Large surfaces in C3D?

[akbowbender]

I have a LDT project for 65 miles of new road. The terrain models are from LIDAR gridded to 15'. The largest tin file is just over 300mb. Can C3D handle this size file?

Autodesk seems to be hanging it's hat on C3D, so I guess I can't expect much improvement in stability or new features in LDT. It can handle the large surface files, but it does have stability problems at times. If C3D can handle large surfaces, I may give it another try.

[Anonymous]

I'm pretty sure I could pick you out of a canoe.

John P.

[Anonymous]

My math may be a bit off but assuming they shot a 500' swath we're talking
about less than a million points here. I didn't think the ceiling was that
low.

Maybe my machine just isn't up to the task (Dell Precision M70, 2.3 ghz, 2.0
gb RAM) but I just attempted to build a surface from 2' contours (39,000+)
and crashed in just under 40 seconds. (I think with a bit more processor
speed I can get it down to under 30.) Nothing else was running and I even
disabled all options for minimizing flat areas.

Don't get me wrong - I'm not at all really interested in who can come up
with the most extreme dataset but this is a definite problem in the
petroleum industry and increasingly becoming a problem with advancing
"normal" survey technology. Can you share with us the prognosis (if any) for
future releases and/or service packs?

--
Mike Norton
Total CAD Systems, Inc. - Houston, Texas
"Anthony Governanti" wrote in message
news:5257672@discussion.autodesk.com...
Ok folks, I got the LIDAR point file from "akbowbender" and was able to get
it to work pretty well. its a 55 MB text file of XYZ coordinates. here is
what I recommended:

Ok, so I was able to work with the surface and get an alignment and profile
generated.



Here are some best practice suggestions when working with large amounts of
data:

1.. Use the Point File option for building the surface
1.. Create a new surface in a new drawing
2.. Use a Styl
e that just displays the Border (no contours or triangles)
3.. Expand the surface definition, then right click on the Point File
collection
4.. Choose Add.
5.. Select the appropriate file format (if there isn't one that matches
you can define one) - in this case, I used ENZ (XYZ) space delimited
6.. Browse and select the point file
7.. Select OK
2.. Once the surface is built (this took ~10 minutes on my machine) zoom
extents an
d save the file
3.. Optional step - close Civil 3D to clear out the memory on the system,
then reopen that drawing.
4.. Change to a style that displays you Contours - I used both 25'/125'
and 5'/25; intervals, and was able to get the display to come up in about 30
seconds
5.. Create your alignment
6.. Sample for your profile
7.. Create a Profile View - use a style for the profile view that has the
Minor grids turned
off, and doesn't Clip the grids to the profile. Clipping
takes up some memory, and will cause a delay, and sometime a lock up if the
system runs out of physical memory.


I was able to get a working drawing with these steps. Now we can further
improve performance by taking advantage of the Projects and data
referencing, but for now, I think this should help.





I've attached two screen shots - one of the contours at 5'/25' interval; and
one on th
e profile - Alignment length is 97,977 feet! Drawing size is 121
MB, but it opens it less the 30 seconds.



AG

"Anthony Governanti" wrote in message
news:5255646@discussion.autodesk.com...
I want to test this for myself. Can you send me a text file of the points
gridded at 15'?

anthony dot governanti at autodesk dot com

AG
wrote in message news:5253912@discussion.autodesk.com...
I have a LDT project for 65 miles of new road. Th
e terrain models are from
LIDAR gridded to 15'. The largest tin file is just over 300mb. Can C3D
handle this size file?

Autodesk seems to be hanging it's hat on C3D, so I guess I can't expect much
improvement in stability or new features in LDT. It can handle the large
surface files, but it does have stability problems at times. If C3D can
handle large surfaces, I may give it another try.

[Anonymous]

Back row, right side, ICE in San Rafael. 5'3"+/-, dark hair, and absolutely
stunning in a short black dress!

--
Mike Norton
Total CAD Systems, Inc. - Houston, Texas
"Dana Breig Probert" wrote in message
news:5257806@discussion.autodesk.com...
I'll bet you couldnt pick me out of a lineup.

--
Dana Breig Probert
http://civil3drocks.blogspot.com/
CADapult Ltd
Empowering Design With Innovative Solutions
www.cadapult.net

[Anonymous]

um, maybe you are descripting anthony, cause i was back left, blonde hair,
larger than life. I dont even own a black dress.

