>> Does it make sense?

No, the problem is not related to the output resolution of raster images (actually it will be the same with vector outputs like svg or PDF). On my computer I see no clipping with the code from comment #4. The title is stuck on top of the image though so adding a bit of tolerance would not hurt, e.g.

alpha = get (0, "screenpixelsperinch")/72 * 1.05;

Sorry, the example from comment #3 was untested and "labelfontsizemultiplie" should obviously be replaced by "titlefontsizemultiplier":

figure(2)
titlemultiplier = get (0, "defaultaxestitlefontsizemultiplier");
alpha = get (0, "screenpixelsperinch")/72;
plot(1,1)
ht = title(sprintf('first line\nsecond line'))
set(gca,'titlefontsizemultiplier', alpha*titlemultiplier, ...
    'positionconstraint','innerposition', ...
    'titlefontsizemultiplier', titlemultiplier)
print('good.png')

in which way I can add more room for the title?

BTW I am under Linux and not Windows

I have a script that produces many output images from many plot kinds (plotyy, plot with subplots, etc). I am trying to align horizontally the plots by setting the first and third element of the "position" axis property. I do other customizations such as change of the fontsize, for this reason I need to force "positionconstraint" to innerposition too.

The plots look fine in the figures while print() produces an output image that has a portion of the title text clipped when using multiline. The minimum code snippet shows the difference in the result:

figure(1)
plot(1,1)
title(sprintf('first line\nsecond line'))
set(gca,'positionconstraint','innerposition')
print('bad.png')

figure(2)
plot(1,1)
title(sprintf('first line\nsecond line'))
set(gca,'positionconstraint','outerposition')  % not needed
print('good.png')

No annotation is added after the set() call so all the graphic elements should have a defined position. If I invert the set() and title() calls the result is the same, of course in that case the dimensions and position of the plots would be slightly different.

With print() I expect to have the same result visible in the figure.