English|中文
The main reason for this project is that I encountered a scenario where the STEP and IGES models need to be displayed on the Web, but the web3d libraries on the market do not support this format, and the direct display of STL files uploaded by users will consume a lot of bandwidth or CDN Traffic, converted to compressed gltf would be more appropriate.
Demo assets model effect compare:
| model type | file path | Convert time | Origin size | After size |
|---|---|---|---|---|
| stl | assets/test.stl | 2368.890ms | 7.6 MB | 86 KB |
| iges | assets/test.iges | 1641.226ms | 1 M | 111 KB |
| stp | assets/test.stp | 2969.200ms | 5.1 MB | 217 KB |
| fbx | assets/Samba Dancing.fbx | <1000ms | 3.7 MB | 614 KB |
support input format： STL/IGES/STP/OBJ/FBX
support output format： GLTF/GLB
I organized my thoughts into a blog: STEP and IGES models are converted to the web-friendly glb format
PS: My blog is write by Chinese, if you are non-Chinese native speaker, you should take a Google Translate tool for well.
Project status: stable
- Basic project structure planning and interface design
- Conversion and compression code implementation
- Add obj format to darco gltf
-
Related API implementation(not so useful, droped) - docker image packaging
- write easy to use convert.sh
- online convert preview
- [bug] stp convert to gltf is too large
- grpc support
- rpc should response error detail
- rpc docker server logs output problems
Support GRPC, convert code refactor, fix bugs.
Add English document, fix bugs.
Basic commit, core feature complete, support shell convert.
I tried to use assimp, but the result under the test of stl/iges/obj conversion is not good. I used https://hub.docker.com/r/dylankenneally/assimp docker environment for testing, you can have a try on it.
Model conversion is a very performance-consuming and slow-speed service. The upload and download of the model will consume bandwidth. If it is deployed directly on your own server, it will be a very bandwidth-intensive and CPU-consuming task. For the most common method to upload and download large files is to introduce OSS and CDN with dynamic expansion of queues and back-end services, but the deployment cost and implementation cost will be relatively high.
Due to the trouble of environment configuration and other reasons, the command line mode still needs to rely on docker. The command line mode is suitable for simple invocation on the server side. The conversion process blocks the processes to be synchronized and cannot be deployed in a distributed manner to increase concurrency.The most recommended way is to use grpc with docker deployment to make rpc, which can be synchronous or asynchronous, and will be easy to extend.
PS：When there are too many simultaneous conversion models in the command line mode or a single model is too large, there is a risk that the server providing the web service is stuck
You can convert model online (<100MB) powered by modelbox-sdk，preview link: https://wangerzi.gitee.io/modelbox-sdk/examples/index.html
Based on GRPC, and it will be more convient to build a dynamically expanded service cluster, we support uploading zip/model source files, for the compatibility of each model, response files are all in zip, you need to decompress it after got it.
You should run server-side rpc service by docker, please make sure 8999 port is usable and wj2015/3d-model-convert-to-gltf:latest image is up to date, command:
$ docker pull wj2015/3d-model-convert-to-gltf:latest
$ docker run -d -p 8999:8999 wj2015/3d-model-convert-to-gltf:latestWhen using grpc in this project, please copy server/rpc/protos/converter.proto , and generate a code template according to the language of caller and enjoy it. Official document: Support Language
If this project is helpful to you, you can commit another examples and PR, such as php/golang/Nodejs rpc call examples.
| name | code | comments |
|---|---|---|
| Python rpc client usage example | server/examples/python/rpc_client.py | convert and save |
Download the convert.sh, and grant execution authority, execute the following command, the second param should choose in stl|stp|iges|obj|fbx, please determine according to the file type
The script depends on the docker environment, so you should prepare the Docker environment first. Command mode is not support zip file convert, beacause docker volume will auto sync picture or mtl assets to docker container.
convert.sh stl inputpath.stl outputpath.glb # convert to glb single bin file
convert.sh stl inputpath.stl outputpath.gltf # generate gltf fileIn the assets directory, there are four test files test.stl test.stp`` test.igs E 45 Aircraft_obj.obj Samba Dancing.fbx, copy it to the project path, and you can see the convert result.
If you got this error when you use php-fpm or other language executor to execute convert.sh, you can add your execute user to docker group to avoid this problem.
usermod -a -G docker nginx
docker: Got permission denied while trying to connect to the Docker daemon socket at unix:///var/run/docker.sock: Post http://%2Fvar%2Frun%2Fdocker.sock/v1.40/containers/create: dial unix /var/run/docker.sock: connect: permission denied.
