Here are three grids of external saturation vs internal saturation.

In the RED grid, I see only a faint contrast throughout and nothing in the first three columns.

In the GREEN grid, I see everything but the first column and good contrast.

In the BLUE grid, the central square is yellow, which isn't all that easy to see against whitish background, but it is definitely there on all but the first column.

See attachment.

Attachments:

best cb tester w...nb

Hi Hedwig,

I've attached the notebook, but there is something wrong with many of my notebooks in Version 14.1. It has something to do with the Dynamic function. When I evaluated the notebook, the sliders in the app were not staying put. Also, beware of the first line; set it to 1==0 or delete it.

Eric

Attachments:

best cb tester w...nb

Hi Hedwig,

Which version of Mathematica are you using? I recently bought a new computer to run Windows 11 and also installed version 14.1 of Mathematica. Could this computer be at fault?

I can't run the colorblind test, and it's been almost a decade, so I can't give you much help.

I remember that a datum can be removed by clicking on it in the graph. This is useful because outliers do occur, especially with inexperienced users.

What version of Mathematica are you using?

Eric

Hi Doug,

(Weird, the previous post wasn't on line when I went back to edit it, so this post repeats some of the points.)

Now that the computer colour blind measuring test is pretty much finished, we can consider your original observation: that the medium of the test doesn't matter. It could be a laser printer, ink-jet, or a monitor.

The computer test could be implemented on paper as a grid of surrounding saturation vs interior saturation for each colour. the line demarcating seen/unseen would be the curve of the computer test

I'd forgotten about colour blind testing being used to exclude people from certain jobs. It would have been a terrible thing if I had not been allowed to work at the university because of my colour blindness, considering what I have accomplished.

The alternative use of the computer test would be to:

1) create "sunglasses " that normalize the colours

2) create illuminators that normalize the illumination.

3) create a camera with inverse response in the non-defective colours so that the image would be normalized.. This would be the best way, because the response is sensitive to saturation, which a filter or illuminator would not be. We have the responses as equations, so these could be applied directly to RGB values.

In fact, if the corrections were applied to the RandomImage[] display, it should be possible to get three coincident straight lines. Or would it? The green and blue values would be distorted to make them follow the red response. Everything would be multiplied by some coefficient to bring the maximum to 1.0. That will be the next thing to do.

The sunglasses gave a good response on the our test, and I was declared "not colour blind" in the colour arrangement test. But they made no difference in the Ishihara. That will be the ultimate condition for colour correction--see the hidden figures in the Ishihara.

Attached is the test with the controls below the image. This makes it more useful for touch screens.

Eric

Attachments:

Hi Patrik,

Amazing! I also went to StackExchange and found some code that would have required a lot of finagling to accomplish anything.

This might be the final version. I did fix up the Initialize to set the vals to zero instead of empty.

When I said a "small editing menu," I was just leaving room for ideas that haven't yet occurred to me. Nothing comes to mind, though.

Thank you for your help. I'm curious; how did you learn Mathematica?

The slightly modified code is attached.

Eric

Attachments:

Hi Patrik,

Yup, it's really good code!

The constant scintillation wasn't really what I wanted, once I got used to it. The problem before was that you could only really see the inner square at low contrast when the mouse was moving. Here is some code with the Refresh[] image only while the mouse button is down. (In the attachment.)

I removed the inner braces and the 1 in {{"MouseUp", 1} :> (AppendTo[valsr,... What was your intention?

I've been trying all morning to find a way to edit the graphs with a mouse. Especially with the new automatic collection of data at a MouseUp, it's possible to get some erroneous data points. What I'd like to do is have a little editing menu (delete a graph point for now). The Tooltip knows when you are at a point in a ListPlot, but it doesn't return the point. Do you know of a way of accessing the graph like the Tooltip does?

Here is the effect of the scintillation while the mouse button is held down:

You can see the effect by the much smaller scatter in the red graph.

Eric

Attachments:

Hi Patrik and Doug,

Doug, this isn't a test in the sense of Ishihara. It's more like a visual field test.

Patrik, I love your code. Adding constant scintillation was quite easy using Refresh[]. I put the ranges of all sliders back to {0,1} but there isn't really enough room to move them accurately unless you use the alt key trick. Notebook attached.

Eric

Attachments:

Patrik and Eric,

In reply to your question (Eric), I am really tense during the test because I am trying to be absolutely consistent in "when I see it". If anything, I push it towards "I definitely see it".

If this technique made it to ophthalmology, they might want a verifiable answer. Than would mean we would have to randomize the shape of what we are displaying. In that regard, I would follow Ishihara and use two numerals. Approaching always from the invisible side, when the subject calls out the correct numerals, you have their value.

You guys are so far ahead of me in coding, I am sorry that I can be of no help. It's also going to take some time for the biophysics of saturation to sink in. I also wanted to make it clear that asking about a CDF did not mean I wanted to close off the development.

I am very happy that you made so much progress.

Doug

Wow! What an awesome thing you guys got going on here!

It isn't necessary to click on the Get colour values button. All that is required is to sense the mouse up when the slider is released. Suggestions?

I think that an event handler would be the way to go. They are associated with their respective expression so the "MouseUp" event will only trigger when the mouse is over that expression. So putting one of these on each of the sliders does the trick. However, I don't know if you can directly manipulate the Manipulate sliders. An alternative is to skip the Manipulate function and implement your own sliders. I did that in the attached Notebook.