--
Dana Breig Probert
http://civil3drocks.blogspot.com/
CADapult Ltd
Empowering Design With Innovative Solutions
www.cadapult.net
----------------------------------------------

[Anonymous]

Well, actually this file has 1.8 million points, but determining the ceiling
wasn't the issue here. In this particular post, the user wanted to get his
surface built from the 55 MB coordinate file, and was having issues with
crashing. And the actual size of this swath is roughly 6000' wide and
100,000' long.

After I experimented, I was able to get it to build without the crash, and
was able to do design work with it by using the techniques I posted. This
is not to say that it'll work for everything, but in this case, I hope it
helps him get his job done.

As far as the future, I can't say anything detailed (as you are already well
aware Mike). Are Lidar data and point clouds becoming more and more common,
yes. Are we aware of this trend, yes. When will we deal with massive
amounts of point data? I can't say, but rest assured we are looking at
solutions.

As far as the 39000+ contours, have you tied to weed the vertices down? Run
Map Cleanup and all the usual tricks? How many acres are you dealing with,
and can you break it down into separate drawings?

AG
1.8GHZ
2.0 GB Ram
128 MB Video (not a certified driver)

"Mike Norton" wrote in message
news:5258321@discussion.autodesk.com...
My math may be a bit off but assuming they shot a 500' swath we're talking
about less than a million points here. I didn't think the ceiling was that
low.

Maybe my machine just isn't up to the task (Dell Precision M70, 2.3 ghz, 2.0
gb RAM) but I just attempted to build a surface from 2' contours (39,000+)
and crashed in just under 40 seconds. (I think with a bit more processor
speed I can get it down to under 30.) Nothing else was running and I even
disabled all options for minimizing flat areas.

Don't get me wrong - I'm not at all really interested in who can come up
with the most extreme dataset but this is a definite problem in the
petroleum industry and increasingly becoming a problem with advancing
"normal" survey technology. Can you share with us the prognosis (if any) for
future releases and/or service packs?

--
Mike Norton
Total CAD Systems, Inc. - Houston, Texas
"Anthony Governanti" wrote in message
news:5257672@discussion.autodesk.com...
Ok folks, I got the LIDAR point file from "akbowbender" and was able to get
it to work pretty well. its a 55 MB text file of XYZ coordinates. here is
what I recommended:

Ok, so I was able to work with the surface and get an alignment and profile
generated.



Here are some best practice suggestions when working with large amounts of
data:

1.. Use the Point File option for building the surface
1.. Create a new surface in a new drawing
2.. Use a Styl
e that just displays the Border (no contours or triangles)
3.. Expand the surface definition, then right click on the Point File
collection
4.. Choose Add.
5.. Select the appropriate file format (if there isn't one that matches
you can define one) - in this case, I used ENZ (XYZ) space delimited
6.. Browse and select the point file
7.. Select OK
2.. Once the surface is built (this took ~10 minutes on my machine) zoom
extents an
d save the file
3.. Optional step - close Civil 3D to clear out the memory on the system,
then reopen that drawing.
4.. Change to a style that displays you Contours - I used both 25'/125'
and 5'/25; intervals, and was able to get the display to come up in about 30
seconds
5.. Create your alignment
6.. Sample for your profile
7.. Create a Profile View - use a style for the profile view that has the
Minor grids turned
off, and doesn't Clip the grids to the profile. Clipping
takes up some memory, and will cause a delay, and sometime a lock up if the
system runs out of physical memory.


I was able to get a working drawing with these steps. Now we can further
improve performance by taking advantage of the Projects and data
referencing, but for now, I think this should help.





I've attached two screen shots - one of the contours at 5'/25' interval; and
one on th
e profile - Alignment length is 97,977 feet! Drawing size is 121
MB, but it opens it less the 30 seconds.



AG

"Anthony Governanti" wrote in message
news:5255646@discussion.autodesk.com...
I want to test this for myself. Can you send me a text file of the points
gridded at 15'?

anthony dot governanti at autodesk dot com

AG
wrote in message news:5253912@discussion.autodesk.com...
I have a LDT project for 65 miles of new road. Th
e terrain models are from
LIDAR gridded to 15'. The largest tin file is just over 300mb. Can C3D
handle this size file?

Autodesk seems to be hanging it's hat on C3D, so I guess I can't expect much
improvement in stability or new features in LDT. It can handle the large
surface files, but it does have stability problems at times. If C3D can
handle large surfaces, I may give it another try.

