Welcome to the Onshape forum! Ask questions and join in the discussions about everything Onshape.
First time visiting? Here are some places to start:- Looking for a certain topic? Check out the categories filter or use Search (upper right).
- Need support? Ask a question to our Community Support category.
- Please submit support tickets for bugs but you can request improvements in the Product Feedback category.
- Be respectful, on topic and if you see a problem, Flag it.
If you would like to contact our Community Manager personally, feel free to send a private message or an email.
Large model local computing, mesh features
alex_godwin198
Member Posts: 7 ✭
I am attempting to use medium sized 3D scans I took of my car, for example the wheel well and suspension components of just one corner. That model is too large for OnShape and makes creating parts within that model basically impossible. Its a slide show with features taking a long time to generate. So there is no way to do a full car sized model in Onshape without reducing the mesh significantly. I found meshes over 50 mb OnShape cannot deal with which is unfortunate because thats not even a large mesh.
Is there ever going to be a way to utilize local hardware to help the program like how Fusion 360 does? That program can run the things I mentioned no problem which is why I compare them. I much prefer OnShape over Fusion so if this was ever solved it would be awesome. My current work flow is to bring the models into Fusion, use their remesh and reduce mesh features then bring it into OnShape, since OnShape has no such equivalent features that I could find.
Answers
Is your GPU enabled in chrome? this should help with large models usually
Hello Alex, your large mesh probably has millions of triangles and needs a mesh decimation tool. Fusion has these built in, but for onshape you'll have use some free tools like MeshLab or the tool in Blender. And then Meshmixer to repair it. I'm working on an Onshape native mesh to cad tool ParameshAI and it will also include a free mesh decimator and cleaner.
I started working on something similar after the most recent CNC Kitchen video for Bumpmesh, also using the OCCT kernel to do the heavy lifting of surface generation and stitching. I got an early working build using the MakerVolume functions in the kernel but that approach is a great way to make a geometric landmine of exponential complexity for more than the most basic of parts. I'm currently reading into the rest of the boolean functions in OCCT to find a more robust approach than my initial attempt but it's a new kernel to me so I'm having to learn from scratch what works and what doesn't. The plan for mine is to develop a tool that allows for more user control over the surface painting and what the groups of mesh faces get interpreted as and just dump the result as a featureless .step file though and not try to reconstruct features.
Derek Van Allen | Engineering Consultant | MeddlerOh nice! Using pure OCCT sounds pretty hard. I'm using cadquery which is basically a python script layer over OCCT. I'm also segmenting services for meshes but my software will generate it automatically. But the idea of leveraging software for surface painting seems like a good idea for complicated parts
Yes GPU acceleration is on and working. Its not doing as much as Fusion or other software does that more fully utilizes the local hardware. It would be nice if OnShape had an app that would run local on your PC instead of a browser because it seems to be limited on power because of the cloud only compute.
HI, the model is about 70mb after using Fusion to cut it down. It runs perfeclty smooth in fusion and slolidworks, its becuase Onshape is cloud based and doesnt seem to be using very much local hardware resources very well. What Im saying is other CAD software can open and spin these scans no problem, only Onshape has issues with it. I am using scans more and more and I need the detail in the scans, I cant just remesh it until its fast in Onshape becuase that is too simplified and looses details. Ive tried, Onshape seems to struggle on any scan file over 60 mb. That severely limits the usage of scan data in Onshape. For the particular scan I am using if I reduce the mesh until the model spins smoothly then the hole detail is no longer round.
To the Onshape support people, is there a way to get better performance or "is this it" in terms of compute performance? Even after remeshing in other programs getting files to under 50 MB, its still substantially slower in Onshape compared to other CAD programs for the exact same ~45MB scan file. Are you ever going to develop a local app that utilizes local hardware better than GPU acceleration in a browser? Or maybe even just improve the current GPU acceleration and allow it to use more of the GPU. I never see it go above low GPU usage levels and its not even coming close to using very much RAM, so this tells me majority of the compute is being done by the cloud. Since you limit cloud resources, allowing us to utilize our local hardware would save users time but also save your server load.
Yes, we are limited by the amount of GPU RAM the browser can address (Firefox does the best job). No, we will not be creating a local app.