W3C home > Mailing lists > Public > w3c-wai-ig@w3.org > April to June 2015

Re: Accessibility of SVG - low vision issues?

From: Olaf Drümmer <olaflist@callassoftware.com>
Date: Tue, 26 May 2015 22:33:04 +0200
Cc: Olaf Drümmer <olaflist@callassoftware.com>, "chaals@yandex-team.ru Nevile" <chaals@yandex-team.ru>, Vedran Vucic <vedran.vucic@gmail.com>
Message-Id: <BBCC1197-A03C-477A-A929-99827F7E8B3A@callassoftware.com>
To: WAI Interest Group <w3c-wai-ig@w3.org>
A few thoughts… (not all of them specific to low vision):

- SVG is horribly limited when it comes to expressing semantics; as a a consequence it is usually just used as a vector painting format (i.e. exclusively focusing on visual presentation without any noteworthy content or semantic representation). This makes it extremely difficult to make SVG images [sic! if I may…] useful beyond creating a rendered presentation for visual consumption; for example: bar and pie charts are usually generated from tabular data - but it is very cumbersome to keep such data and its internal relationships around in an SVG encoded chart.

- the business charts in the collection listed here suffer from bad overall design; even I as a user without visual disabilities it annoys me to no end how difficult some of these charts are to consume. Despite the fact that there is usually ample space no use is made of such space. Some of the most annoying details are (and I can only guess they are so much more annoying to people with low vision):
- put the actual value next to the entity representing such value; e.g. for a bar chart, put the applicable number next to each end of the bar
- for pie charts, put the values for both axes next to the applicable slice
- also for pie charts, make each slice sufficiently distinct from its neighbours without relying on colour or a (in most cases) mild contrast only, instead / in addition  use a line around each slice, or a gap between slices, or both (probably half of the pie charts become partially meaningless once printed on a black and white laser printer)

- now, for any SVG that is not visually representing content that can also be represented as a list (like an org chart) or as tabular data (like most business diagrams/charts) but rather represents cyclical or complex internal relationships, I have yet to encounter an approach that can make the visual presentation as such decently accessible to a user with low vision; in such cases the only solutions from my point of view are 
- to semantically enrich the SVG data (which as far as I can tell is hardly possible as of today, at least not in an interoperable fashion)
- have such enrichment follow certain rules/paradigms such that
- suitable tools (browser? AT? ….?) can assist by means of 
	- providing additional context information (could be as easy as a tooltip telling me where I am or highlighting the 'current object' or the 'parent object' or 'the next object'; could be as complex as providing a structured text based [re]presentation of a possibly multiply nested structure at the current 'location')
	- offering navigational options (where do you want to move to next? One level up, to the neighbour to the right, ….); this is one of the  most under-utilised aspects, as I think spatial navigation can easily be learnt, is easily understood and can easily be memorised by many types of users; it would probably also allow for decent keyboard or gesture driven navigation

- last but not least: in all cases where diagrams are generated from well structured data, it is always preferable to also offer such data, whether as a table alongside with the chart/diagram, or as a semantic structure 'invisibly' encoded into the graphical representation, such that it can be retrieved and presented as deemed necessary by a given user, and as enabled by the tool at hand.

All in all I have to say that I believe as of today SVG is not up to the task of achieving much in this corner. The most needed extension of SVG would be some approach to make it possible/obvious how to encode semantics, both for nodes and for structure/relationships. Once such approach exists, authors/authoring tools and users/consuming tools could take advantage of it. Whether for users with low vision or whoever else.

Problems going back to bad functional design (one of the diseases most difficult to cure) will probably and unfortunately stay with us forever… 

And….:  don't waste much energy on interactive SVG (at least not from an accessibility perspective). Rather, make SVG semantically richer, so tools and user agents can establish rich interaction especially with more complex SVG as needed. 


On 26 May 2015, at 18:02, chaals@yandex-team.ru wrote:

> Thank you Vedran, and Wayne...
> Please look at the SVG images in each of
> https://www.w3.org/wiki/SVG_Accessibility/School_Graphs
> https://www.w3.org/wiki/SVG_Accessibility/Cool_Graphs
> https://www.w3.org/wiki/SVG_Accessibility/Directed_Graphs
> https://www.w3.org/wiki/SVG_Accessibility/Mapping_Graphs
> https://www.w3.org/wiki/SVG_Accessibility/Other_Graphs
> https://www.w3.org/wiki/SVG_Accessibility/Schematic_Graphs
> and see what issues there are for low vision users.
> One important task is writing them down -  in reply to this thread, or editing the wiki page directly if you have a login, or write to me and getting me to pass it on...
> Another useful thing to do is to try out SVG implementation in whatever browsers you use, and figure out what works and is helpful, and what doesn't work and causes problems. Please remember to specify what browser/OS/version you are using. I suspect we have a limited understanding of how magnification really works across the web platform, and getting some systematic sense of it would be good.
> For example, some browsers zoom text, some zoom pages, some reflow things and some may not, etc.
> Also, if you have examples of images that we are missing - currently we've got nothing interactive and there are a lot of interactive SVG things out there - which we can publish (i.e. are public domain or similar), we'd be grateful for them...
> cheers
> Chaals
> 26.05.2015, 08:53, "Vedran Vucic" <vedran.vucic@gmail.com>:
>> I am willing to help. Please let me know what I am supposed to do.
>> Thanks
>> Vedran
>> On Tue, May 26, 2015 at 1:30 AM, <chaals@yandex-team.ru> wrote:
>> Hi,
>> the SVG accessibility task force is trying to do a bunch of things. One of them is to collect somewhat representative SVG images found "in the wild", and figure out what issues they pose to users - which is a useful thing when we try to figure out how to solve those problems.
>> There is something of a collection at <https://www.w3.org/wiki/SVG_Accessibility#Images_for_playing_with_to_determine_use_cases_and_requirements> although there is a lot that we probably need to add.
>> Anyway, if people want to look over that, and provide comments, whether here, on the github repository, or to the SVG accessibility task force, that could be a helpful thing to do…
>> cheers
>> chaals
>> --
>> Charles McCathie Nevile - web standards - CTO Office, Yandex
>> chaals@yandex-team.ru - - - Find more at http://yandex.com
> --
> Charles McCathie Nevile - web standards - CTO Office, Yandex
> chaals@yandex-team.ru - - - Find more at http://yandex.com
Received on Tuesday, 26 May 2015 20:33:31 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 25 May 2017 01:54:15 UTC