[Anonymous]

And my black dress was at the cleaners that day.....um....nevermind
"Dana Breig Probert" wrote in message
news:5258550@discussion.autodesk.com...
um, maybe you are descripting anthony, cause i was back left, blonde hair,
larger than life. I dont even own a black dress.

--
Dana Breig Probert
http://civil3drocks.blogspot.com/
CADapult Ltd
Empowering Design With Innovative Solutions
www.cadapult.net
----------------------------------------------

[Anonymous]

Hi Mike,

We are currently quality control testing a program to filter a Point Cloud
based on an offset from an alignment.

At its simplest level of operation it will create one new file with all
points within a user nominated corridor half width.

At its most complex level of operation it will copy the original data file
into 31 segments:

Points not opposite the alignment
10 chainage ranges left of the user nominated corridor half width
10 chainage ranges right of the user nominated corridor half width
10 chainage ranges within of the user nominated corridor half width

This should allow the user only to work with relevant data.

The program should be ready for release to www.civil3Dtools.com within a
week or so.

By working logically you should be able to create data sets which are
relevant to a specific area.
We have run the program on 2,800,000 points over a 140km alignment (time
under 10 min on a midrange computer)
It is currently with a client who has a 30,000,000 point cloud and I'm
awaiting his report.
While under development I used a 100,000 point cloud which was processed in
18 seconds.

The program does not lose any data. We are considering an algorithm for a
"lossy" process which would create DTM suitable for site planning - as
distinct from design where the output of the current program should be used.

The sort of thing I have in mind is to throw the data into a data base and
use SQL queries to extract cells of information and compute the centroid for
each data set.

I would not expect any real development in that field for several months due
to other commitments as CADApps.


--

Laurie Comerford
CADApps
www.cadapps.com.au
"Mike Norton" wrote in message
news:5258321@discussion.autodesk.com...
My math may be a bit off but assuming they shot a 500' swath we're talking
about less than a million points here. I didn't think the ceiling was that
low.

Maybe my machine just isn't up to the task (Dell Precision M70, 2.3 ghz, 2.0
gb RAM) but I just attempted to build a surface from 2' contours (39,000+)
and crashed in just under 40 seconds. (I think with a bit more processor
speed I can get it down to under 30.) Nothing else was running and I even
disabled all options for minimizing flat areas.

Don't get me wrong - I'm not at all really interested in who can come up
with the most extreme dataset but this is a definite problem in the
petroleum industry and increasingly becoming a problem with advancing
"normal" survey technology. Can you share with us the prognosis (if any) for
future releases and/or service packs?

--
Mike Norton
Total CAD Systems, Inc. - Houston, Texas
"Anthony Governanti" wrote in message
news:5257672@discussion.autodesk.com...
Ok folks, I got the LIDAR point file from "akbowbender" and was able to get
it to work pretty well. its a 55 MB text file of XYZ coordinates. here is
what I recommended:

Ok, so I was able to work with the surface and get an alignment and profile
generated.



Here are some best practice suggestions when working with large amounts of
data:

1.. Use the Point File option for building the surface
1.. Create a new surface in a new drawing
2.. Use a Styl
e that just displays the Border (no contours or triangles)
3.. Expand the surface definition, then right click on the Point File
collection
4.. Choose Add.
5.. Select the appropriate file format (if there isn't one that matches
you can define one) - in this case, I used ENZ (XYZ) space delimited
6.. Browse and select the point file
7.. Select OK
2.. Once the surface is built (this took ~10 minutes on my machine) zoom
extents an
d save the file
3.. Optional step - close Civil 3D to clear out the memory on the system,
then reopen that drawing.
4.. Change to a style that displays you Contours - I used both 25'/125'
and 5'/25; intervals, and was able to get the display to come up in about 30
seconds
5.. Create your alignment
6.. Sample for your profile
7.. Create a Profile View - use a style for the profile view that has the
Minor grids turned
off, and doesn't Clip the grids to the profile. Clipping
takes up some memory, and will cause a delay, and sometime a lock up if the
system runs out of physical memory.


I was able to get a working drawing with these steps. Now we can further
improve performance by taking advantage of the Projects and data
referencing, but for now, I think this should help.





I've attached two screen shots - one of the contours at 5'/25' interval; and
one on th
e profile - Alignment length is 97,977 feet! Drawing size is 121
MB, but it opens it less the 30 seconds.



