Upsampling TIN

classic Classic list List threaded Threaded
8 messages Options
Reply | Threaded
Open this post in threaded view
|

Upsampling TIN

joolek
Hi Experts, 

I've tried so many software's without success.
Is it possible to upsample existing TIN? What I mean is for example divide each triangle into two for example. Not interpolation but somehow... subdivide or add more vertices between existing one?
Thank you for all your time
J

_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
Reply | Threaded
Open this post in threaded view
|

Re: Upsampling TIN

b.j.kobben

The main question is actually: why would you want to do that? You just end up with a larger data structure (more triangles) to show exactly the same surface model. You would not have any more detail, i.e the smaller triangles inside the original large one would have exactly the same azimuth and slope and therefore add no terrain detail...

 

yours,

 

--

Barend Köbben

 

 

 

On 04/10/2019, 09:41, "Qgis-user on behalf of joolek" <[hidden email] on behalf of [hidden email]> wrote:

 

Hi Experts, 

 

I've tried so many software's without success.

Is it possible to upsample existing TIN? What I mean is for example divide each triangle into two for example. Not interpolation but somehow... subdivide or add more vertices between existing one?

Thank you for all your time

J


_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
Reply | Threaded
Open this post in threaded view
|

Re: Upsampling TIN

ginetto
IMHO tension surface simulation for example can be an application where a TIN up sample could be necessary

Luigi Pirelli

**************************************************************************************************
* LinkedIn: https://www.linkedin.com/in/luigipirelli
* Stackexchange: http://gis.stackexchange.com/users/19667/luigi-pirelli
* GitHub: https://github.com/luipir
* Book: Mastering QGIS3 - 3rd Edition
* Hire a team: http://www.qcooperative.net
**************************************************************************************************


On Fri, 4 Oct 2019 at 10:14, [hidden email] <[hidden email]> wrote:

The main question is actually: why would you want to do that? You just end up with a larger data structure (more triangles) to show exactly the same surface model. You would not have any more detail, i.e the smaller triangles inside the original large one would have exactly the same azimuth and slope and therefore add no terrain detail...

 

yours,

 

--

Barend Köbben

 

 

 

On 04/10/2019, 09:41, "Qgis-user on behalf of joolek" <[hidden email] on behalf of [hidden email]> wrote:

 

Hi Experts, 

 

I've tried so many software's without success.

Is it possible to upsample existing TIN? What I mean is for example divide each triangle into two for example. Not interpolation but somehow... subdivide or add more vertices between existing one?

Thank you for all your time

J

_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user

_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
Reply | Threaded
Open this post in threaded view
|

Re: Upsampling TIN

joolek
Hi, 
I am more interested to have a point in between rather than the triangles, basically interpolate the data twice, to have an extra vertex between the existing one.
J

On Fri, 4 Oct 2019 at 09:23, Luigi Pirelli <[hidden email]> wrote:
IMHO tension surface simulation for example can be an application where a TIN up sample could be necessary

Luigi Pirelli

**************************************************************************************************
* LinkedIn: https://www.linkedin.com/in/luigipirelli
* Stackexchange: http://gis.stackexchange.com/users/19667/luigi-pirelli
* GitHub: https://github.com/luipir
* Book: Mastering QGIS3 - 3rd Edition
* Hire a team: http://www.qcooperative.net
**************************************************************************************************


On Fri, 4 Oct 2019 at 10:14, [hidden email] <[hidden email]> wrote:

The main question is actually: why would you want to do that? You just end up with a larger data structure (more triangles) to show exactly the same surface model. You would not have any more detail, i.e the smaller triangles inside the original large one would have exactly the same azimuth and slope and therefore add no terrain detail...

 

yours,

 

--

Barend Köbben

 

 

 

On 04/10/2019, 09:41, "Qgis-user on behalf of joolek" <[hidden email] on behalf of [hidden email]> wrote:

 

Hi Experts, 

 

I've tried so many software's without success.

