I found a new AI tool Shakker, the best image-to-image tool. You can try it via https://www.shakker.ai, it can help you:
-Remix: Upload a picture. Switching the prompts, you can create stunning images in the same style.
-Style Transfer: Shakker not only extracts the style but also switches among various styles.
Besides, Shakker also offers Object Control, Composition Control, Collage Redrawing, etc.
Three Ways to Get it
Online Version: https://zhuyu1997.github.io/open-pose-editor/
sd webui extension: https://github.com/nonnonstop/sd-webui-3d-open-pose-editor
Local Version (If get network issue): click download
Basic Workflow:
Edit Pose => Generate Depth/Normal/Canny Maps => Drag images into ControlNet panel
Project link:
GitHub: https://github.com/ZhUyU1997/open-pose-editor
Gitee (China): https://gitee.com/891085309/open-pose-editor
Feature:
Pose Editing: Edit the pose of the 3D model by selecting a joint and rotating it with the mouse.
Hand Editing: Fine-tune the position of the hands by selecting the hand bones and adjusting them with the colored circles.
Depth/Normal/Canny Maps: Generate and visualize depth, normal, and canny maps to enhance your AI drawing.
Save/Load/Restore Scene: Save your progress and restore it later by using the built-in save and load functionality.
Adjust Body Parameters: Adjust various body parameters such as height, weight, and limb lengths to create a custom 3D model.
Redo, Undo
Usage: Youtube Video
Scene Navigation:
Rotate Scene: Click and hold the blank space, then move the mouse while holding down the left mouse button.
Move Scene: Click and hold the blank space, then move the mouse while holding down the right mouse button.
Body Manipulation:
Rotate Body: Click on any joint to select it, then hold down one of the colored circles and move the mouse to rotate the selected joint.
Hand Editing: Click on the red dot to select the hand bones, then rotate them by holding down one of the red circles and moving the mouse.
Adjust Output Resolution:
Adjust Output Resolution in Menu: Change the "Width" or "Height" in the menu to control the output resolution.
Adjust Output Resolution:
Change Browser Window Size: Adjust the size of your browser window to control the output resolution.
Other Functions:
Switch to Move Mode: Press the X key to switch to move mode, allowing you to move the entire body.
Delete Body: Press the D key to delete the entire body.
Redo: Ctrl + R / Shift + Ctrl +C
Undo: Ctrl +C
Contact:
email: [email protected]
三种方式
sd webui 插件: https://github.com/nonnonstop/sd-webui-3d-open-pose-editor
本地版本：点击下载
项目主页：
GitHub: https://github.com/ZhUyU1997/open-pose-editor
Gitee (China): https://gitee.com/891085309/open-pose-editor
国内网站： https://www.liblibai.com/#/model/508
参考视频：
AI绘画免费3D骨骼工具，让女神摆出你想要的姿势，手指所有关节都可调整 https://www.bilibili.com/video/BV1QL411R7wE/
【在线3d版poenpoes编辑器，让AI更容易画出你想要的动作来，controlnet绝配】 https://www.bilibili.com/video/BV1ro4y1B751
基本工作流程
编辑姿势 => 生成深度/法线/Canny 贴图 => 将图片拖到 ControlNet 面板中
不要把这个编辑器和那个 webui 扩展搞混了，它们不是一个东西
功能
姿势编辑: 通过选择关节并使用鼠标旋转编辑 3D 模型的姿势。
手部编辑: 通过选择手部骨骼并使用彩色圆圈微调位置来精调手部位置。
深度 / 法线 / Canny 地图: 生成和可视化深度、法线和 Canny 地图，以提高 AI 绘图的质量。
保存/加载/还原场景: 使用内置的保存和加载功能保存进度并在以后恢复。
调整身体参数: 调整各种身体参数，如身高、体重和肢体长度，创建自定义的 3D 模型。
重做，撤销
用法
场景导航:
旋转场景: 单击空白处并按住鼠标左键拖动鼠标。
移动场景: 单击空白处并按住鼠标右键拖动鼠标。
身体操纵:
旋转身体: 单击任何关节以选择它，然后按住彩色圆圈之一并移动鼠标以旋转所选关节。
手部编辑: 单击红点以选择手部骨骼，然后按住红色圆圈之一并移动鼠标以旋转它们。
调整身体参数:
选择身体: 单击身体以选择它。
打开身体参数: 单击菜单中的“Body Parameters”以调整身体参数。
调整输出分辨率:
在菜单中调整输出分辨率：更改菜单中的“宽度”或“高度”以控制输出分辨率。
切换到移动模式: 按 X 键切换到移动模式，允许您移动整个身体。
删除身体: 按 D 键删除整个身体
重做: Ctrl + R / Shift + Ctrl +C
撤销: Ctrl +C
个人QQ:891085309
bug/feature反馈交流QQ群：272950545
Description
Detect 3D pose from image
sd webui extension: https://github.com/nonnonstop/sd-webui-3d-open-pose-editor
Poses library
Support undo, redo
检测图片3D姿势
sd webui 插件: https://github.com/nonnonstop/sd-webui-3d-open-pose-editor
姿态库
支持撤销重做
FAQ
Comments (43)
I love the "set random pose" option, but it doesn't work in the webui extension :(
can you add a way to add/delete bones? openpose can be used to generate centaurs, digitigrade characters, dinosaurs, etc, it is not limited to human anatomy.
also, for less weird users, a hand pose library would be nice to have.
very cool tool by the way.
