Photogrammetry – Managing the Data

How much?

Photogrammetry produces a lot of data. Here’s a walk-through of the kind of volume needed to create the High Seas Fleet pinnace found last month in Scapa Flow:-

Volume of data by event

From beginning to end we have:-

• Dive – 598 RAW files = 15.32Gb
• Initial processing – 598 derived JPEG images = 5.76Gb
• Build the model – Agisoft Photoscan project files = 15.61Gb
• Exported TIFF Orthophoto = 1.07Gb

All of this is for a relatively small target.

The Value of Managing Data

Years as working as a professional photographer developed a really slick and quick workflow to get the RAW files managed and the JPEGs processed really quickly. Embedding meta data into the images seems tedious at the time, but when it comes to finding things quickly in future precious time is saved.

For example, there are now 130,000 RAW files under management and finding ‘the one’ takes seconds. I would shudder to think how the unmanaged cope.

The workflow for model management is now honed, with long term backup and storage/retrieval working well.

But the stumbling block is the orthophoto.

Limits of LightRoom

Most photographers will be unaware of the physical limit Lightroom imposes, but any file larger than 64,000 pixels x 64,000 pixels in size or 512 gigapixels in total, triggers this error:

Which is easy to trigger when the native size geo referenced orthophoto TIFF file is recording a resolution of 1.48mm per pixel.

Yes, 1.48mm per pixel. In the TIFF image you can count the spines on the sea urchins…if you really want to.

And for bigger sites it is easy to export a TIFF that neither Lightroom or Photoshop can actually open, let alone view.

What to do?

The value of image management is priceless so saving the orthophotos without some management/search/retrieval method is not an option. There are a few things we can do:

• Generate a lower resolution orthophoto. Photoscan Pro allows us to accept the default resolution, or to set a value that will import into Lightroom or be opened by Photoshop.
• Generate the orthophoto at the higher resolution, and then export a derived and lower resolution TIFF image and manage this version in Lightroom.
• Export the higher resolution image and (if Photoshop can cope) resize the image and import into Lightroom.

Summary

Generally I opt for the first choice. With a management system in place I’m confident I can  go back into Photoscan Pro and recreate a higher resolution orthophoto when the software catches up with what we can create and view.

So whilst its a problem, I’m not complaining. The orthophotos derived from the models provide a level of detail beyond the model and really give a divers-eye view of the subject with all the water drained away, such as this one of the UB 116: