Autodesk 123D Catch : Where's the Catch? The Ease of “Scanning” 3D Objects with photographs from your Camera
123D Catch is a handy cloud-based application that will generate 3D models from photo image sets. It can give you useful models, & it is free. CADplace takes a second look, and shares a few lessons-learned.
CADplace : Tom Lansford
The new Siemens/Fujitsu workstation arrived for testing. I wanted to work through the evaluation of the system using a variety of software and hardware tools, and I wanted to have a bit of fun with my own models.
Since the 123D Catch client-program only works on Windows and I am usually Mac-based, I took advantage of this review to test 123D Catch more and generate some models at the same time. Autodesk gave a demonstration of the product to CADplace and with a first “hands-on” session, the application looked promising.
One topic to examine – very important in my opinion – is the applicability of photo-geometric data generation to different kinds of models, so that you know when this tools can be used to your best advantage.
Setting up 123D Catch
First, you'll find the link to 123D Catch Downloads on this page. The download and the installation is simple. You'll be giving Autodesk you email address since 123D Catch can process your model while you continue with your work and send an email when your model is ready.
You are ready to start with your first project. There are some useful Autodesk videos to watch before you start taking your first pictures.
Tips on your photographs and why light matters
The basic idea behind a photograph was to expose a film to light and let a chemical process run its course. Today, most of you use a digital sensor in place of film in your camera, but the concept remains the same. Photographs are based on light. Keep this in mind when you consider your pictures for 123D Catch.
Autodesk videos on line have some good tips such as maintaining the same exposure for your photos, and avoiding over-exposed & under-exposed photos. Over and under-exposed photos provide less visual information for 123D Catch to analyze which reduces the quality.
If you are outside, you want photographs without stark shadows, so avoid bright sunny days or else shoot in a shaded area. Likewise, if you are shooting indoors, then you want to have lighting as even as possible surrounding your objects. Try to avoid “hot” spotlight types of lighting. Instead, use as much reflective indirect lighting as possible, and try to minimize shadows.
“Stitching” and automated versus manual workflow
The best case scenario means that 123D Catch takes your pictures and sends back your 3D model – no changes or adjustments required. Up to this point, I have not found the 3D object and the set of photographs which does not require some manual stitching.
“Stitching” is the process of analyzing multiple photographs, recognizing the common objects, calculating the perspective relationships between them. In this way, 123D Catch is able to create a 3D space from your pictures and then give you the geometry and the textures.
After you upload your first photo-set to the 123D Catch cloud-based servers, the model you receive may seem to have very little in common with the object you photographed ! Don't be surprised. With just a little bit of manual stitching, your model will start to take shape.
123D Catch guides you through the manual stitching process. Essentially, you identify the same point on the object in multiple photographs. When each photograph has enough common points accurately identified, 123D Catch can generate a good 3D mesh.
Which brings us back to your photographs
If you think about manual stitching before you take your pictures, then you will see that you want to have very good, precise reference points in your photos. Maybe your object already has many such identifiable points naturally.
If you want to make life easy for yourself, consider adding your own fixed-reference points. For example, before photgraphing the 3Dconnexion SpaceMouse Pro, I placed multiple, small, round, colored stickers at the 4 corners of the SpaceMouse Pro. This gave every photograph 2 or 3 easy reference points for manual stitching.
More on uploading to the cloud & generating meshes
In the next part, I'll go into the cloud processing and mesh generation, and we'll take a look at exporting different file formats as well as generating animations and youtube uploads.
First 3D models
The first model testing was about trying objects and scenes with different characteristics. The SpaceMouse Pro has smooth surfaces, buttons, knobs, lights and text. The red bear has rich textures and bright colors.
You can see that the SpaceMouse Pro has reflective surfaces and a deep, monochrome (non-patterned) surface. So far, this has proven the greatest difficulty for me in producing a good, quality, continuous mesh.
|The bear has good, even lighting and exposure||The results on the SpaceMouse Pro look quite nice in some areas||But I have had problems generating geometry for the large, smooth, monochromatic palm-rest on the SpaceMouse Pro|