[vtk-developers] Code review request: biquadratic quadinterpolation

[John Platt]

Hi Marcus,

I looked at vtkQuadraticQuad::EvaluatePosition() (VTK5.10) . This assumes 
the same points for the origins of the linear quads as vtkBiQuadraticQuad...

    subId 0 :  (    0,    0 ) - point 0,  linear quad { 0, 4, 8, 7 }
             1 :  ( 0.5,    0 )            4,                    { 8, 4, 1, 
5 }  ** inconsistent **
             2 :  ( 0.5, 0.5 )            8,                    { 8, 5, 2, 
6 }
             3 :  (    0, 0.5 )            7,                    { 7, 8, 6, 
3 }

It looks like a bug in vtkQuadraticQuad which has been copied into 
vtkBiQuadraticQuad.

John.

----- Original Message ----- 
From: "Marcus D. Hanwell" <marcus.hanwell at kitware.com>
To: "John Platt" <jcplatt at dsl.pipex.com>
Cc: "David Thompson" <david.thompson at kitware.com>; "VTK Developers" 
<vtk-developers at vtk.org>
Sent: Friday, January 24, 2014 6:57 PM
Subject: Re: [vtk-developers] Code review request: biquadratic 
quadinterpolation


> Hi John,
>
> Thanks for your input - I really appreciate it. I will see if I hear
> back from the contact David Thompson pointed out, and also try to find
> some time to look at the vtkQuadraticQuad to ensure they are
> consistent.
>
> Marcus
>
> On Thu, Jan 23, 2014 at 6:21 PM, John Platt <jcplatt at dsl.pipex.com> wrote:
>> Hi Marcus,
>>
>> I tested this change using 5.10 and it did not affect colour mapping,
>> cutting or clipping. Unfortunately, I don't do any probing so I cannot 
>> check
>> this.
>>
>> There would appear to be a dependency on the starting point for the 
>> linear
>> quads with the 'adjust parametric coordinates' section in
>> EvaluatePosition(). Here, the parametric coordinates from the linear quad
>> are converted to parametric coordinates in the quadratic quad. The 
>> origins
>> of each linear quad are as follows ...
>>
>>  (    0,    0 ) - point 0
>>  ( 0.5,    0 )            4
>>  ( 0.5, 0.5 )            8
>>  (    0, 0.5 )            7
>>
>> So I would say that the proposed change is consistent with these origins.
>>
>> vtkQuadraticQuad uses the same linear quads as the old vtkBiQuadraticQuad 
>> so
>> it could be a problem here also.
>>
>> HTH
>>
>> John.
>>
>> ----- Original Message ----- From: "Marcus D. Hanwell"
>> <marcus.hanwell at kitware.com>
>> To: "David Thompson" <david.thompson at kitware.com>
>> Cc: "VTK Developers" <vtk-developers at vtk.org>
>> Sent: Thursday, January 23, 2014 9:16 PM
>> Subject: Re: [vtk-developers] Code review request: biquadratic
>> quadinterpolation
>>
>>
>>> On Thu, Jan 23, 2014 at 3:48 PM, David Thompson
>>> <david.thompson at kitware.com> wrote:
>>>>
>>>> Hi Marcus,
>>>>
>>>>> There is a topic on Gerrit I am having trouble finding a reviewer for,
>>>>> http://review.source.kitware.com/#/t/3848/
>>>>
>>>>
>>>> Soeren Gebbert (soerengebbert at googlemail.com) is the original 
>>>> developer.
>>>> I'm really swamped right now, but it would be great to verify with 
>>>> Sandia
>>>> that the change does not mess with the Exodus reader's numbering of 
>>>> nodes
>>>> within the cell... several changes have done so in the past.
>>>>
>>> It would be nice to figure out what the expected behavior is too,
>>> because this doesn't break any tests I can see. It is a single line
>>> change, I can email the original developer but don't have much more
>>> time I can put into this.
>>>
>>> Marcus
>>> _______________________________________________
>>> Powered by www.kitware.com
>>>
>>> Visit other Kitware open-source projects at
>>> http://www.kitware.com/opensource/opensource.html
>>>
>>> Follow this link to subscribe/unsubscribe:
>>> http://www.vtk.org/mailman/listinfo/vtk-developers
>>>
>>>
>>
>>
>