In this tutorial you will learn how to start editing and rendering your my3dscanner project files, visualize and evaluate your model. The photo above illustrates a typical result which you will get after this tutorial and basically with no additional work.
The sample project used in this tutorial was made with 7.5 MP camera. I took source photos of an architectural fragment in the Summer Yard of the Forbidden City in Beijing, China. Processing by my3dscanner took 49 minutes. Fusion rate of 9 photos were 100%, number of points 455K.
Well, the front photo above is a bit different from what I initially see in my3dscanner in-browser viewers:
Viola! point cloud viewer tells me if the project was successful at all. Balloon It! mesh viewer gives me an idea of mesh quality and if the mesh has any potential for further processing. Just looking at my model I think it is worth trying to create something interesting
I downloaded point cloud and mesh to my computer and unzipped the files. Then I started MeshLab and opened .ply (File -> Open), then .obj as a new layer (File-Open as new layer) and opened a layer dialog.
Let’s look at the point cloud: not bad, good density, but a lot of holes: my points of view were too low to capture the top (unfortunately I am a short guy). Left part of the object was in shadow which has resulted in a low density surface. The problem with color point cloud is that it can be very misleading. A project may look fabulous as color point cloud but have no value as a solid 3D model.
Looking at mesh as a wireframe makes a lot of sense since in a mesh viewer a model usually looks like a shell or a balloon. This is because my3dscanner tries to build a watertight mode. But now I can see the object, not the balloon.
Mesh looks too smooth. Let me explain: our engine automatically removes outliers, generates normals, fills holes, smooth the surface but at the cost of some details loss. My3dscanner radically removes the scanning noise (unwanted points due to improper photographing) as successfully as intrinsic surface details. Noise and actual object’s details appear similarly unwanted to our merciless engine. But it accurately retains the main shape of the object. We will add some lost details later.
My3dscanner delivers all the volumetric and color information needed for visualization. Let’s transfer point cloud color information to the mesh (facets). This is very easy in MeshLab: Filters -> Sampling -> Vertex Attribute Transfer. In a dialog window chose .ply as a source and .obj as target. Check transfer color and normals, click Apply. This is how your mesh will now look:
In a Layer dialog click on either layer with right mouse button, choose Flatten visible layers and click OK on a default. You will end up with one merget object.
Let’s start some editing. I want to crop the beast from the wall. It’s easy as well: select an area you want to keep, invert the selection (Filters -> Selection -> Invert Selection: Apply) and delete it (Delete current set of faces and verticles). We end up with 167K faces model.
At this point you may want to reduce triangle count: Filters -> Remeshing… -> Quadric Edge Collapse Decimation.
Save your .obj as obj, ply, wrl and even stl or 3ds (for 3ds you will have to decimate your model). If you want to go directly to Meshmixer or Blender, I recommend to save as .ply (check Color and Normal, uncheck Binary on save).
We will continue editing this model soon. But I have a question for you: is this a dragon, a crocodile or an elephant? We need to decide before we start some exiting 3D sculpting and painting work in the next tutorial.