By calling in other languages, you can synchronously determine whether the output file exists to determine whether the conversion is successful, such as:
<?php
$out = 'out.glb';
$input = 'test.stl';
$type = 'stl';
// please be careful about execution attack
shell_exec('convert.sh '.$type.' '.$input.' '.$out);
if (file_exists($out)) {
echo "convert result:" . $out;
} else {
echo "convert failed";
}Under the docker host machine installed with docker, run the following command to pull the image (about 4G)
docker pull wj2015/3d-model-convert-to-gltfInside the container and execute conda run -n pythonocc python convert.py [stl|step|iges|obj|fbx] input.stl out.glb can convert model synchronous.
If there is a demand for multi-machine load, you can use nginx proxy / docker container with rpc to do a simple load balancing, or use message queue with producer and consumer. The HTTP API or queue needs to implement your own logic.
At first install docker and docker-compose, refer to official documents: Docker Desktop
Then, enter to environment/ documents, execute docker-compose up, the execute result is as follows to indicate success
user@MacBook-Pro environment % docker-compose up
Recreating 3d-model-convert-to-gltf-app ... done
Starting 3d-model-convert-to-gltf-redis ... done
Attaching to 3d-model-convert-to-gltf-redis, 3d-model-convert-to-gltf-app
3d-model-convert-to-gltf-redis | 1:C 09 Oct 2020 03:03:29.150 # oO0OoO0OoO0Oo Redis is starting oO0OoO0OoO0Oo
3d-model-convert-to-gltf-redis | 1:C 09 Oct 2020 03:03:29.150 # Redis version=6.0.1, bits=64, commit=00000000, modified=0, pid=1, just started
3d-model-convert-to-gltf-redis | 1:C 09 Oct 2020 03:03:29.150 # Warning: no config file specified, using the default config. In order to specify a config file use redis-server /path/to/redis.conf
3d-model-convert-to-gltf-redis | 1:M 09 Oct 2020 03:03:29.152 * Running mode=standalone, port=6379.
3d-model-convert-to-gltf-redis | 1:M 09 Oct 2020 03:03:29.152 # WARNING: The TCP backlog setting of 511 cannot be enforced because /proc/sys/net/core/somaxconn is set to the lower value of 128.
3d-model-convert-to-gltf-redis | 1:M 09 Oct 2020 03:03:29.152 # Server initializedIf there are port conflicts, initialization failures and other abnormal situations, please check and search the information according to the error information.
Create a new terminal, execute docker ps for this current execute docker containers, the execution result is as follows to indicate success
user@MacBook-Pro 3d-model-convert-to-gltf % docker ps
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
69b684ed7755 wj2015/3d-model-convert-to-gltf "conda run -n python…" 3 seconds ago Up 2 seconds 3d-model-convert-to-gltf-appNext, enter the container to execute the command and enter the pythonocc conda environment. Executing the script in this environment can facilitate code changes and debugging
wangjie@MacBook-Pro 3d-model-convert-to-gltf % docker exec -it 3d-model-convert-to-gltf-app /bin/bash
(base) root@5efd6ef96814:/opt/3d-model-convert-to-gltf# conda activate pythonocc
(pythonocc) root@69b684ed7755:/opt/3d-model-convert-to-gltf# python server/convert.py
Params not found, format: python convert.py [type] [file path] [out file path]Mainly for developers who cannot run docker, you can try to use this method to build a development environment.
Create conda virtual environment:
conda create -n 3d-model-convert-to-gltf-pythonocc -c dlr-sc -c pythonocc pythonocc-core=7.4.0rc1
conda activate 3d-model-convert-to-gltf-pythonocc
pip install -r server/requirements.txtYour local node version need 12.0.0, or got error when you run the gltf-pipeline command, and you should install gltf-pipeline and obj2gltf packages.
Install nvm by this script(MacOs or Linux), you can download .exe executable file from https://github.com/coreybutler/nvm-windows
curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.35.3/install.sh | bashInstall and use node 12.0.0, with other packages.
nvm install 12.0.0
nvm use 12.0.0
npm install -g gltf-pipeline obj2gltfThen, download FBX2glTf from https://github.com/facebookincubator/FBX2glTF and put it to environment dir.
PS: you should rename FBX2Gltf to fbx2gltf for Unified invoking
Understand the code and the file structure, submit the PR after the modification. Welcome to email me [email protected].
3DModelConvertToGltf is licensed under the Apache License, Version 2.0. See LICENSE for the full license text.