The Magic behind Glanz & Gloria Miss Switzerland 360° Superzoom

Published on SF.TV:
and MSN.CH:

After the mountains in Deep Zoom Zermatt and the paintings in Deep Zoom van Gogh, Cézanne and Monet I realized that it would be cool to replace the traditionally slow loading and low resolution Flash 360° object viewers with a Silverlight Deep Zoom one.

First, I wrote a prototype of the player to check the speed and it was quickly clear that it would be perfectly suitable for a 360° player. Next, we had to find a good subject. We tried cars but soon we realized that a nice lady would be even better to spin around :). Starting with a shooting of a colleague and uploading the player to Azure, we went to the Swiss television “SF” to see what they thought about it. SF is the producer of Miss Switzerland election and they quickly took the opportunity to implement a new fun experience in their portal.

Deal done and ready to go. A couple of weeks after we were on a flight to Madeira Island (Portugal), carrying more than 80 KGs of photo equipment and with the high hopes to achieve our goal.

The whole idea is to show the 12 candidates in super high resolution without any Photoshop correction and let you judge the Misses like you were standing 10 cm away from them.

The solution is composed by three parts:

  • A panorama showing all 12 girls together
  • A 360° view of each girl
  • A Microsoft Pivot collection with sortable and searchable information about the girls with a high resolution picture.

The big challenge consisted on how to get very high resolution pictures of the girls. For the panorama the only way was to take as many pictures as possible and stitch them together. We opted for a Canon 5D II, 21 Mpixels with a 400mm F4 lens on a Manfrotto geared head. We took 4 rows of 20 pictures each for a final output of 650 Mpixels.


Eventually, we used the latest Microsoft ICE stitching software that had a couple of killer features making the project easier to achieve: force a matrix (e.g. 4 x 20) when stitching and exporting the result in very large (> 4GB) Photoshop file for the final touchup. The challenge in this case was to make the reflection on the pool match the Misses, considering that the reflections were taken a few minutes after the Miss photo. In two cases we had to use the Photoshop CS5 Puppet-Wrap feature to move the feet in order to match the model.

clip_image004 + clip_image006 = clip_image008
The challenging reflection problem.

Exporting from Photoshop to Deep Zoom was done with the Microsoft Research HDMake 64 bit plug-in.

For the 360° we wanted to achieve a head shot and a full body shot on the same picture requiring a minimum 50 Mpixels resolution. Stitching would be problematic, risky and time consuming. We opted to use a medium format digital camera, the Hasselblad H3D 50 Mpixels camera. At a whopping 33’000 USD body only and needing a spare one too, wasn’t a cheap choice but definitely worth it.

clip_image010 clip_image012
Noemie - left the full frame; right 100% pixel crop

We wanted to create an almost surreal light and be sure that we could repeat the shooting if something went wrong. We opted to use one Elinchrom Ranger AS Speed with a Deep Octa as a main and one Elinchrom Quadra with the standard reflector as a rim light.
Once the setup was done the shooting was quite straight forward, each miss should rotate at 30° steps standing on the center of the check board. The camera’s resolution is so high that we even could up-res the photos from 50 to 100 Mpixels using Bicubic algorithm for higher zoom possibility. We had 2 hours to take 12 photos for all the 12 candidates. Boy, what a stress! Not processing though, since we had Deep Zoom Composer.

clip_image014 clip_image016
Hans Stuhrmann, photographing Kerstin

The third Miss project should be a pivot which is a collection of the candidate’s data combined with more pictures. These were done with the available light around or staged outdoor with strobes.


Shooting was done with the Hasselblad H3D, a Canon 1D IV and a 5D II. Lighting was with two head Elinchrom Ranger and a two head Elinchrom Quadra. We used the Excel Pivot Collection tool to create the pivot datasets.

Myself photographing Jennifer P.

The solution was built in Silverlight using Expression Blend 4 and Visual Studio 2010.

Some interesting figures:

  • 650 Mpixel picture for the panorama
  • Each Miss 360 is composed by 12 * 100 Mpixel photos, equals a 1.2 GPixel dataset
  • Final deep zoom solution is composed of 337’488 JPGs equals 4.8 GB
  • 53 hours of CPU computation to prepare the images

Equipment used:

  • 2 Hasselblad H3D 50 Mpixels
  • 1 Hasselblad 80 mm
  • 1 Hasselblad 210mm
  • 1 Hasselblad 35mm
  • 2 Canon 5D II
  • 1 Canon 1D Mk IV
  • 1 Canon 70-200 F2.8 IS II
  • 1 Canon 24-105 F4 IS
  • 1 Canon 16-35 F2.8 IS
  • 1 Canon 400mm F4 DO
  • 1 Elinchrom Ranger AS Speed
  • 2 Ranger Batteries
  • 2 Ranger A Head
  • 1 Elinchrom Ranger Quadra
  • 2 Rangers Quadra A Head
  • 2 Quadra Batteries
  • 1 100cm Elinchrom Deep Octa
  • 1 100m Softbox
  • 4 Lightstands
  • 2 Tripod
  • 1 Manfrotto 400 Gearhead
  • 2 Manfrotto 405 Gearhead
  • 2 Windows 7 64bit Notebooks
  • Microsoft Visual Studio 2010 Pro
  • Microsoft Expression Blend 4
  • Microsoft Expression Design 4
  • Microsoft Research ICE
  • Microsoft Research HDMake
  • Microsoft Deep Zoom Composer
  • Adobe Photoshop CS5 64 bit

The hotel room transformed into a warehouse.

The team

  • A creation of Ronnie Saurenmann, Microsoft Switzerland
  • Project acquisition: Stefano Mallè, Microsoft Switzerland
  • Web tracking, MSN integration: Aurel Hosennen, Sébastien Donzel, MSN Switzerland
  • Photography: Ronnie Saurenmann, Hans Stuhrmann (
  • Swiss Television SF: Bernhard Michel, Markus Krucker
  • A special thanks to: Luca Argentiero, Daniella Roth, Patrick Schuler

The obligatory fun photo: myself, Jasmin, Hans in a “James Bond” pose.


Ronnie Saurenmann

Comments (1)

  1. Miss Swiss says:

    Did you know that during her study Miss Switzerland was sharing an appartment with other flatmates thanks to the free webportal

Skip to main content