Elgin Cathedral QTVR

NT_2_final

Project: Elgin Cathedral QTVR
Phase(s): Data capture and processing

Status: Closed

Time: October 2015
Location: Elgin Cathedral, Elgin
Equipment: Nikon D800, Nikon D810
Software: Photomatix, PT GUI, Photoshop

 

survey photos.png

Documenting Elgin Cathedral using laser scanning and 360° photography.

The very first project I was involved in as a member of the Digital Documentation Team -in both fieldwork and processing stages- was the production of QTVR panoramas of eight rooms in the two towers of Elgin Cathedral. The QTVRs would be valuable as records of the existing condition of these interiors, before the installation of a new interpretation suite.

The first phase of the project required a visit to Elgin Cathedral with two other members of the DigiDoc team. The fieldwork was carried out over two days and including 360 photography of the rooms in the two towers as well as laser scanning parts of the Gordon Aisle.

The standard process for the production of QTVRs involves taking a series of photographs from the exact same location and rotating so as to cover the entire panorama. To ensure that all photographs are taken from the exact same spot, the camera is mounted on a standard surveying tripod using an accessory called a “nodal ninja”. Photographs are then taken at specific angles to ensure sufficient overlap in creating the panorama.

Inside the towers, two very narrow spiral staircases lead to generous-sized rooms (one per level). The lighting conditions inside the spaces were diverse, including natural light coming from the windows (some glazed, some louvred) and strong spotlights. The effect of different light sources on the surfaces of exposed masonry walls and vaults created a mixture of very dark and very bright areas as well as everything in between. Producing a 360 panorama of one room by merging several photographs of the same (or very similar) light settings into one image, required the use of HDR photography.

HDR images consist of a number of photographs taken from exactly the same position but with varying exposure settings (brackets). Different exposure values (EV) are suitable for lighter/darker areas; by capturing the same scene with different EV settings we made sure that the entire range of very-dark-to-very-light surfaces had been documented clearly in one of the photographs. In this project, an HDR software package called Photomatix was used to merge the bracketed photographs into even-toned HDR images.

HDR example
Example of HDR image. The three photographs on the left were taken in the third floor of the South tower using a nodal ninja mounted on tripod. The photographs were taken with different exposure values: +1, +2, +3. Using Photomatix, the three original photographs were merged to create the HDR image shown on the right.

Since the HDR images would be used to create 360 panoramas, it was important to ensure that process of creating the HDR files would be repeated using the exact same settings for every image in each panorama. Using even slightly different settings for each image in a panorama (room) would result in differences in brightness /contrast / colour / tone between images, i.e. a visually incoherent panorama. To ensure an even result across multiple images, a special preset was created for each different set of photographs (each room) and applied to all images in the set.

Photomatix settings
Screenshot showing the editing process in Photomatix. The three original photographs have been merged into an HDR image (preview in the middle). The method used and the specific settings are shown in the window to its left. These were saved as a preset and applied to all the images in that set.

 

PTGUI was the software used to create 360 panoramas from the HDR images exported from Photomatix. Although PTGUI technically offers the option to process bracketed images into an HDR panorama, in practice this workflow results in bad alignment between the images, which are very difficult to fix. Both workflows (Photomatix->HDR to PTGUI and PTGUI->HDR Panorama) were tested in the beginning of the project and the decision came in favour of the former, based on HDR image quality, alignment results, processing times, and ease of use.

PrScr PTGUI HDR problems Detail
Screenshot from the Detail Viewer in PTGUI, showing the alignment problems (marked in red) when attempting to create HDR panoramas from the original images in PTGUI. Fixing these issues proved very time-consuming and the results were less than satisfactory. For these reasons, this particular workflow was abandoned.

Producing panoramas in PTGUI is a relatively straightforward process. The main objective is to get the best alignment of individual images possible. This can be controlled through the use of control points (CP) between pairs of images, which can be added automatically or manually and removed again manually or by using an algorithm to spot the outliers in the group. The optimisation algorithm gives an indication as to the quality of the alignment, but my experience was that it cannot be trusted completely. Visual control of the alignment is absolutely critical! This meant a repeated process of checking the panorama for misaligment, jarred edges, discolourations etc. When finished editing, the panorama was exported as a simple image file (JPEG) or as a 360 QTVR, which can be viewed with the Quick Time player.

SWT_G
360 panorama of the ground floor room, South Tower.

The final step was to correct the flaw in the floor of the panorama, which resulted from the lack of photographic coverage of the area under the tripod, known as the “cone of denial”. PTGUI does not have a built-in feature for allowing this, which meant that a workaround had to be devised.

The idea was to export cube faces based on the stitched panorama from PTGUI, edit the floor image in Photoshop to correct the cone of denial and then re-import the 6 cube faces (with the corrected floor face) into PTGUI to produce a new (and final) 360 spherical (equirectangular) panorama based on the cube faces.

Web
Diagram showing the cubes faces created from a 360 panorama of the room on the ground floor of the South Tower, Elgin Cathedral.

Any number of options are available for correcting the cone of denial with this process: a simple black circle can be applied to the area or a logo or even text giving information about the space. For this project, the final option was considered more appropriate, particularly since the rooms looked very much alike, so including location info was considered helpful for interpretation.

fixed cone
[Comparison of cube face (floor) from the ground floor room South Tower, before and after fixing the cone of denial in image editing software (Photoshop).

The QTVRs produced for Elgin Cathedral document the condition of the interiors of the two towers prior to the installation of the interpretation suite. More than that, the interpretation team plans to display the QTVRs in an interactive platform within the new suite, showing visitors the original layout of the rooms before any alterations.

NT_2_final
Stereographic (“little planet”) panorama of the room on the second floor, North Tower.

 

Advertisements

One thought on “Elgin Cathedral QTVR

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s