Google (?) PhotoSketch – Rapid Photo Modelling

Photomodelling has been a slightly disregarded technology since Adobe acquired and subsequently dropped CANOMA (see our post on Great Software from the Past). The release of Google Earth and more recently Microsoft’s Virtual Earth have bought the issue of rapid photomodelling to the forefront of the race to capture cities digitally.

To date Microsoft have been leading the way with their photorealistic automatic capture of cities such a New York (see Populating the Digital Earth) with the aim of 500 new cities in the next year. While this is impressive it lacks the ‘community owned’ approach of Google and thus numbers of users.

As such it is without surprise that Google has just released a video detailing its new product ‘PhotoSketch‘ – a combination of automatic camera calibration from photographs and the simplicity of modelling using SketchUp. PhotoSketch addresses a number of the points needed for rapid photomodelling for Google Earth and general city model development.

The presentation is 58 minutes long but well worth the time, grab a cup of tea, sit back and take a look at PhotoSketch:

Aimed as ever with Google at the average user rather than the high end photogrammetric market it has the potential to change the modelling industry.

Now where are all those photographs we took for CANOMA?

Thanks to The Google Earth Blog for the link..

Author Bio

Andy is Professor of Digital Urban Systems at the Centre for Advanced Spatial Analysis at University College London.


  1. Anonymous - July 18, 2007

    Great bunch of people at the Google Site.

  2. Anonymous - July 18, 2007

    Not sure this is yet a Google product. It is a video of a talk given via the Google Tech talk series. The developer/researcher is lead by G Wolberg out of NY.

  3. Andy - July 18, 2007

    Yep good point – I’ve put a (?) on the Google title..

    Thanks anon..


  4. Richard - July 21, 2007

    I think the guy talking about the level of model detail has a fair point. If the model simply isn’t rich/detailed enough the end product will look fairly amateurish.

    I am more interested in the idea of the model being derived from the texture, and as such having as much detail as the images. Applying the texture to the model to get the end product in this way limits the capabilities to the skill of the modeler.

  5. phil - April 29, 2009

    Looks like this is getting close to being available:

  6. gw - July 7, 2009

    The PhotoSketch plugin is now available!
    Check it out at

Leave a Reply to gw Cancel reply