Is it possible to upsample existing TIN? What I mean is for example divide each triangle into two for example. Not interpolation but somehow... subdivide or add more vertices between existing one?

Thank you for all your time

J

_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user

_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
Reply | Threaded
Open this post in threaded view
|

Re: Upsampling TIN

b.j.kobben

?? a point in between what? A TIN is a set of triangles, where each corner represents an X,Y,Z point in space (on a surface usually). So where do you nwat to have and extra vertex...? Sounds like you are not wanting aq TIN in the end, but something else...

 

Barend

 

On 04/10/2019, 10:47, "joolek" <[hidden email]> wrote:

 

Hi, 

I am more interested to have a point in between rather than the triangles, basically interpolate the data twice, to have an extra vertex between the existing one.

J

 

On Fri, 4 Oct 2019 at 09:23, Luigi Pirelli <[hidden email]> wrote:

IMHO tension surface simulation for example can be an application where a TIN up sample could be necessary

 

Luigi Pirelli

**************************************************************************************************
* LinkedIn: https://www.linkedin.com/in/luigipirelli
* Stackexchange: http://gis.stackexchange.com/users/19667/luigi-pirelli
* GitHub: https://github.com/luipir
* Book: Mastering QGIS3 - 3rd Edition

* Hire a team: http://www.qcooperative.net
**************************************************************************************************

 

 

On Fri, 4 Oct 2019 at 10:14, [hidden email] <[hidden email]> wrote:

The main question is actually: why would you want to do that? You just end up with a larger data structure (more triangles) to show exactly the same surface model. You would not have any more detail, i.e the smaller triangles inside the original large one would have exactly the same azimuth and slope and therefore add no terrain detail...

 

yours,

 

--

Barend Köbben

 

 

 

On 04/10/2019, 09:41, "Qgis-user on behalf of joolek" <[hidden email] on behalf of [hidden email]> wrote:

 

Hi Experts, 

 

I've tried so many software's without success.

Is it possible to upsample existing TIN? What I mean is for example divide each triangle into two for example. Not interpolation but somehow... subdivide or add more vertices between existing one?

Thank you for all your time

J

_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user


_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
Reply | Threaded
Open this post in threaded view
|

Re: Upsampling TIN

Nicolas Cadieux
In reply to this post by joolek
Hi,

You cannot invent new data between point without doing an interpolation. Your choice of interpolation will determine the quality of this data.  The only way I can think of is to create a raster from this TIN.  Then, resample (re-interpolated) this raster with a finer pixel and then make a new TIN.  This may create a finer TIN but this will significantly alter the original data.  You would need to evaluate the results and keep a backup of the old files.

Alternative, I would try with Mesh Lab.  It’s open source and probably has Mesh densification methods that would be more adapted.  Cloud Compare also has interesting algorithms.
Nicolas

> Le 4 oct. 2019 à 03:40, joolek <[hidden email]> a écrit :
>
> Hi Experts,
>
> I've tried so many software's without success.
> Is it possible to upsample existing TIN? What I mean is for example divide each triangle into two for example. Not interpolation but somehow... subdivide or add more vertices between existing one?
> Thank you for all your time
> J
> _______________________________________________
> Qgis-user mailing list
> [hidden email]
> List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
> Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
Reply | Threaded
Open this post in threaded view
|

Re: Upsampling TIN

Falk Huettmann
Hi,
I agree with Barend and Nicolas here.

It's a typical and essential question that we get all the time, like,
can we re-grid and subsample pixels ?
=>You can, technically, but you stay just within the same data,
and thus, create nothing new or meaningful.

In other words, it should not be done.

There is a lot wrong with re-sampling, gridding, within DEMs, with merging grids,
reprojecting and interpolation surfaces. And GIS techs are pretty guilty of that,
so are their supervisors.
The tools we provide should point that out clearly and not allow for it.

Key to those questions remain 'ethics', metadata of course (!) and then always a comparison as a benchmark for 
accuracy and to see the actual gain of such re-samplings.

