isotrim (large)surface- getting a untrimmed surface in the end

Ok ..I have a huge D.E.M surface out of which I want to parcel out a test patch. I also want this patch to be divided by a set unit. I able to write a script for it, but due to the large number of sub-surfaces, the script crashes( specifically if the length of division is lowered below 50m). Is there any other way?

The thing is if you split the surface you get a trimmed surface, nd if you iso trim the surface, the whole patch gets divided. Thus I wrote the script in a way that divides the whole surface and selects the cells that fall within the test patch curve. But now due to very high number of divisions, thus the script fails there.

Any help is much appreciated!

Best,

Aditya

Attachments:

surface divide (landscape).gh, 2.6 MB

Replies to This Discussion

Permalink Reply by Aditya Tognatta on October 24, 2016 at 10:42am

Any luck guys?

Permalink Reply by Aditya Tognatta on October 24, 2016 at 3:22pm

???

Permalink Reply by peter fotiadis on October 25, 2016 at 4:05am

My Dear Watson:

More is certainly more (like Godzilla) but ... well ... 500+K divisions ????

BTW: I have a slightly used CRAY for you > 10M down payment > yours > hurry > life's short.

Permalink Reply by Aditya Tognatta on October 25, 2016 at 4:27am

Dear Lord, your absence was certainly missed here. I'm aware of the Godzilla,although I just need to extract a mouse from it. Is there a way just to divide the patch I want, perhaps! just divide the domain corresponding to the test patch only ( but how do we figure that out?).

Thanks in advance.

Aditya

Permalink Reply by peter fotiadis on October 25, 2016 at 4:44am

Indeed I was out in the wild lately (searching the Truth Out There > none found, mind > blame Karma).

Your thingy is 4*4 km > not a Godzilla by any means. I'll prepare a small demo with regard mouse matters as soon as I can find a proper triple espresso (NOTE: Huan Valdez, what else?) + a decent cigar (NOTE: Ramon Allones, what else? > notes of marzipan and citrus blah, blah).

more soon, Lord of Darkness, Prince of Desert Rats.

Permalink Reply by Aditya Tognatta on October 25, 2016 at 8:04am

Dear Lord,

Were you able to trap the mouse?

Permalink Reply by peter fotiadis on October 25, 2016 at 9:39am

Well ... the mouse is one thing BUT the fact that I'm out of Ramon stuff (WHAT a disaster > panic > should I commit seppuku? ) that's the big thingy of the day.

Moral: My kingdom for a decent cigar.

Attachments:

surfacedividelandscape_V1.gh, 2.7 MB

Permalink Reply by peter fotiadis on October 26, 2016 at 3:04am

On the other hand maybe a more elaborated approach could skin the cat better:

1. Create a grid (polylines, points, cats, dogs) that visually gives you a clue about what the current "tessellation" (sizeU/V) means. This could be "arbitrary": first locate a greater area of interest and then subdivide further.

2. Get the mouse (surface segment) due to a boolean between the surface and a box/brep made from a given tree "cell". This yields canonical pieces (if this is required).

Added the grid but not step 2 (it's just an idea) therefor the method for the mouse remains as in V1.

BTW: For this particular case ... potential "delays" MAY are due to some OpenGL related thingy (meaning: borrow from a friend some decent Kepler Quadro [K2200 and on, a K4200 is way better - Note: SLI is kinda 3rd marriage] and test the C#).

BTW: Without // processing (NOT an one minute job, mind) and having in mind your true goal (water reservoirs etc etc) I foresee agony/pain/frustration/tears/despair.

Plan B: Forget water reservoirs > try collecting red pepper Vodka.

Attachments:

surfacedividelandscape_V2.gh, 2.7 MB

Permalink Reply by Aditya Tognatta on October 26, 2016 at 5:50am

Hi Peter, The mouse got caught eventually, it did require manual tuning( moving the start u and v to the required location was a pain. The rebuild made the surface super curvy( almost like a scenic landscape we drew as kids, with the sun rising behind the mountains). I have the Quadro K4000M so I can use the cuda cores. But the cool thing was the mouse was an untrimmed surface, once I divided it according to the cell size I wanted. It was fine. Perhaps! not the rebuild the surface might just work /look better.

Will try V2 ASAP.

Best,

Aditya

Permalink Reply by peter fotiadis on October 26, 2016 at 8:58am

Rebuild > super curvy ?? Strange (pair of K4200's on duty + the useless SLI "bridging" [works on top MCAD stuff, mind like NX/Catia]). Here's how it looks with a "reasonable" rebuild U/V count.

Anyway the rebuild is optional.

Permalink Reply by peter fotiadis on October 26, 2016 at 9:30am

BTW: Your trimmed thingy VS the BrepFace underlying surface (as used by the 1st C#).

Permalink Reply by Aditya Tognatta on October 27, 2016 at 8:37am

Greeting from the land of the mortals The definition works like a charm. I had already setup my definition; just had to change the surface to do the analysis. Your script worked like a charm. But now I'm exploring another domain ( I want to convert my rhino model into a D.E.M *.txt file that I can upload into RAMMS ( software for ca

Attachments: