Universal Viewer Discussion Documents

uv-discuss.md

Update, March 2019

Many of the discussions below have been synthesised into this slide deck: https://docs.google.com/presentation/d/1zAmXDlfvuwlQtp9uVgFLS7RVcfNqTcyEJybnNdQ2TgY

The links below are more comprehensive and are kept here for posterity. The other document worth considering is this:

Universal Viewer design principles
(in progress) Inspired by discussions at UVCON; an attempt to set out some guiding principles for the UV; comments sought!

One or two of the issues mentioned in this doc (arising from UVCON) have been addressed in the UV.

---

2018 Discussion, mainly for Washington

Stanford UX proposals
Google-doc discussion of Stanford's UX proposals, prepared for UVCON.

UV Search, revised design for comments
Updated search UI proposal from Stanford, after UVCON

• intended to be extension of current design language
• concerns about search in footer - not the place for it?
• could put it in a tab - but that ties it to the content panel. Search button could expose input field on click.
• Concerns about mobile version

UV Search, Adobe XD in response to above

• Experiments with white and black backgrounds
• concern over sliding panel, dependency on content panel

UV Search, 'Fully panel-ized'

• Frees search from dependency on content panel, but problem: if there is a content panel you have to close search to get to its CTA
• Suggestion - vertically stacked icons - like the Adobe XD UI here (look at XD, not the UV)
• but: "It means you can't have Metadata and Contents open at the same time, though, and I think that came out as a requirement in London"
• How crucial is multiple panels? And does this really prevent that? And what about mobile? Relationship between overlays (dialogs) and panels (shift over)

Look at conflicts between panel and popups - NLW Map And with a book

• user needs to be able to interact with the image/object as well as the results list, which made a panel seem like a necessity. But I think it's possible to allow a popup to stay present without being modal
• How does this relate to the textual content of an object?
• still chevrons, which don't test well

We need to discuss the relationship between "Search" and another as yet non-existent component of the UI, which is the non-primary content of the object. This may be transcriptions, comments, other annos as per #4. And consider the BL's AV textual content requirements.

Radical suggestion - is there really a "Search panel" at all? Or is search instead an important piece of interaction within a "Content" panel (NB - not "Contents" which is TOC)

Universal Viewer design principles
(in progress) Inspired by discussions at UVCON; an attempt to set out some guiding principles for the UV; comments sought!

---

Conversations in Slack followed... Locations of buttons, grouping of functionality in tabs/panels/stacked icons; where Search lives

The Farringdon Interpretation
How the UV deals with structural navigation in time-based media (a very detailed dive into one design principle)

Funding the Universal Viewer
There's a big chunk of work to do. How do we pay for it?

---

Related documents (not UV-specific, but of UV interest)

IIIF and Born Digital

IIIF and 3D

Comment on above from Jennifer Vine, May 21st:

This feels like we're back where we started, though, @tomcrane - with search enclosed inside another feature that may or may not always be exposed. If the textual content panel is fixed open, as in the screenshot, then it makes complete sense to search within that panel. But if the textual content is not exposed, where does search live, and how is it visible to the user? What happens in the NLW map use case? Or am I misunderstanding your suggestion?

Response from Ed:
I like the 'AdobeXD solution' - when the panels are collapsed, you get mutually exclusive vertically stacked icons. So if the annotations panel in Tom's image above were closed, you'd see a generic search icon. Clicking on it would expand the panel and focus on the search input field.
I mentioned this approach at the UV half-day on Monday and there were no strong objections. Feedback has been entirely positive.

In case (when) it disappears from Slack, I'm adding this image here:

Additional links for discussion:

Prototypes

• Text content of image-based content: https://5bd334fbb13fb141d1fca696--canvas-panel.netlify.com/developer-stories/viewer1.html
• Text content of time-based media: https://openhypervideo.github.io/iiif-interactive-transcript/

And this is the "Adobe XD UI" style mentioned above.

It's worth noting that this is the approach taken by the emerging Mirador 3 designs.
https://groups.google.com/forum/?hl=en#!topic/iiif-discuss/F3TPeKmuj4U

An assumption, to be challenged:

• The UV offers a consistent, comfortable, familiar UI across different types of media.
• If I've used the UV to view a manuscript from a BL catalogue search result, I'll feel at home if I encounter the UV when looking at the National Radio Archive

If this consistency is not a requirement, then the NRA doesn't need to use the UV. But I think it is; reduction of viewers in the BL estate is not just a code/configuration management issue, it's a UX issue too (again, this assumption may be challenged).

Problems to solve:

• Sparkline for results doesn't test well and runs out of room
  • Is mistaken for interactive navigational element (drag to change position in object)
• Search is of textual content because text terms are what we feed it
• ...but textual content is (or is part of):
  • plain text with arbitrary result precision (dynamically returned annos, e.g., Wellcome)
  • fragments (lines or paras of text, or temporal fragments of broadcasts)
  • complex annotations from crowdsourcing or machine learning processes (NLW tithes, libcrowds?)
• We want to view textual content, or more broadly, annotation content
• and we want to search the object, which indirectly means searching that annotation content
• If "view content" is a panel, how does it relate to search? As per jvine, a Search call-to-action shouldn't be hidden inside a content panel that may not be revealed.
• Possible solution is "Adobe XD style" where [0] is a CTA to both view and search content (and various flavours of search may be available along with the content, including none at all).
  • you have to be looking at (have revealed) textual content to search it.

--
[0] = magnifying glass symbol.

• And finally, what is our process for deciding all this within the UV community? What's the approach to product management?