An addon uses SMPL's poses and global translation to drive cartoon character in Blender.

Overview

Blender addon for driving character

The addon drives the cartoon character by passing SMPL's poses and global translation into model's armature in Blender. Poses and global translation can be obtained from ROMP or any other 3D pose estimation model. If the model outputs poses and global translation at a high FPS, you can drive cartoon characters in Blender in real time.

Demo

image image

The first demo uses ROMP outputs from the video, which is stored in a file.

The second demo uses ROMP outputs from the webcam in real time.

How to Use the add-on

Data Requester

This addon is a data requester that sends a data request over TCP to 127.0.0.1:9999.It gets one data at a time from the data server.

After running the addon by pressing ctrl+E in Blender, it keeps asking for data until the server closes the TCP connection. You can also press A to close the TCP connection.

Data Server

The data server is bound to 127.0.0.1:9999. After receiving a request from the data requester, one data is sent to the requester.

I've written server.py as a example data server (you only need to know a little about Python TCP to understand it).

Real-time data server can be found in ROMP.You just need to run webcam_blender.sh.

Data Format

The data is a Python list of four elements in the form of [mode,poses,global translation,current keyframe id].

  1. Mode is an integer, 1 for insert keyframe, 0 for no insert keyframe. Insert keyframes and animation rendering can be carried out later. Keyframes are generally not inserted in real-time mode to make real-time driving cartoon characters smoother.
  2. Poses is a list of length 72.
  3. Global translation is a list of length 3. If you don't need global translation, just go into [0,0,0].
  4. Current keyframe id is an integer.If you insert keyframes, you should set it to the correct keyframe id. If you don't insert keyframes, just set it to 0.

Steps

  1. Install the addon in Blender
  2. Run data server
  3. Press Ctrl+E in Blender to run addon
  4. Press A in Blender to stop addon or wait until the data transfer is complete

In step 3, you'd better select Armature, otherwise bugs may occur. Also, the mouse must be placed in the 3D viewport area(where the model is), otherwise the addon will not run.

Something about Blender

If you're not familiar with Blender, I've placed a blender project in the resources folder to help you.All you need to do is open it and follow Steps to achieve the effect shown in the Demo.(It's better to know something about animation in Blender.)

If you need a video background in the demo, select Compositing in the top menu bar, click Open Clip in the Movie Clip, and select your video.

图 2

If you are familiar with Blender and want to use your own models, you should make sure it's armature is SMPL's skeleton. The armature should name Armature and each bone has the same name as the bones in demo model(Only the 24 bones of SMPL skeleton are needed, and the fingers don't need to change their names).

图 3

Comments
  • How to run .fbx file to control the charater

    How to run .fbx file to control the charater

    Hello. I have successfully run the demo of ROMP, which exported .fbx file. And currently I want to use the .fbx to control the character. Can you provide steps for the video demo? I can only see the camera one

    opened by CheungBH 31
  • [simple-romp] How to use it?

    [simple-romp] How to use it?

    I tried to use it with simple-romp but it did not work. I already created an issue at the ROMP repository and described my problem here and here in detail.

    The author of the ROMP repository answered there:

    About live blender driving, please refer to this repo. https://github.com/yanch2116/CharacterDriven-BlenderAddon My colleague is responsible for maintaining this funciton now. Best regard.

    and closed the issue.

    @yanch2116 So how to solve it?

    My goal is to send the positions and quaternions of ROMP (or any other SMPL based solution) over the VMC protocol to other application (not Blender). Any ideas, whats the best way to do it?

    opened by vivi90 26
  • Could I ask for a detailed instruction on how to change the 3D Character?

    Could I ask for a detailed instruction on how to change the 3D Character?

    Also, I would like to know if there is a way to replace the 3D Character before I run the script and process the video so that I don't have to manually change it every time?

    opened by XXZhe 12
  • Can't connect ROMP into addon

    Can't connect ROMP into addon

    I use ROMP v1.1 on Windows machine. But CDBA works with version 1.0 of ROMP I read installation and using documentation of ROMP v1.0 but i can't figured out all How can i use this addon properly? I'm not good at programming i'm animator and interested in your project. Can you write steps how to install and use ROMP with this addon,please?

    image

    opened by hasanleiva 11
  • Hi, Looks like change another mixamo character not work

