An update for you all...
After a bit of a delay, due to some urgent work for a major display manufacturer, we have at last got around to looking in detail at the evaluations Zoyd performed.
As per the message I sent him directly, and as I also posted in this thread, we have indeed found a 'rounding error' we managed to introduce into LightSpace some time back, and had missed.
If Zoyd (or Spectracal for that matter) had been polite enough to provide the test data Zoyd use we would have found this instantly...
But no matter, because as Zoyd stated. the error was basically invisible.
However, also as I posted, we strive for perfection, so were keen to understand what the error was.
Here is the data in an easy to understand graphical form.
First the error in the existing release version of LightSpace, as tested by Zoyd.
This first image shows the 'noise' as a 3D cube representation of the data.
It is easy to see the 'noise' at the edge of the cube,
And this as a 1D graph, where the noise shows as colour distortions of the graph plot.
And the following images are exactly the same profile data with the new version of LightSpace we will be releasing shortly (it's undergoing Beta testing at the moment).
The difference is rather obvious.
There may be other rounding errors we have missed, so we are running more tests before we release this new version.
But, as above, the initial results seem to suggest the main (but invisible) issue has been solved.
Steve
Steve Shaw
Mob Boss at Light Illusion