> Wouldn't it be possible to have the test machine generate the ref. image
> itself, for instance the first time a (new) visual test is run?
>
> Yes that's a possibility indeed.
> This would solve any enviro discrepancy (in addition to differences in
> rendering, for instance a small difference in fonts used or in
> anti-aliasing
> or any other detail, may generate a false positive).
>
The font should be the same. Antialising is something we should test
indeed. Maybe reference files should be created with antialiasing off.
>
> Also, a 1-pixel difference -- which seems to cover the majority of
> differences in the example test run -- may be rather irrelevant. Any change
> to make the test (a little bit) more tolerant?
I tried to tweak the call to image magick compare without success. I also
tried to manipulate the resulting file without success. Any idea is
welcome, but I would prefer to solve the root cause.
> Ultimately, a visual test
> makes sense for differences which are *visible*: I think my eye is not
> untrained, but I could not spot any *real* difference between the reference
> in the test images and the sample test run.
>
>
Then the tests are passing and it's the goal after all :)
lasconic
------------------------------------------------------------------------------
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
the latest Intel processors and coprocessors. See abstracts and register >
http://pubads.g.doubleclick.net/gampad/clk?id=60133471&iu=/4140/ostg.clktrk
_______________________________________________
Mscore-developer mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/mscore-developer