The code became:

DynamicModule[{saturation = 0.5, r = 1.0, g = 1.0, b = 1.0},
 Column[{
   Labeled[Slider[Dynamic[saturation], {0, 1}], "S: ", Left],
   Labeled[
    EventHandler[
     Slider[Dynamic[r], {0.75, 
       1}], {{"MouseUp", 
        1} :> (AppendTo[valsr, {1 - saturation, r (1 - saturation)}]; 
        r = 1.0; g = 1.0; b = 1.0; saturation = RandomReal[];)}, 
     PassEventsDown -> True], "R: ", Left],
   Labeled[
    EventHandler[
     Slider[Dynamic[g], {0.75, 
       1}], {{"MouseUp", 
        1} :> (AppendTo[valsg, {1 - saturation, g (1 - saturation)}]; 
        r = 1.0; g = 1.0; b = 1.0; saturation = RandomReal[];)}, 
     PassEventsDown -> True], "G: ", Left],
   Labeled[
    EventHandler[
     Slider[Dynamic[b], {0.75, 
       1}], {{"MouseUp", 
        1} :> (AppendTo[valsb, {1 - saturation, b (1 - saturation)}]; 
        r = 1.0; g = 1.0; b = 1.0; saturation = RandomReal[];)}, 
     PassEventsDown -> True], "B: ", Left],
   Row[{Column[{
       Button["Initialise", valsr = {};
        valsg = {}; valsb = {};],
       Button["Draw graphs", drawGraphs[valsr, valsg, valsb]]}],
     Dynamic[(image = 
        ImageData[
         RandomImage[{saturation, 1.0}, {100, 100}, 
          ColorSpace -> "RGB"]]; 
       ImageCompose[Image[image, ImageSize -> 200], 
        Image[Partition[
          Transpose[{r, g, b}*
            Transpose[Flatten[image[[25 ;; 75, 25 ;; 75, All]], 1]]], 
          50], ImageSize -> 200]])]}, Spacer[2]]}]]

Is there a reason why the red slider in your code went from 0 to 1 and all the others went from 0.75 to 1?

Best regards, Patrik

Attachments:

Eric,

Here are two more trials. I have a hardcover copy of Ishihara, and I pass.

Doug

Hello everyone,

Here is a functional app to measure colour blindness.

Instructions

1) Click on Initialise to set the data collection arrays to zero.

2) A random saturation level for the image will be selected. Move one of the r, g, or b sliders until the central square appears. Keep adjusting the slider until the square is just visible. Click on the corresponding Get colour values button.

3) A new random saturation will appear. Continue until you have enough data points to generate smooth graphs.

4) Click on Draw graphs. A new notebook will appear with the graphs, data arrays, and the approximating equation.

5) Examine the graphs for regions that need more data. Go back to the tester and manually set the Saturation slider to the value needed. This can be done by positioning the slider visually so that it is matches the horizontal axis of the region needing data.

Discussion

This app is an implementation of a technique due to Douglas Youvan.

There is a significant improvement that eludes me, so I'm asking for help. It isn't necessary to click on the Get colour values button. All that is required is to sense the mouse up when the slider is released. Suggestions? I know that EventHandler[] is probably the way to go, but how do you relate the MouseUp to a particular slider?

It would be interesting to get some graphs from both normal and colour blind people. There are other types of colour blindness than red/green. (I always thought that my green vision must be defective, too. Not so.)

And suggestions for improvements and other features would be welcome.

A notebook is attached.

Eric

Attachments:

Hi Doug,

Here is the result with my amber-tinted sunglasses:

And that's why the fall colours explode with them on!

Eric

Hey Eric,

Before I comment on your recent work, let me upload another program here.

This started at: http://et-al.com/pdf/document3.pdf . Look at Figure 5, panel F and the math that goes with it.

I re-wrote that code in Examples 3 and 4 in my Pseudocolor e-book in 2006: http://youvan.com/Examples/Example_4._Counterfeit_Detection_Based_on_Ink_Color.htm . Here, I've attached "otho 1" which is running in version 10.3. The main thing to note in the code is that on images with homogenous backgrounds the matrix will get identical pixel values and blow up from singularity. That's why you see me adding small random numbers.

I've now uploaded "ortho 2 click", and it is much easier to use. You get to pick three pixels (colors) and then orthonormalize the entire image.

Doug

Attachments:

Hi Doug,

Have you come up with anything? This morning I discovered RandomImage[], and in true MMA style, it's very fast. Instead of a cross in the square, I used a square inside a square. I also found an interesting thing about the saturation (correct term?) of the background image. When it is at full saturation, I can only barely see the interior square when r is at zero.

Here's the code:

        Manipulate[
           (
              imag=ImageData[RandomImage[{saturation,1.0},{100,100},ColorSpace->"RGB"]];
              ImageCompose[Image[imag,ImageSize->200],
              Image[Partition[Transpose[{r,g,b}*Transpose[Flatten[imag[[25;;75,25;;75,All]],1]]],51],ImageSize->200]]
            ),
            {{saturation,0.5},1,0},{{r,1.0},0.0,1.0},{{g,1.0},0.0,1.0},{{b,1.0},0.0,1.0}
        ]

The reason I can see the green and blue so well is that I have lens implants in both eyes. The ophthalmologist showed me a yellow glass lens that he keeps in his office to show his patients how discoloured the lens becomes with age. It took a few months to get over the extreme blueness of my new vision. But I also remembered how blue snow used to look when I was a child.

Eric

Attachments:

Hi Douglas,

The fall colours have finished in Nova Scotia. Viewed directly, there isn't anything very spectacular, but when I put on my amber-tinted sunglasses, the reds become vibrant. So the sunglasses seem to give me normal colour vision.

Not only could MMA test for colourblindness, it could actually measure it and compensate. I've always wondered what normal people see.

Eric