Colour Differentiation of Structure Levels...

[Mark_SAI]

I sometimes find it easier to understand a block diagram if I colour the background:

 

Which got me thinking... what if I had something like this:

It is not immediately obvious to me the levels of the structures... so perhaps I should colour the levels progressively darker:

Now it looks like you have multiple layers of tinted glass stacked up, and gives it an almost 3D look! I can now immediately tell which structure is innermost.

I think the enviroment should have the ability to specify colours of structure depth:

What do you think?

[AristosQueue (NI)]

The problem is wires.

Using color is always a problem in LabVIEW because many of our users are color blind. When we use color near wires, the problem gets really bad. We test our wires against a white background. It is the *only* background that the majority of users can consistently distinguish wire color. Even the usually distinct orange and blue wires used for numeric types appear identical to many eyes when placed against backgrounds other than white. 

You would end up having to have the diagram colors be different for each individual user instead of being part of the diagram itself. Setting that up could be tricky, and even then I think we'd make the defaults be all white all the time.

[AristosQueue (NI)]

PS: I noticed which block diagram you're using for your post. It's nice to see that someone is using my Maze demo. 🙂

[altenbach]

Any diagram background color other than white strongly interferes with wire visibility of at least one datatype.

(In rare cases where I use background fills, I use near white pastel colors.)

For this reason I don't like the suggestion. A properly designed diagram is clear without colors. 😉

[PJM_Labview]

I personally like to use light gray shade alternatively with white (as seen below).

I do think that it help readability (even for "properly designed diagram" :P).

PJM

[AristosQueue (NI)]

I agree with the selective use of color helping a diagram... but I think it is selective, and perhaps specific to the data types being used in that particular VI. I don't think any automated system will work.

[smercurio_fc]

What if the person looking at it is color-blind?

Never mind. Saw that it was already mentioned. 

Message Edited by smercurio_fc on 08-21-2009 05:41 PM

[altenbach]

> I personally like to use light gray shade alternatively with white (as seen below).

Agreed, there is nothing wrong with this. You use a very light shade. The idea suggests a gradient of increasingly dark shades, which is guaranteed to interfere with objects and wires.

Case in point is the shown example:

Look at the boolean wires at the indicated locations (A..E). In some places it is almost impossible to see the wire or tell it's color, depending on the background.

.

Message Edited by altenbach on 08-21-2009 03:59 PM

[Mark_SAI]

> Look at the boolean wires at the indicated locations (A..E). In some places it is almost impossible to see the wire or tell it's color, depending on the background.

This suggestion is definately one for people with good eyesight- I can see those boolean wires perfectly fine! 🙂

This feature was only ever going to be an optional extra idea, that you access when you untick "Use Default Colors" and the default colors for all the structure levels would be white.

[altenbach]

It's not just eyesight. Single-pixel lines on a LCD display necessaily cause certain problems because the R,G, and B fields differ in position. (see also e.g. cleartype). If your eyesight is too good (e.g. when you are looking at the LCD with a microscope), things look actually worse!

Things also vary significantly with contrast and gamma settings, LCD, vs. CRT, etc. (My (corrected) eyesight is perfect but my laptop is 1400 pixels at only 15 inch diagonal) 

Then we have the optical illusion that a perceived shade depends on its surrounding. The boolean wire at B looks significantly lighter and less green at e.g. B. Similarly, the marked grey squares in the following picture are identical.

I also believe that structure level is not important information that needs to be encoded somehow. This informations is already fully given by the 2D arrangement, unless your entire diagram is many screens big and you only see a small fraction. There are probably better color schemes to make the diagram more logical by coloring based on function, (e.g. UI handling in light green, exception handling in light red, pure math in light yellow, DAQ and IO in light blue, etc. or whatever ;)) Filling an area with a desired color is trivially simple. I don't think we need a built-in hand-holding. 😄

-

Message Edited by altenbach on 08-24-2009 08:41 AM

[elset191]

Altenbach's image got me thinking of another illusion that I had seen recently that is too cool not to post.

http://blogs.discovermagazine.com/badastronomy/2009/06/24/the-blue-and-the-green/

The "blue" and the "green" are actually the same color.