This morning I've found this app from autodesk. It's called 123D Catch. ==> http://www.123dapp.com/catch It creates a point cloud mesh from pictures. It needs 8-70 pictures from different angles with the same focal size (don't change zoom). Here is my first test: I've set my phone on picture interval and walked around my car and took some random pictures. Loaded the pictures in the program and let it do its thing for an hour. It doesn't work well on reflective and transparent objects. And the surface needs some references (non uniform colors, text, ...). So my car wasn't the best subject. I think a race car has better results.
Lol we have same car. Yeah this is a nice app, especially for small objects. If you take care to make a proper 360 around, possibly with not too big distance between steps, it's gonna work fine to generate reference mesh. Fun.
Don't know... I've installed the iPhone app an hour ago and I've "scanned" my garden with it. The app sends the pictures to the autodesk cloud server. There they render the mesh and the UV map. (this takes a few hours, i'm still waiting) Once it's finished, it sends the mesh back to your phone so you can preview it or i can go to my laptop and download the 3D result with the 123D catch PC version and Export it to FBX and load it in Maya, Max, Blender, .... whatever you are working with.
These things are producing a network of triangles which is not really optimized for game engines, but if you manage to clean up that mesh, it's for sure possible to use it for a track. Isn't really that much different from what you do with real point clouds converted to triangle networks. The only (big) difference is you cannot really use this technology to scan a track, but it's a nice way to grab some localized detail you want to port into your 3D scene. Limit of this specific software is it's limited to 70 pictures per asset, if I remember well. Of course we are saying about low frequency network....as you can't really model a gravel trap in full 3D, stone by stone, or a square meter of that gravel will be heavier than a full track plus a full grid of cars... Maybe more... (and btw this technology hasn't the resolution to offer such details)
I've known about it for long, bit as Tuttle says it's not really suited for "scanning" tracks. The result is very impressive though, even with the cars all white and slightly reflective body (y) If only I owned a drone, good camera etc. it would be cool for making building reference models
Thanks for the link. Astonishing piece of software. These type of features make me think that we will see Google drones in the future 3d scanning the whole planet. Enviado desde mi ONE A2001 mediante Tapatalk
For car modders... I've combined 2 scans and matched them in Maya. BTW: it doesn't have to be a full 360 scan...
Looks like laser scanning has some very inexpensive competition. The accuracy needs a bit of work, but this has a lot of potential.
Yes it needs some improvement. Detailed textures are quite finely scanned whereas flat textured surfaces seem to present more errors. What it would also need is something to simplify the mesh up to a reasonable level without losing much detail. As a reference for building something over it it can be very helpful. Enviado desde mi ONE A2001 mediante Tapatalk
It's not about the flat surfaces, it's about reflections and lack of visual details for picking reference points.
I didn't refer to flat surfaces, but to flat textures. As you say lack of detail for reference. Reflections provoke the same lack of reference to match different POV. Enviado desde mi ONE A2001 mediante Tapatalk