    Hi, Looks like change another mixamo character not work

    I have ROMP driven SMPL model looks OK, but when using CDBA (I using script locally) driven mixamo model, result looks wrong:

    image

    I am using this bone mapper in your repo:

    bones_mixamo_smpl_mapper = {
        "Hips": "Pelvis",
        "LeftUpLeg": "L_Hip",
        "RightUpLeg": "R_Hip",
        "Spine2": "Spine3",
        "Spine1": "Spine2",
        "Spine": "Spine1",
        "LeftLeg": "L_Knee",
        "RightLeg": "R_Knee",
        "LeftFoot": "L_Ankle",
        "RightFoot": "R_Ankle",
        "LeftToeBase": "L_Foot",
        "RightToeBase": "R_Foot",
        "Neck": "Neck",
        "LeftShoulder": "L_Collar",
        "RightShoulder": "R_Collar",
        "Head": "Head",
        "LeftArm": "L_Shoulder",
        "RightArm": "R_Shoulder",
        "LeftForeArm": "L_Elbow",
        "RightForeArm": "R_Elbow",
        "LeftHand": "L_Wrist",
        "RightHand": "R_Wrist",
        "LeftHandIndex1": "L_Hand",
        "LeftHandMiddle1": "L_Hand",
        "RightHandMiddle1": "R_Hand",
        "RightHandIndex1": "R_Hand",
    }
    bones_smpl_mixamo_mapper = {v: k for k, v in bones_mixamo_smpl_mapper.items()}
    bone_name_from_index_character = {
        k: bones_smpl_mixamo_mapper[v] for k, v in bone_name_from_index.items()
    }
    
    

    Do u know why?

    also the hand look not right.

    image

    opened by jinfagang 11
  • Detection variable: outputs = {'poses': poses, 'trans': trans[0]}

    Detection variable: outputs = {'poses': poses, 'trans': trans[0]}

    Hey, Yanchxx,

    May I ask, what is the output variable, I think "trans" is the translation between camera and object.

    What are the poses, is it the location x, y, z, or rotation x, y, z, degrees?

    Thank you 👍

    opened by zhangby2085 10
  • 导入fbx时骨骼方向乱了

    导入fbx时骨骼方向乱了

    @yanch2116 你好,非常酷的工作!我基本把整个项目跑通了,但我有两个问题想进一步请教一下你。 、 1.导入fbx时骨骼的方向错乱了,参见:https://www.bilibili.com/read/cv2520452 。但是这个文章里的方法没有完全解决方向错乱的问题,所以我想请问一下你,你是怎么解决这个问题的呢? 2.我想请教一下,如何将编辑骨架让其和smpl的骨架一致呢? 望不吝赐教!万分感谢!

    opened by syguan96 8
  • Use keyframes to prevent pose shaking

    Use keyframes to prevent pose shaking

    I found that the avatar will shake drastically when running the webcam demo. And I set the mode =1 to insert keyframe record the webcam results for playback.

    Now, if we set one more condition for inserting keyframe, only the frame_idx % 3 == 0 or frame_idx % 5 == 0. This would allow the avatar to move along these keyframes much smoother.

    However, is there a way to let the webcam demo runs in real-time using the keyframe strategy I said? This skip keyframe strategy only seems to work with the recorded playback. The character is still moving on each frame when we are actually running in real-time.

    opened by ZhengdiYu 4
  • Can I rotate the scene while running scripts?

    Can I rotate the scene while running scripts?

    Hi, I'm able to run to demo with blender now, but it turns out that the view is locked while the script is running.

    If there's a way to enable rotation?

    opened by anzisheng 4
  • Which version of Blender are you using?

    Which version of Blender are you using?

    I met the following error while running Beta.blend. I'm using blender 2.83.9. What's the expected version?

    Read blend: E:\Workspace\blender_test\addons\CharacterDriven-BlenderAddon-master\blender\Beta.blend 0 meshes freed Error: File written by newer Blender binary (290.0), expect loss of data!

    opened by sylyt62 4
  • multiple people

    multiple people

    Hey, yanch2116, very impressive job done in visualizing 3D characters. I tested and it works, one question to ask, the code romp_server.py has the setting.show_largest=True, I am thinking, what is the condition with multiple people in the webcamera, when there are many people, the blender character keeps on shifting. Are they ways to solve this? 1, keep on the object tracker in single object_ID 2, visualize multiple 3D characters as input in the camera

    opened by zhangby2085 3
Owner
犹在镜中
犹在镜中
implementation for paper "ShelfNet for fast semantic segmentation"

ShelfNet-lightweight for paper (ShelfNet for fast semantic segmentation) This repo contains implementation of ShelfNet-lightweight models for real-tim

Juntang Zhuang 252 Sep 16, 2022
Cookiecutter PyTorch Lightning

Cookiecutter PyTorch Lightning Instructions # install cookiecutter pip install cookiecutter

Mazen 8 Nov 06, 2022
Learning to See by Looking at Noise

Learning to See by Looking at Noise This is the official implementation of Learning to See by Looking at Noise. In this work, we investigate a suite o

Manel Baradad Jurjo 82 Dec 24, 2022
3DMV jointly combines RGB color and geometric information to perform 3D semantic segmentation of RGB-D scans.

