The complexity of visual elements in digital connections.
Shelley Bernstein

It’s great to see a site with experimental options for exploring the collection. We’ve been working with a small NZ gallery on an online collection site and have been encountering some of the same challenges.

We’ve added colour, image orientation and subject tags (Google Vision API) as computer generated options for exploring the collection. Visitors can add multiple filters, including more than one colour, so we’ve had the same difficulty where adding many results filters makes it harder to spot why particular works are in the results.

We sort the best matches to the top by default, but the colour matches aren’t always perfect. Adding several colour filters at the same time can make the matches quite tenuous. At this stage we’re leaving the multiple colour selection option in as there are some good results with two (and sometimes three) colours.

Our site is also just at the soft launch stage. Here’s an example results page:

The colour and image orientation filters are at the bottom left in the desktop view. The Google Vision image labels are shown just on the detail page for each work, along with some natural language text we generate based on the object type, production place and production date.

With the Barnes site, are you snapping to a palette of just 20 colours, or is the 20 colour palette just the interface for filtering the results? We’re using a 139 CSS colour palette in the filtering, but this does result in less matches when filtering. We’ll be writing a post later this month on the changes and challenges in our project.

Like what you read? Give Paul Rowe a round of applause.

From a quick cheer to a standing ovation, clap to show how much you enjoyed this story.