Im going to show how to digitalize a spectral power distrubution (SPD) graph in order to analyze it with the tool.
Obviously, you need to have a SPD graph to analyze.Im going to use MS paint to do it. To edit a graph with it, righ click the file and select the option "Edit". It open directly MS paint with the graph in it:
You need to do some steps to digitalize it easily:
1-From the scale in the X axis of the graph, wich is given in nanometers (nm), calculate its width in nm (end of the axis minus start of the axis). Almost all SPDs cover either 400-700nm or 380-780nm (preferable), thus they have a width of 300 or 400nm.
The example has a with of 785-380=405nm. But ill ignore the last 5nm waveband, as the analyzer tool only has a range up to 780nm.
2-Remaining operations are performed better using a larger graph size and the squared mode. To do it, resize the image to 400% by clicking on the top bar on "view", select "zoom" and click on large size.
You can do it directly by pushing Ctrl+Av Pg. By pushing Ctrl+m after it, you put on the squared mode (or go again to "view", "zoom" and "show squares").
3-Adjust the horizontal size (x axis) of the pic in the way it has same pixels than nm. Thus each pixel will correspond to a one nm waveband. To do it, first you need to know the width in pixels of the graph:
3a-Put the pointer on the left edde of the x axis and take note of the pixel number, wich is shown on the bottom right. It shows two numbers splitted by a ",". The first correspond to the x axis coordinate and the second is the y axis coordinate. Now we are interested on the first number (x axis).
(The screenshots dont show the pointer. Its on the left bottom corner of the graph. As graph's lines are 3 pixels thick, i put the pointed on the middle pixel)
Do the same putting the pointer on the last pixel of the x axis. The second minus the first gives you the graph width on pixels.
So in the example, it has a width of 848-60=788 pixels.
3b-Now divide the width on nm by the width on pixels. Multiply it by 100. Round the entire part of the figure from the closest decimal (sum 1 if the first decimal is 5 or higher).
Example: 400nm / 788pixels=0.507*100=50.7 wich is rounded to 51.
Click on "Image" on the top bar, "expand or reduce" (or directly, Ctrl+w). A dialog box appears. Enter the figure you get on the last operation on the first (upper) box (expanding horizontally). It dont mind if the figure is lower than 100. It works the same way.
Now you should have the same width on pixels than in nm. But as the expand tool not accept decimals, sometimes there is one or two pixels of difference. So check the pixel width of the graph again (step 3a). If its not the same than nm width, repeat step 3b (likely you'll get 99 or 101). If it still dont works, reduce size to any figure, as 57%, and start again. But it usually works at the first or at least, the second try.
Matching accurately pixels and nm is important on 1nm resolution graphs. In the example, its a 5nm wavebands resolution, so it dont mind. Each 5nm width column is clear, and you can check there is 8 5nm columns in each 40nm division of the graph.
4-Optional. Highlight a single pixel for each nm on the SPD curve. This step does easier entering data on the tool and help reducing errors associated to hand digitalizing a graph. Just select a color with the stronger contrast with the color of the graph (by clicking it on the bottom left of Paint) and go clicking on a single pixel that represent better the emission on that nm.
Very often,lines are some pixels thick, so you need to choose if highlighting the upper, middle or lower pixel. As far as you use always same criteria, its valid. Differences due different criteria are under the error margin associated to digitalizing a graph. On other ocasions, emission raises/drops sharply on a few nm, resulting on a vertical line with several pixels on the same nm waveband. In this cases, seek for the darker pixel. If there is some, try to see whats the shape of the line to select the most accurate or select a criteria, as the half point of darker pixels and use same criteria in all cases.
While this step is advisable on SPDs using 1nm wavebands, those using longer waveband of 5-10nm, that appears as columns often dont require doing it, as the top of the columns is clear.
(continue)