Figure thumbnail as visual clue

[hlesaint]

As we humans are particularly good in identifying and differentiating visual clues, the similarity of the extract's visual appearance requires reading the sentence, to (1) identifying the method, kind of result, or species, and (2) re-identifying extracts you've already seen.

Having a thumbnail of the figure on the extract would help to identify and differentiate extracts immediately. In this case the option "figure" could be removed, and the thumbnail could be displayed at the very right (behind author).

Do you think this is computationally demanding and would slow down the search?

If this is an option, we could think about a way to avoid forcing a potentially large or complex figure to be ridiculously small. For instance, in the extract creation and modify extract page, a semitransparent form could overlay the figure, so you can choose the actual image section you want to show as a thumbnail (like e.g., for your profile pic in Facebook). The shape could be a square with round edges to match the overall style.

What do you think?

[hlesaint]

Of course, mouse over the thumbnail would expand the figure.

[hlesaint]

The longer I think about this, the more I'm convinced that this is really important. I fully underestimated the power and opportunity of visual clues to facilitate and speed up the usage.

[tgetgood]

Making a thumbnail from the figure is pretty straight forward. The UI to choose the thumbnail from the full figure is less straight forward.

If you're just looking for a visual cue, then shrinking the image might be enough in and of itself.

It would probably help to move the thumbnail to the left or right and put it next to the text. That would let us make it a little bigger.

As it is right now, nothing stops you from adding multiple figures to an extract. Should I change that to enforce just one?

[hlesaint]

Great, looking forward to this. I also thought to move the thumbnail to the right to slightly enlarge it.

I think one figure suffices. Enforcing only one figure may cause you to break too complex extracts down into the real key findings. That's good way to go.

[hlesaint]

Some options for a possible implementation:

[hlesaint]

[hlesaint]

I think I'd go for option 1 or 3. The figure dimensions are another clue for fast identification / differentiation.

[hlesaint]

Option 3. The reason is clear: This is the data, followed by its description. The data is what we want to see first.

It is very interesting how this thumbnail changes my feeling to our entire approach. Displaying the thumbnail is crucial. Without it I almost feel blinded to the actual content.

[hlesaint]

On a glimpse you see so much detail that you could not write in a few sentences: type of data, type of method, species, etc.

[hlesaint]

And you directly know where to hover for inspecting the data in more detail, it really attracts the cursor.

[tgetgood]

3 is probably best. One more option would be to deform the images into squares for the thumbnails instead of cutting them off, but I think we're better off just shrinking to the bigger of width or height.

[hlesaint]

Nice! Thanks for the quick fix. It really makes a huge difference.

What do you think about simply shrinking the figure. For tall narrow figures details may get lost, but the full image might be better to memorize /recognize by its individual shape (like a symbol), and differs more from others.

Therefore, I think not filling the space available can be considered an advantage, and it looks fine in the mock demo above.

In this case we could center position the figure with white spaces left/right. Maybe we should go for center position in general (also top/bottom).

[hlesaint]

[tgetgood]

That would be better, I agree. I couldn't figure out how to accomplish it, so in the interest of moving on I just deployed what I could get working.

We should also vertically centre short, wide figures.

[tgetgood]

I've put in place another interim fix that I think gets the styling right. It doesn't respond properly to zooming in and out, but it's not very noticable most of the time.

What I really want is for the size of the image to depend on the size of the extract summary box. This way extracts with tall, narrow figures and near the maximum amount of text will show more of the figure an look balanced. This is surprisingly difficult to get right.