[Paraview] colormap for logarithmic quantities

John Doe ufnoise at gmail.com
Tue Jul 8 15:17:30 EDT 2008


Actually the data would vary between 0 and 1e19.  In my example, the
donor doping varies between 0 and 1e19.

Using an asinh filter, as opposed to log(abs()) filter, would ensure
that the data would never blow up.  A log(abs(x) + 1e-308) filter
would also never blow up.

I think this would be a nice feature for the program as a checkbox, as
opposed to requiring the use of the calculator each time.  I'll file
an enhancement request.

Regards,

Juan

On Tue, Jul 8, 2008 at 2:09 PM, Kent Eschenberg <eschenbe at psc.edu> wrote:
> I didn't mean "do your data values make sense"; I'm sure they do.
>
> What I questioned is whether changing the lower bound of the log color scale
> as you suggest makes sense.
>
> Remember, changing the color scale does not change the data -- the original
> values, both positive and negative, will be passed through the color lookup
> table to find their color.
>
> I think with the change you suggest all values from -1e+15 to +1e+15 would
> get the same color, the one for +1e+15. The colors would only vary between
> +1+e15 an 9.999e+18. Does that make sense?
>
> If you want to work with the absolute value of your data first pass it
> through the Calculator and take the absolute value. If needed also set a
> floor so values don't get too close to 0 where the log blows up.
>
> Kent
> Pittsburgh Supercomputing Center
>
> John Doe wrote:
>>
>> In semiconductor devices, the net doping is a signed quantity which
>> can very exponentially.  It is typical in most device simulation
>> packages as it refers to the signed number of donors (positive) and
>> acceptors (negative).
>>
>> Other candidates for a colormap options would be a "Use log of the
>> absolute value", or even "Use asinh scale".
>>
>> Regards,
>>
>> Juan
>>
>> On Tue, Jul 8, 2008 at 1:23 PM, Kent Eschenberg <eschenbe at psc.edu> wrote:
>>>
>>> That would prevent the error but would it make sense?
>>>
>>> Kent
>>> Pittsburgh Supercomputing Center
>>>
>>> John Doe wrote:
>>>>
>>>> When I try to view a color map logarithmically, I get the following
>>>> error:
>>>> ERROR: In
>>>> /home/berk/Work/ReleaseBuilds/ParaView3/VTK/Common/vtkLookupTable.cxx,
>>>> line 120
>>>> vtkLookupTable (0x8ff2ae0): Bad table range for log scale: [-1e+15,
>>>> 9.999e+18], adjusting to [1, 10]
>>>>
>>>> Would it be reasonable to request that the log option for the colormap
>>>> takes the absolute value of the quantity for which the log is taken?
>>>>
>>>> Thanks,
>>>>
>>>> Juan
>


More information about the ParaView mailing list