Thanks for your feedback. This request needs more complex editing interaction. I consider implementing it if I am free
@babmutz629 I will add it in next version
That's really great but more feet options please. Control feet fingers etc.
上手有难度，比如我想移动某个关节，而不是旋转。期待更成熟的版本。
刚刚支持IK, 可以移动四肢，更多的控制，后面会添加
I didn't understand how to get the best out of joining 4 ControlNET, does anyone have a tutorial?
Why do you give a low rating because of something that has nothing to do with what you're rating?
You just need to learn to use ControlNET properly. Try removing Canny and reducing "Guidance End" to 0.7 (for each).
@TernaryM01 I have figured out how to use it now. Because of your comment I edited my review to 1 star.
可以讓手部的深度圖，全白嗎(沒有深度或降低深度)，這樣到depth lib會比較好用，不然只能用canny。
为什么还需要depth lib，生成的深度图效果不好？
is it possible to select more than one point ? i'd like to rotate the entire skeleton, but not sure how
you can click center of body
@yzhu the small red ball under the neck ? it seems to only rotate the neck or another limb, not the entire body at once
@yzhu ive been trying all day to do this, are you sure i can rotate the entire body or select more than one point ? the point in the middle under the neck seems to move the neck or all other limbs but not all of them together, otherwise there doesn't seem to be a centre of body
I understand you problem, it is a bug. caused by recent change. thanks for your patient test
@Aili I fixed it by adding a center gray point. Hope it is useful for you
@yzhu thank you so much, works perfectly now
Hi yzhu, pose editor looks great, can i embed it in a small site i created to upload controlnet poses https://controlnetposes.com/controlnet-poses
Of course. If you meet issues. please feedback on below link
@yzhu thanks alot!
Without Inverse Kinematics this is frustrating as hell to use.
Thanks for your feedback. I already added four control points of IK, it is a good start
@yzhu In that case the next frustrating thing is a lack of documentation. There's plenty of room to put some quick how-to details actually inside the app, nice and visible to new users.
@blister0fingers790 Ok, but it will be done after new ui release, current status can't support complex interaction. BTW, it is deployed on github, I can't pay more for it, too many media element will influence visit speed.
@yzhu I don't know what you're imagining but I was referring to a text box with basic controls and information... I doubt that will slow it down.
hello, since last webui update, i cannot re select any skeleton nodes in the webui app, i can load it a first time, but when i try to modify any part, it "freezes" and cannot select nothing, is it a known issue?
Ohhh, the error happens when while you are working in the 3dopenpose tab, you select the created image in the txt2img tab, then you select another tab different than the 3dopenpose tab, but when you go to the 3dopenpose tab, the skeleton freezes. It unfreezes when in the txt2img tab, select the image with the skeleton instead of the generated image. I don't remember it happened that way?
Can you explain more on the below link? Thank you in advance.
https://github.com/nonnonstop/sd-webui-3d-open-pose-editor/issues/new
@yzhu hello, it was due to the latest webui update, i re rolled back to the previous ver and now it's working right
I get a canny, normal, depth and pose image file. Is there a specific order I should use when I put in the Automatic1111 UI?
It's a good thing for beginners.
In my opinion, the most valuable part of this extension is the "Detect Pose From Image" since you have hand detection, but I've yet to have this ever work. I always get an error "Oops, something went wrong..."
I don't think anyone would be crazy enough to rotate one skeleton after another to control the hands pose.
If hand posture templates, or control sliders, or IK controllers cannot be provided, then the hand skeleton feature is not an improvement, but a burden.
controlnet v1.1 now supports hand and face I will be really happy to see it here
I had zero luck trying to create a turn around, each time the back view renders as front in web ui
UPD. Never mind, just added "turned back to camera" and it works :)
seem like send to controlnet text2img function not working properly, can't send anything to controlnet model number 1. Have to manual download the map in order to use
Thank you for the great editor. It would be immensely more helpful if you make a section/library for preset poses, especially for hand poses. Thank you again for your work.
First of all, thanks a lot for that great editor! As some other mentioned, it would be helpful to have presets and it would be also great to have two cameras: one for the final export and one for editing.
this is just way too non user friendly. the biggest downside is how bad the image detection is.
感谢 插件版太烂了