How do you set a new default colormap for Matlab?

I want the default colormap for all of my projects to be jet? Can anyone tell me how I set this to be the default?

[Stephen23]

Why jet should be avoided:

https://blogs.mathworks.com/steve/2014/10/13/a-new-colormap-for-matlab-part-1-introduction/

https://blogs.mathworks.com/steve/2014/10/20/a-new-colormap-for-matlab-part-2-troubles-with-rainbows/

https://blogs.mathworks.com/steve/2014/11/12/a-new-colormap-for-matlab-part-3-some-reactions/

[Huzeyfe Yilmaz]

This is not a good statement. Although parula may have some advantages with some data sets, it also has disadvantages with others. As a physicist, I think it's just not a good idea to use a limited visible spectrum. Others (including those who make decisions on default colormaps in Matlab) may think otherwise, but it doesn't mean that jet should be avoided.

[Stephen23]

"This is not a good statement."

Actually it is a good statement: jet introduces artifacts into data that do not exist, does not represent how data are distributed, and is misleading when printed in grayscale. This topic has been extensively discussed in many journals and blogs: you are welcome to go and read them. For example this research showed that "...a perceptually appropriate color map leads to fewer diagnostic mistakes than a rainbow color map."

Of course we are not restricted to use awful, misleading, old default jet and the much better replacement default parula: there are many inbuilt colormaps and also many to download on MATLAB FEX that have been designed to optimally represent all kinds of data. They also include ones designed to have a wide range of hues:

https://www.mathworks.com/matlabcentral/fileexchange/?term=colormap+uniform

I very much doubt that jet's irregularities really represent your data distribution, and that its uneven lightness and chromacity really correspond to features of your data.

jet should be avoided. Use a better colormap (it does not have to be parula)

[James LaRue]

"Actually", the comment by Huzeyfe "is a good statement".

"jet" has its place in certain circumstances. Matlab's pixel-centric matrix visualization is one prime reason I stay with it and sometimes, well most times, I type in colormap jet. As an Applied Physics Signal Processor I have worked (Physics, Biology, Acoustics, Imagery, Data-bases, Electroencephologram, Electro-cardiograms) and I get clues through visualizing in 'jet' that I completely did not see with 'parula' (I learned something today); and yes those clues are relevant.

So, if one is a developer, maybe 'parula' is enough, but if one is in the investigative science side, 'jet' is my suggestion. www.DataPlasticity.com for examples of how I visualize things with 'jet' and I know would not come out as "nicely perceived" with autumn, summer, fall, or parula.

"Awful", "misleading"? Seems a little strong to me.

What I don't like is when I am coding and then SUGGESTIONS come up but cover-up what I am coding (it's a developer thing I am sure), and while I am here, not sure I like the movement of the zoom/pan icons to the top right of the plots; just a comment.

[Stephen23]

"So, if one is a developer, maybe 'parula' is enough, but if one is in the investigative science side, 'jet' is my suggestion"

So even given research that demonstrates that use of a rainbow colormap produces higher diagnostic errors, you would be happy for your doctor to diagnose you using one? In any case, why on earth are you restricting yourself to only comparing PARULA and JET, or for that matter, only inbuilt MATLAB colormaps? With minimal effort on the internet you could no doubt find much much more suitable colormaps for your own work. There are thousands out there, start looking!

'"Awful", "misleading"? Seems a little strong to me.'

Awful: sadly compared to many other colormaps, rainbow is quite garish and unattractive. There are many better colormaps on the internet (e.g. ColorBrewer is a good place to start, but there are many more for specific fields of research).

Misleading:

• "Evaluation of Artery Visualizations for Heart Disease Diagnosis" https://ieeexplore.ieee.org/document/6065015
• "Rainbow Color Map (Still) Considered Harmful" https://ieeexplore.ieee.org/abstract/document/4118486
• "Data visualization: the end of the rainbow" https://ieeexplore.ieee.org/abstract/document/736450
• "True Colors of Oceanography: Guidelines for Effective and Accurate Colormap Selection" https://www.jstor.org/stable/24862699?seq=1
• "Self-Corrected Perceptual Colormaps" https://domino.research.ibm.com/library/cyberdig.nsf/papers/A9B2B2048BB83B68852574440050B52F/$File/rc24542.pdf
• "Why We Use Bad Color Maps and What You Can Do About It" https://www.ingentaconnect.com/content/ist/ei/2016/00002016/00000016/art00022
• etc. etc.

jet should be avoided. Use a better colormap (it does not have to be parula)

[Bob photonics]

@Stephen Cobeldick

This alone is why jetmap is still useful for some people, because there's no automatic built-in function for this:How to display data with a focus on a different area

Non of these colormaps are bad, seem are ugly (subjectively), it's all about who uses it and how it's used.

It's misleading because either the person using it doesn't know exactly what they are doing or because they want to mislead their audience. Very easy example, is Word is easy to use, so people mess up the layout. Latex is harder to use but because it's so strict it's hard to mess up the layout. Still if one uses Word properly then it isn't really bad, many people complain that word is bad, but in reality it are the users who are bad at using it. Same with using a colormap Jet or whatever colormap you want to use.

So yes it's good jet is not the standard, I agree 100%. However that does not make Jet a bad colormap in specific situations it can be useful. Every paper can scream that it's misleading however much they want, doesn't make it a bad colormap if used in under the right circumstances.

More importantly doing ANY kind of quantative data analysis based on a plot with color is complete idiocricy, you can only use surface plots for qualitative analysis, hence why in certain situations Jet can be useful.

Lastly if something is published in color than it should be observed in color and not grayscale. You either publish it in Grayscale or in Color and the only time it should be compatible is when you NEED to publish in both. Otherwise you're losing out on the improvements you can have because of color.

[Stephen23]

"It's misleading because... " of the banding due to piecewise linear interpolation in RGB colorspace.

https://ai.googleblog.com/2019/08/turbo-improved-rainbow-colormap-for.html

JET

TURBO

Highly recommended reading:

https://www.mathworks.com/content/dam/mathworks/tag-team/Objects/r/81137_92238v00_RainbowColorMap_57312.pdf

Answer 1

[Massimo Zanetti]

You need to set it default at root level. Type this in the command line:

set(0,'DefaultFigureColormap',feval('jet'));

[Amir Semnani]

Thanks bro, it solved my problem, but each time I close MATLAB and open it, I should run the command, is there any permanent solution?

[Johnny B]

You can put the command above [and you can just use "jet" in the place of "feval('jet')"] in your startup.m file. If you don't already have a startup.m file -- and you should -- it's as simple as putting a file with that name in the directory in which you start MATLAB.

One annoyance I've noticed is that setting the default figure colormap causes a figure to open. For a long time, I've had the commands to set DefaultFigurePaperPositionMode and DefaultFigureToolbar in my startup.m, along with several settings for default axes properties. But now that I've added the command to set DefaultFigureColormap, I get a blank figure opened when I start MATLAB.

[Johnny B]

... and now I've figured out the problem. The 'jet' command (and presumably all the other colormap commands) gets the current figure's colormap size to decide how large of a matrix to return. It calls 'gcf', which is guaranteed to return a valid figure handle, so a figure gets created. I changed the command to

set(0, 'DefaultFigureColormap', jet(64))

and the gratuitous figure no longer appears.

[Irl Smith]

One minor glitch: apparently when this is set at level 0 (or at groot -- is that the same thing?) in startup.m, it cannot be changed by a subsequent set(0, ..., even in startup.m. This "stuck" behavior does not occur if the set(0... occurs in the command window.