getframe difference in 2014b.... bug?

2 Ansichten (letzte 30 Tage)
Sven
Sven am 18 Dez. 2014
Kommentiert: Sven am 11 Mär. 2015
The following code run in 2014a and 2014b produces remarkably different results. I think it's a bug.
figure('Position',[100 100 200 500],'Color','y');
I = imread('rice.png');
imagesc(I)
axis image
colormap(gray)
gf = getframe(gca);
figure, imshow(gf.cdata)
In 2014a it successfully captures ONLY the current axes:
In 2014b it grabs beyond the axes limits to the extent of the figure:
  • The figure "Color" parameter makes no difference - I just used yellow for clarity.
  • The axis image line does the damage. Without it both 2014a and 2014b capture only the axes extents (as they should).
Can anyone suggest a workaround that produces identical results in both 2014a and 2014b? I have code that runs on both systems and is expected to produce similar figures using getframe. I understand that there can be pixel border effects when using getframe, these minor differences are fine, but the results shown are major differences that I want to avoid.
And if someone can confirm that they get similar results I'll submit a bugfix request.
Thanks, Sven.
  2 Kommentare
matt dash
matt dash am 18 Dez. 2014
Confirmed that it does the same thing for me.
The problem seems to be that axis image changes the view without changing the axes position property, which is what getframe is using to calculate the extent of the screen capture. Because of this, it will also break any other getframe-like function that uses the axes position property. (There are several on the file exchange... usually i recommend them when someone has a property with getframe, but for your case something else will be needed.)
Sven
Sven am 11 Mär. 2015
Note also that this bug persists in 2015a.

Melden Sie sich an, um zu kommentieren.

Akzeptierte Antwort

Kelly Kearney
Kelly Kearney am 18 Dez. 2014
I'll also confirm, but older versions of Matlab also responded to axis image without changing the axis's Position property, so that's not the underlying problem. Rather, it must be a change in whether getframe considers aspect ratio properties and modes in the same way.
You can get around the bug by giving getframe precise coordinates to capture; plotboxpos will give you those coordinates:
figure('Position',[100 100 200 500],'Color','y');
I = imread('rice.png');
imagesc(I)
axis image
colormap(gray)
drawnow;
gf = getframe(gca);
un = get(gca, 'units');
set(gca, 'units', 'pixels');
rec = plotboxpos(gca);
set(gca, 'units', un);
gf2 = getframe(gcf, rec);
figure, imshow(gf.cdata)
figure; imshow(gf2.cdata);

Weitere Antworten (0)

Kategorien

Mehr zu Graphics Object Programming finden Sie in Help Center und File Exchange

Produkte

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by