(but let's agree that TINs are only one way of getting a surface, often not the best one,
when compared to the 'truth'. Playing around with them for best output is not a bad idea)

  Very best
    Falk Huettmann




On Fri, Oct 4, 2019 at 4:00 AM Nicolas Cadieux <[hidden email]> wrote:
Hi,

You cannot invent new data between point without doing an interpolation. Your choice of interpolation will determine the quality of this data.  The only way I can think of is to create a raster from this TIN.  Then, resample (re-interpolated) this raster with a finer pixel and then make a new TIN.  This may create a finer TIN but this will significantly alter the original data.  You would need to evaluate the results and keep a backup of the old files.

Alternative, I would try with Mesh Lab.  It’s open source and probably has Mesh densification methods that would be more adapted.  Cloud Compare also has interesting algorithms.
Nicolas

> Le 4 oct. 2019 à 03:40, joolek <[hidden email]> a écrit :
>
> Hi Experts,
>
> I've tried so many software's without success.
> Is it possible to upsample existing TIN? What I mean is for example divide each triangle into two for example. Not interpolation but somehow... subdivide or add more vertices between existing one?
> Thank you for all your time
> J
> _______________________________________________
> Qgis-user mailing list
> [hidden email]
> List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
> Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user

_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
Reply | Threaded
Open this post in threaded view
|

Re: Upsampling TIN

Saber Razmjooei
Hi,

If you want to see your TIN as raw data, without interpolation and conversion to gridded format, best to have it supported as a "mesh layer":

To add support for it, you need to have it in MDAL:

There are processing tools to convert it to regular vector or raster (at user-defined resolution), but you should be able to work with the native nodes/edges/surfaces directly from QGIS.

Regards
Saber

On Fri, 4 Oct 2019 at 18:09, Falk Huettmann <[hidden email]> wrote:
Hi,
I agree with Barend and Nicolas here.

It's a typical and essential question that we get all the time, like,
can we re-grid and subsample pixels ?
=>You can, technically, but you stay just within the same data,
and thus, create nothing new or meaningful.

In other words, it should not be done.

There is a lot wrong with re-sampling, gridding, within DEMs, with merging grids,
reprojecting and interpolation surfaces. And GIS techs are pretty guilty of that,
so are their supervisors.
The tools we provide should point that out clearly and not allow for it.

Key to those questions remain 'ethics', metadata of course (!) and then always a comparison as a benchmark for 
accuracy and to see the actual gain of such re-samplings.

(but let's agree that TINs are only one way of getting a surface, often not the best one,
when compared to the 'truth'. Playing around with them for best output is not a bad idea)

  Very best
    Falk Huettmann




On Fri, Oct 4, 2019 at 4:00 AM Nicolas Cadieux <[hidden email]> wrote:
Hi,

You cannot invent new data between point without doing an interpolation. Your choice of interpolation will determine the quality of this data.  The only way I can think of is to create a raster from this TIN.  Then, resample (re-interpolated) this raster with a finer pixel and then make a new TIN.  This may create a finer TIN but this will significantly alter the original data.  You would need to evaluate the results and keep a backup of the old files.

Alternative, I would try with Mesh Lab.  It’s open source and probably has Mesh densification methods that would be more adapted.  Cloud Compare also has interesting algorithms.
Nicolas

> Le 4 oct. 2019 à 03:40, joolek <[hidden email]> a écrit :
>
> Hi Experts,
>
> I've tried so many software's without success.
> Is it possible to upsample existing TIN? What I mean is for example divide each triangle into two for example. Not interpolation but somehow... subdivide or add more vertices between existing one?
> Thank you for all your time
> J
> _______________________________________________
> Qgis-user mailing list
> [hidden email]
> List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
> Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user


--
Saber Razmjooei
www.lutraconsulting.co.uk
+44 (0)7568 129733

_______________________________________________
Qgis-user mailing list
[hidden email]
List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user