[MDAL-Developer] 3D Meshes

Sun Jun 27 10:03:00 PDT 2021

Hi,
My comments between lines below.

Le dim. 27 juin 2021 à 12:23, Paul Harwood <runette at gmail.com> a écrit :

> Re-opening this and adding back the wider group( I had not noticed that
> they had dropped off).
>
> The question is about saving 3D meshes with volumetric datasets. The
> context is around creating a python interface and using it for 3D
> volumetric data.
>
> There are a number of questions coming out of this work - I will raise
> individual issues and PRs where appropriate but I wanted to discuss the
> issues wider:
>
> 1 There seems to be a general lack of drivers capable of writing any sort
> of data - according to the flags and therefore also according to the C API
> which checks the flags. UGRID, for instance, was mentioned but the Ugrid
> driver capabilities do not include the ability to write ANY data. Is that
> correct? It seems to contradict what has been said elsewhere.
>

UGRID does not write dataset for now, but it is on wish list...

> There are three drivers that can write meshes (2DM, SELAFIN and UGRID). Of
> these only one can write data - SELAFIN - and only on vertices. This is
> very limited. Is it correct?
>

Correct, only SELAFIN can write dataset AND mesh.
Indeed it is very limited.

> The flags also say that DataOnFace can be written by the TUFLOW FV driver
> but that meshes cannot, and that data can be written by the DAT drivers.
> But none of that helps with creating a new mesh with data of any type (let
> alone 3D).
>

Not sure TUFLOW FV can write anything.

> 2 There are NO drivers that admit to the ability to write DataOnVolumes
> datasets.
>
> Is this true?
>

I think it is true.

>
> 3 The C API MDAL_G_addDataset() method explicitly excludes (i.e. there is
> a validation check to enforce the exclusion) creation of DataOnVolume
> datasets. This does make the conversation earlier in the email chain about
> how to create these datasets incorrect. Is there a reason for this
> exclusion and is there a way around it.
>

 Since there is a check about the supported data location just before this
enforced exclusion, I don't see any reason, but maybe Peter can know about.

> Thanks for any pointers / comments
>
> Paul
>
> On Thu, 17 Jun 2021 at 10:22, Paul Harwood <runette at gmail.com> wrote:
>
>> Ok, thanks. Got it now
>>
>> On Thu, 17 Jun 2021 at 10:20, Peter Petrik <
>> peter.petrik at lutraconsulting.co.uk> wrote:
>>
>>> base net is same as 2D mesh, verticals are just datasets defined in
>>> https://github.com/lutraconsulting/MDAL/blob/42eb61f880f99ca0aa89a72f5afc5193b2f061e2/mdal/api/mdal.h#L648
>>>
>>>
>>>
>>> On Thu, Jun 17, 2021 at 11:07 AM Paul Harwood <runette at gmail.com> wrote:
>>>
>>>> To be more precise - I can see how to add volume based datasets but not
>>>> how to add the vertical structure - i.e. the number of volumes per face of
>>>> the base net.
>>>>
>>>> On Thu, 17 Jun 2021 at 08:18, Paul Harwood <runette at gmail.com> wrote:
>>>>
>>>>> We probably need to change or clarify the README which says that 3D
>>>>> data is not supported for UGRID. Does that column mean something else ?
>>>>>
>>>>> Is the support read-only? I cannot immediately see how to use the C
>>>>> API to write a layered 3D mesh.
>>>>>
>>>>> On Thu, 17 Jun 2021 at 07:27, Peter Petrik <
>>>>> peter.petrik at lutraconsulting.co.uk> wrote:
>>>>>
>>>>>>
>>>>>> https://ugrid-conventions.github.io/ugrid-conventions/#3d-layered-mesh-topology
>>>>>> is supported
>>>>>>
>>>>>> On Wed, Jun 16, 2021 at 8:38 PM Paul Harwood <runette at gmail.com>
>>>>>> wrote:
>>>>>>
>>>>>>> Sorry - in that last email when I said "is UGRID unstructured 3d
>>>>>>> MESH supported"? I meant "is UGRID LAYERED 3d mesh supported"!
>>>>>>>
>>>>>>> On Wed, 16 Jun 2021 at 19:35, Paul Harwood <runette at gmail.com>
>>>>>>> wrote:
>>>>>>>
>>>>>>>> Peter
>>>>>>>>
>>>>>>>> Can I just confirm. I look at the GH readme and it says that only
>>>>>>>> XMDF supports 3d (i.e. I guess TUFLOW). There is an issue about UGRID that
>>>>>>>> mentions a PR. Also the Readme says that XMDF support for 3D is read-only.
>>>>>>>>
>>>>>>>> Is UGRID Unstructured 3D mesh supported?
>>>>>>>>
>>>>>>>> Is any 3D RW?
>>>>>>>>
>>>>>>>> On Wed, 16 Jun 2021 at 14:59, Paul Harwood <runette at gmail.com>
>>>>>>>> wrote:
>>>>>>>>
>>>>>>>>> That makes sense. Thanks. I think that ugrid is what I want
>>>>>>>>>
>>>>>>>>> On Wed, 16 Jun 2021, 14:45 Peter Petrik, <
>>>>>>>>> peter.petrik at lutraconsulting.co.uk> wrote:
>>>>>>>>>
>>>>>>>>>> I think we only support TUFLOW 3d data and UGRID specification
>>>>>>>>>> for 3D stuff, which limits your options quite a bit :)
>>>>>>>>>>
>>>>>>>>>> P.
>>>>>>>>>>
>>>>>>>>>> On Wed, Jun 16, 2021 at 3:28 PM Paul Harwood <runette at gmail.com>
>>>>>>>>>> wrote:
>>>>>>>>>>
>>>>>>>>>>> Hi all
>>>>>>>>>>>
>>>>>>>>>>> I am thinking about voxel based 3D data in meshes.
>>>>>>>>>>>
>>>>>>>>>>> I am aware of this
>>>>>>>>>>> https://github.com/qgis/QGIS-Enhancement-Proposals/issues/158
>>>>>>>>>>>
>>>>>>>>>>> The question I have is a simple one :
>>>>>>>>>>>
>>>>>>>>>>> Absent any other constraint, what would be the "best" format to
>>>>>>>>>>> use to hold and transfer voxel based 3D mesh data?
>>>>>>>>>>> _______________________________________________
>>>>>>>>>>> MDAL-Developer mailing list
>>>>>>>>>>> MDAL-Developer at lists.osgeo.org
>>>>>>>>>>> https://lists.osgeo.org/mailman/listinfo/mdal-developer
>>>>>>>>>>>
>>>>>>>>>> _______________________________________________
> MDAL-Developer mailing list
> MDAL-Developer at lists.osgeo.org
> https://lists.osgeo.org/mailman/listinfo/mdal-developer
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/mdal-developer/attachments/20210627/cd214aa7/attachment-0001.html>