[OpenSCAD] Light source and default camera position different on command line
nop.head at gmail.com
Sun Sep 16 06:00:24 EDT 2018
Yes I have reviewed them and they all see to just differ in lighting, apart
from the few that fail on my machine due to different Z fighting. If I
update those for the lighting change then presumably they will fail on
somebody else's hardware. Is that OK?
I am thinking the vector mode can only do a subset of the gimbal mode
because it can't rotate around Y. If so, to avoid GUI and command line
diverging again, we can remove all the vector camera code and just convert
command line vector parameters to gimbal parameters.
If we do unify the camera modes we have some decisions to make. As well as
lighting differences the default view is at a different angle and distance.
The command line looks towards the first quadrant at a shallower angle
[70.53, 0, 315]. The GUI looks at the second quadrant with a steeper
downwards angle [55, 0, 25] The default distance is 500 on the command line
but 140 in the GUI.
I think looking to the first quadrant is more logical, so the azimuth
should always be 315. To avoid big changes to the tests I think we should
adopt 70 for the altitude. That will make tiny changes to the tests but the
lighting means they all need regenerating anyway.
Distance 500 means you can see about +/- 100 in the Y direction. 140 shows
about +/- 30. I propose to make the GUI distance 500 to match the command
line and the tests. When you reset the view it is better to be able to see
Although I have fixed the lighting bug I don't totally understand why. I
think when you call gluLookAt with the centre at the origin and the eye on
the X axis you will end up with a matrix similar to the identity matrix but
with a few different signs. I don't understand how that can change the
lighting without altering the view in any other way. If the modelview
matrix is different in any way, surely the view should change spatially?
On 16 September 2018 at 00:55, Marius Kintel <marius at kintel.net> wrote:
> > On Sep 15, 2018, at 8:18 AM, nop head <nop.head at gmail.com> wrote:
> > The few tests that failed now pass and all the vector camera tests fail,
> about 40% of the total. Is there a quick way of updating all the test
> images or do I have to manually dig them out and replace them all?
> Regenerating failing tests is a necessary evil, and requires some manual
> ..but we can automatically generate these expected files on the cmd-line:
> $ TEST_GENERATE=1 ctest -R <test regex>
> Some info here, but not too well organized: https://github.com/openscad/
> OpenSCAD mailing list
> Discuss at lists.openscad.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Discuss