3DMV 3DMV jointly combines RGB color and geometric information to perform 3D semantic segmentation of RGB-D scans. This work is based on our ECCV'18 p

Владислав Молодцов 0 Feb 06, 2022
this is a lite easy to use virtual keyboard project for anyone to use

virtual_Keyboard this is a lite easy to use virtual keyboard project for anyone to use motivation I made this for this year's recruitment for RobEn AA

Mohamed Emad 3 Oct 23, 2021
Instant neural graphics primitives: lightning fast NeRF and more

Instant Neural Graphics Primitives Ever wanted to train a NeRF model of a fox in under 5 seconds? Or fly around a scene captured from photos of a fact

NVIDIA Research Projects 10.6k Jan 01, 2023
Code for our NeurIPS 2021 paper: Sparsely Changing Latent States for Prediction and Planning in Partially Observable Domains

GateL0RD This is a lightweight PyTorch implementation of GateL0RD, our RNN presented in "Sparsely Changing Latent States for Prediction and Planning i

Autonomous Learning Group 16 Nov 03, 2022
Vowpal Wabbit is a machine learning system which pushes the frontier of machine learning with techniques such as online, hashing, allreduce, reductions, learning2search, active, and interactive learning.

This is the Vowpal Wabbit fast online learning code. Why Vowpal Wabbit? Vowpal Wabbit is a machine learning system which pushes the frontier of machin

Vowpal Wabbit 8.1k Jan 06, 2023
Scalable training for dense retrieval models.

Scalable implementation of dense retrieval. Training on cluster By default it trains locally: PYTHONPATH=.:$PYTHONPATH python dpr_scale/main.py traine

Facebook Research 90 Dec 28, 2022
DLWP: Deep Learning Weather Prediction

DLWP: Deep Learning Weather Prediction DLWP is a Python project containing data-

Kushal Shingote 3 Aug 14, 2022
Ranger deep learning optimizer rewrite to use newest components

Ranger21 - integrating the latest deep learning components into a single optimizer Ranger deep learning optimizer rewrite to use newest components Ran

Less Wright 266 Dec 28, 2022
Neural network chess engine trained on Gary Kasparov's games.

Neural Chess It's not the best chess engine, but it is a chess engine. Proof of concept neural network chess engine (feed-forward multi-layer perceptr

3 Jun 22, 2022
Reimplementation of the paper `Human Attention Maps for Text Classification: Do Humans and Neural Networks Focus on the Same Words? (ACL2020)`

Human Attention for Text Classification Re-implementation of the paper Human Attention Maps for Text Classification: Do Humans and Neural Networks Foc

Shunsuke KITADA 15 Dec 13, 2021
Code for sound field predictions in domains with impedance boundaries. Used for generating results from the paper

Code for sound field predictions in domains with impedance boundaries. Used for generating results from the paper

DTU Acoustic Technology Group 11 Dec 17, 2022
FridaHookAppTool - Frida Hook App Tool With Python

FridaHookAppTool(以下是Hook mpaas框架的例子) mpaas移动开发框架ios端抓包hook脚本 使用方法:链接数据线,开启burp设置

13 Nov 30, 2022
Face Mask Detector by live camera using tensorflow-keras, openCV and Python

Face Mask Detector 😷 by Live Camera Detecting masked or unmasked faces by live camera with percentange of mask occupation About Project: This an Arti

Karan Shingde 2 Apr 04, 2022
An experiment on the performance of homemade Q-learning AIs in Agar.io depending on their state representation and available actions

Agar.io_Q-Learning_AI An experiment on the performance of homemade Q-learning AIs in Agar.io depending on their state representation and available act

1 Jun 09, 2022
Veri Setinizi Yolov5 Formatına Dönüştürün

Veri Setinizi Yolov5 Formatına Dönüştürün! Bu Repo da Neler Var? Xml Formatındaki Veri Setini .Txt Formatına Çevirme Xml Formatındaki Dosyaları Silme

Kadir Nar 4 Aug 22, 2022
Unsupervised Learning of Probably Symmetric Deformable 3D Objects from Images in the Wild

Unsupervised Learning of Probably Symmetric Deformable 3D Objects from Images in the Wild

1.1k Jan 03, 2023
Fast and robust clustering of point clouds generated with a Velodyne sensor.

Depth Clustering This is a fast and robust algorithm to segment point clouds taken with Velodyne sensor into objects. It works with all available Velo

Photogrammetry & Robotics Bonn 957 Dec 21, 2022