1,000 models, 100 pages, 10 models/page


Building the 3rd Dimension: VisualSize offers the best-in-class technologies for reconstructing 3D information from regular digital pictures. Our technologies use pictures from mass-market, consumer-grade digital cameras, camcorders, and phones without the need of any special equipment, setup, calibration, markers, user training, or manual intervention.


To learn more about Visualsize's 3D offerings and how Visualsize compares to the competitors, please peruse our overview slides. Or, if you are an academic type and prefer reading more rigorous and detailed analyses, please check out our technical report.



     Building 3D Models

Building a 3D model has never been this easy. Take pictures at random locations around an object of interest using your digital camera, camcorder, or phone. Upload them to our program, and it will automatically generate a 3D model capturing both the 3D structure and appearance of the object. You can manipulate, measure, and interact with the 3D object or export it to a CAD program. No special equipments, no tedious setup and calibration, and no markers. Just point, shoot and upload. There is no other solution that comes close. To learn more about the advantages of VisualSize's technologies, please click on "FAQ" below. To learn how Visualsize's technologies compare to others (notably University of Washington/Microsoft's Bundler, Bundler + PMVS2 of University of Illinois and Autodesk's Project Photofly), click on the appropriate buttons below.

[ Try it out! ] (public server courtery of Computer Vision Lab of UCSB)

[ FAQ ]      [ Demo page ]     [ Interactive Demo page ]    
[ Accuracy Study Against Ground-Truthed, 3D Data sets ]    




     3D Measurements

VisualSize Inc. provides photo metrology (measurement) solutions with the highest accuracy in the market. Users can take two pictures of a 3D scene to obtain precise, true 3D dimension measurements (length, angle, area, volume, etc.) of objects in the scene that are 10x more accurate than other photogrammetry solutions.

[ Watch Tutorial and Try it Out! ] (public server courtery of Computer Vision Lab of UCSB)

[ FAQ ]          




     Building Panoramic Images

If you take multiple photos by rotating the camera while standing at a location, Mosaic3D can “stitch” the photos together seamlessly into a panorama with a significantly larger field of view and a much higher spatial resolution. A high-resolution panoramic picture can be sent to a specialty print shop to produce hard copies for wall mounting and display. A low-resolution counterpart can be used in a Web site for a variety of purposes, e.g., product demo, photo tourism, real estate sale. Again, VisualSize’s Mosaic3D technology, like all others provided by VisualSize, works with any consumer-grade digital cameras and does not require special equipment, setup, and training to make it work.

[ Try it out! ] (public server courtery of Computer Vision Lab of UCSB)

[ FAQ ]    [ 360^o panorama demo ]    [ General mosaic demo ]    [ An Android application of geotagged panorama by Brian Dunlay and William Moy of UCSB using our Mosaic3D engine ]




     Photo Browsing

PhotoNAV3D automatically infers the camera movement in a 3D environment based on regular digital-camera photos so that you can navigate the photos based on their 3D relations. PhotoNAV is more robust and able to infer camera motion from much broader scenarios than Microsoft's Photosynth (see the comparison study here).

VisualSize provides PhotoNAV3D to photo sharing sites as a white labeled software service to enable any photo sharing service site to offer, under its own brand, its users the functions that is superior to Microsoft's Photosynth.



[ Try it out! ] (public server courtery of Computer Vision Lab of UCSB)

[ FAQ ]    [ Demo and Comparison Study ]