If you already have CAD 3D model of object, you can upload it instead of photographs. Ideally you should export your object in .obj format, and you should upload also file with associated textures. Alternatively we can also process .step files.
If you don’t have a 3D model, we can easily make the model based on photographs uploaded by you. We use photogrammetry algorithm to create 3D representation. We need to see each point on the object on 3-6 different photographs, made from different angles. Aim is not to produce perfect reconstructions, as for 3D printing, but detailed enough reconstructions that can be annotated and explained for educational purposes. Thus, ~60 photographs made by modern mobile phones should be sufficient. To improve quality of reconstruction, high-resolution camera is can be helpful, as well as larger number of photographs (~250). Please follow the following guidelines for best results:
From which angles should I take photographs?
- do not move the object; instead, move camera around the object
- if the object is small, take at least three series of photographs: move camera in circle around the object, taking photograph every ~10-15 degrees (20-40 photoraphs per circle) and take three series of photographs, one in plane of the object, one from plane lower then the object, and one from plane above the object.
- if the object is large, try to take a series of photographs, “tiling” the views. Try to keep approximately 2/3 overlap between successive views.
- note that one does not to keep whole object always in view of camera - indeed for larger objects with lots of details this is impossible. However taking few photographs where whole or at least bigger sections of object are visible, in combination with many close up photographs of parts of the object, is often useful.
- sometimes is useful to place a pieace of text/ hand written note / printed document or similar next to object if the object is rotationally symmetric. That text/note will break the rotational symettry of the object, and allow easier reconstruction of camera locations during the 3D reconstruction.
How to set up camera optics?
- depth of field should be large, so that whole object of interest and its immediate surroundings is in focus - i.e. set the aperture at f/8 or even more reduced aperture value. It is better to increase ISO if needed, adding bit more noise to photograph, then to open aperture reducing depth of field.
- do not process files or change EXIF information associated with photographs - we need unprocessed photographs and information from EXIF files on camera to do reconstructions.
- do not worry if you capture a lot of surroundings of the object - we cut these in final processing.
How should be light for taking photographs?
- avoid glass and strong reflections
- to prevent strong shadows and/or reflections try using diffuse light if possible (e.g. lamp reflected off the large white surfaces (wall), or if shooting outside, try taking photographs around noon when sun is straight above the object and there are not strong reflections.
Finally, note that even as little as 30 mobile phone photographs can be enough to make model that has educational value for someone on the other side of the world. Even poor models combined with good stories, can stimulate imagination and help understanding. So don’t hestitate to record interesting apparatus wherever you spot one, even if in hurry, and without good camera. Good luck, and thank you for contributing!