Printing Figures Very Slow 2014b vs 2013b

Just installed 2014b – and the printing is a step back to Gutenberg. I have Mac OS X 10.9.3.

Figures with Opengl renderer:

2013b >> tic;print('-dpng','-r300',[outpath,outfile]);toc
Elapsed time is 0.528570 seconds.

2014b >> tic;print('-dpng','-r300',[outpath,outfile]);toc
Elapsed time is 1.786196 seconds.

It does not matter which device or resolution I use:

2013b >> tic;print('-dtiff',[outpath,outfile]);toc
Elapsed time is 0.296846 seconds.

2014b >>tic;print('-dtiff',[outpath,outfile]);toc
Elapsed time is 0.845170 seconds.

Things get even worse if I use the painters renderer:

2013b >> tic;print('-dpng','-r300','-painters',[outpath,outfile]);toc
Elapsed time is 0.476110 seconds.

2014b >> tic;print('-dpng','-r300','-painters',[outpath,outfile]);toc
Elapsed time is 6.714240 seconds.

This is a serious time hit.

Answer 1

[Bruno Pop-Stefanov]

Hi Robert,

Thanks for your feedback. I forwarded your message to the development team in charge of the print function. This is a known issue in MATLAB R2014b and they are actively working on it.

Best,

Bruno

[simsalakim]

Is there already any solution known for this problem?

[Stefan]

Was the printing performance improved in R2015a? I cannot find any info in the release notes.

[Gregory]

It hasn't been fixed in 2015b. We just upgraded matlab on our cluster and a script that use to take 1 to 3 hours to generate 20 large figures now can't even create one in 4 days.

[Jed]

Yeah... I just printed a 600 x 600 image to png. It took over 2 minutes on a very fast machine. 1200x1200 took ~6 minutes... I want 6000 x 6000, but don't want to wait overnight for it!

[Jed]

Update: The problem seems to be the opengl renderer... After initially switching to zbuffer it was much faster, but the result was not what I want (transparancy is gone, e.g.). I then tried on a machine with hardware acceleration enabled (I had disabled it to prevent vnc from crashing on the first machine) and it worked fine (only 2 minutes for 6000x6000).

Answer 2

[XGWS]

This problem seems not fixed in MATLAB R2016a either. But you can get around this problem by using "saveas(gcf,'test.png')". The drawback is that you can not adjust the resolution (e.g., default and fixed: 150dpi for .png)