-
Notifications
You must be signed in to change notification settings - Fork 0
Forks in the road
When developing an application there are many things you want to do but end up not doing. Because you don't have the time or don't have the resources or combinations thereof. On this page we'll keep a list of some of the more major such decisions.
Our evaluation of generous interfaces was made possible by the funding of a number of museums. Those museums have collections and we wanted to make a selection of content that would result in all of them having some of their content included. After some deliberation we decided that a good common content theme was fashion, accessories and history of costume.
This meant that a lot of content was just NOT included. We also decided to aim for a small collection of images in the low thousands, not 10s or 100s of thousands to enable manual data wrangling.
Based on the available and selected content, and lots of gut feeling, we were keen to explore a prototype aimed at users who were looking for visual inspiration for their own (fashion & clothes design) creations.
We initially wanted to include an interactive timeline visualisation in our prototype, but very early we realised we would not be able to combine keyword, colour and timeline exploration and do so within the timeplan of the evaluation. So we opted to focus on colour and keywords. The reason we did this was partially because we had earlier decided to use Google Vision to do the colour extraction and we could then "at the same time" extract keywords. Also, we estimated that a word cloud would be less time consuming to implement than an interactive timeline.
Later on we decided to contract a freelance creative technologist to create a separate/second prototype focusing on timeline exploration and visualisation.
Our evaluation is very focused on the very concept and principles of generous interfaces:
- Show first, don’t ask (provide rich overviews without search)
- Provide samples and clues (using collection content)
- Show relationships (between collection features)
- Provide rich primary content (deliver on the promise)
rather than testing whether the 10 heuristics of usability are 100% followed or not. We know the prototype has a lot of usability issues.
The prototype does not aim to meet the WCAG AA guidelines. Or even the WCAG A guidelines. If we would ever make a real product rather than a prototype we would certainly aim for WCAG AA (actually as government funded organisations we have to).
Also, the prototype doesn't work in Internet Explorer. Sorry, but not sorry.
- Should we allow the user to make very precise colour choices or let them choose from a a smaller palette of generalised colours? Initially we explored the first option and even spent a lot of time building an interactive colour picker. But after some user testing and deliberation we switched directions and adodpted CSS3 colours and a much simpler histogram style colour picker.
- Should we have a very image centric image wall/result grid with no metadata displayed at all? Or a grid that shows some selected metadata (keywords, colours)? We've actually built both and will test both, but our aim is to select just one of them.
- Should adding colours to explore be an AND between the colours or an OR? We tried both and have opted for AND on the principle that filtering should always narrow results, not widen them.
- We thought it was a good idea to let users A. Save and name their saved colour palettes, and B. Select from pre-created and named colour palettes like "Autumn", "Rococo", "80s neon", "Happiness", "Melancholia", "Fire", etc. Some of use still like the idea but we don't have the time to make a good implementation. One of the reasons we think this idea may be a good one is that our early research showed that users looking for inspirational images wish to do so in terms of moods, atmospheres, and feelings.