits,cropped,worstquality,signature,watermark,username,blurry,bad feet,cropped,((((poorly drawn hands)))),poorly drawn face,mutation,deformed,worst quality,low quality,normal quality,jpeg artifacts,signature,watermark,extra finger,fewer digits,extra limbs,extra arms,(((extra legs))),malformed limbs,((((fused fingers)))),((((too many fingers)))),long neck,cross-eyed,mutated hands,polar lowers,bad body,bad proportions,gross proportions,text,error,((((missing fingers)))),missing arms,missing legs,((((pubic)))),((((pubic_hair)))),(((large_breasts))),(((hugh_breasts))),(((gigantic_breasts))),sketch,duplicate,ugly,huge eyes,(text:1.5),(logo:1.5),monochrome,worst face,(bad and mutated hands:1.3),(worst quality:2.0),(low quality:2.0),(blurry:2.0),horror,geometry,bad_prompt,(bad hands),(missing fingers),multiple limbs,bad anatomy,(interlocked fingers:1.2),ugly fingers,(extra digit and hands and fingers and legs and arms:1.4),crown braid,((2girl)),(deformed fingers:1.2),(long fingers:1.2),succubus wings,horn,succebus horn,succubus hairstyle,(bad-artist-anime),bad-artist,bad_hand
使用multi_frame_rendering来渲染
model:tmndMix_tmndMixIIIPruned.safetensors
Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 657235123, Face restoration: CodeFormer, Size: 720x1280, Denoising strength: 0.45,
img2img选图时:添加了controlnet,canny,control weight:1,starting control step:0,ending control step:1
multi_frame_rendering:
initial denosing strength:第一帧的去噪强度,其余帧的降噪强度通过img2img控制
third column(reference) image:None:只使用两张图片,即前一帧和当前帧,不使用第三张参考图片
FirstGen:使用第一帧作为参考图片
Historical:使用当前帧前倒数第2帧作为参考图片
推荐使用FirstGen
enable color correction:使用非FirstGen作为参考图时开启,以减少褪色现象
unfreeze seed:勾选后每次生成图片会自动在基础种子上加1,如果希望你的图片每帧都不一样,可以选
Loopback Source:也就是中间图片,有三种模式
previousFrame:从前一帧生成当前帧
current:从当前帧生成当前帧
first:从第一帧生成当前帧
推荐使用current
lora block weight
两个lora主要是针对脸部改变的
bodyh,bodys,faceh,faces,clothe,act,color,char,background,lowover
bodyh:1,1,1,1,1,1,1,1,0,0,0,1,1,1,1,1,1
bodys:1,1,1,1,1,1,0.2,1,0.2,0,0,0.8,1,1,1,1,1
faceh:1,0,0,0,0,0,0,0,1,1,1,0,0,0,0,0,0
faces:1,0,0,0,0,0,0,0,0.8,1,1,0.2,0,0,0,0,0
clothe:1,1,1,1,1,0,0.2,0,0.8,1,1,0.2,0,0,0,0,0
act:1,0,0,0,0,0,0.2,1,1,1,0,0,0,0,0,0,0
color:1,0,0,0,0,0,0,0,0,0,0,0.8,1,1,1,1,1
char:1,1,1,1,1,0,0,0,1,1,1,1,1,1,1,0,0
background:1,1,1,1,1,1,0.2,1,0.2,0,0,0.8,1,1,1,0,0
lowover:1,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1
(black hair),grey shirt,blue denim shorts,white shoes,(black eyes),abdominal muscle<lora:animeLineartStyle_v20Offset:1:1,0,0,0,0,0,0.2,1,1,1,0,0,0,0,0,0,0><lora:tifaMeenow_tifaV2:1:1,0,0,0,0,0,0,0,1,1,1,0,0,0,0,0,0>
(glove:1.2),ear ring,gradient hair,colored tips,blue hair,hair accessories,stockings,anklet,leg ring,illustration,3d,painting,cartoons,sketch,(worst quality:1.5),(low quality:2),(normal quality:2),lowers
1.视频质量太差
先超分
2.既要任务一致性,又要背景根据自己需求进行变动
人物抠图加背景合成
通过prompt更换背景较差(定向模型训练多是较过拟合的)
3.做舞蹈视频类转换,人物一定要占据视频的主体,人物不能太小,太小生成出来的都不行
dpm++ sde karras,sampling_steps:27,cfg_scale:7.denoising strength:0.45
lora block weight
controlnet canny 0.8,openpose only face 0.5
face editor
1girl,solo,long_hair,shoes,black_hair,socks,dress,bow,apron,(ultra high res,photorealistic,realistic,best quality,photo-realistic),(8k,raw photo,best quality,masterpiece),technological sense,best quality,masterpiece,illustration,CG,unity,wallpaper,official art,amazing,finely detail,an extremely
controlnet+mov2mov,txt2img和mov2mov下的controlnet都可以,但是img2img下的controlnet失效了。在使用img2img时,先把mov2mov移出来即可。