AG

"Anthony Governanti" wrote in message
news:5255646@discussion.autodesk.com...
I want to test this for myself. Can you send me a text file of the points
gridded at 15'?

anthony dot governanti at autodesk dot com

AG
wrote in message news:5253912@discussion.autodesk.com...
I have a LDT project for 65 miles of new road. Th
e terrain models are from
LIDAR gridded to 15'. The largest tin file is just over 300mb. Can C3D
handle this size file?

Autodesk seems to be hanging it's hat on C3D, so I guess I can't expect much
improvement in stability or new features in LDT. It can handle the large
surface files, but it does have stability problems at times. If C3D can
handle large surfaces, I may give it another try.

[Anonymous]

The acceptable simplification varies depending on what the engineer is
needing to design.

For example, in the early planning and pre-planning stages, for grading
concepts couple of feet vertically, and up to 5 feet horizontally on a 1000
acre site may be quite acceptable.

When you're doing rough grading for the same project, your acceptable
accuracy may be 0.5' vertically, and couple of feet horizontally.

In the 'old days' it amounted to the thickness of an unsharpened pencil. On
a 200-scale plan, an unsharpened pencil may amount to 10 feet (one-twentieth
of an inch), on a 40 scale plan this goes down to 2 feet (I'm talking about
horizontal differences).

Vertical is guided little differently, when it comes from aerial surveys.
When an Engineer requests an aerial survey, they usually inform the
photogrammetrist that thy are to be flown for 100-scale drawings. The
photogrammetrists have minimum acceptable accuracy standards for what goes
for 100-scale designs in most civil applications.
You can find a lot of info, and then some here:
http://www.dot.state.co.us/Survey_Manual/Chapter4/Chapter4.pdf#search='minimum%20accuracy%20for%20aerial%20surveys'
AND:
http://www.state.nj.us/transportation/eng/documents/photogrammetry/Section4.htm

USGS HAS IT'S OWN MAP ACCURACY STANDARDS:
http://erg.usgs.gov/isb/pubs/factsheets/fs17199.html

When doing a civil engineering design, you need to think, is this sufficient
to design at the scale I'm working at, and for final construction
documentation, is this the way they are going to build it?
Pads get graded to about a tenth. Sometimes rougher if they're not final
buildable pads but to be sold off and regraded.
Concretework is little more precise, rule of thumb is often 0.03'
When doing highrises, bridges and steelwork applications, you may need to go
more precise.

Weeding contours can be tricky, sometimes it can fall in the area of the
site where a five foot error won't make any difference, and in a different
spot it may get you in trouble, even at preliminary design. For example, if
the site is rather flat, then you have less room for vertical error overall.

When doing a preliminary design, a five foot vertical error may mean that
your planned stormdrain outlet on a relatively flat site goes uphill instead
of downhill, and you only realize later in design, when you get more precise
information that you have no other options, but to put in a sump pump (I've
seen just this happen, due to too much contour trimming)

So, it's a moving target, where a designer really needs to be familiar with
his data and what he needs to do with it to decide how much is too much.




"Dana Breig Probert" wrote in message
news:5257322@discussion.autodesk.com...
I need some feedback on a little experiment I just did.

I took an aerial topo contour file as is vs a weeded contour file...

I am not sure if my results are significant.

http://civil3drocks.blogspot.com/2006/08/how-much-surface-is-too-much.html

--
Dana Breig Probert
http://civil3drocks.blogspot.com/
CADapult Ltd
Empowering Design With Innovative Solutions
www.cadapult.net
----------------------------------------------

[Anonymous]

Another question:

perhaps the topography (contours) are accurate for 100 scale drawings

But we are building TIN from those contours, which in itself is not an
optimal exercise.

And if after weeding, the contours don't change their locations or even have
any noticeable change at all, aren't we still following that 100 scale
drawing accuracy?

Shouldn't out "sharpened pencil" be measured from the actual ground suface
or actual TIN as generated by aerial mapping folks and not the TIN we make
from contours?

If so, how do we do that?

I don't mean to be devil's advocate, i just want to make sure that there
isn't unnecessary data being hauled around.

If I weeded contours and it make ANY visible difference in those contours,
it would be different. But the experiment I did, weeding to 1' didn't even
make the contours look a lick different.

Now the TIN of course had changed. But those 1:100 contours were identical.
Plotted lines nor pencil lines have vertices.

--
Dana Breig Probert
http://civil3drocks.blogspot.com/
CADapult Ltd
Empowering Design With Innovative Solutions
www.cadapult.net
----------------------------------------------

[Anonymous]

Ah, Anthony, you know I was just fishin'. I'll be in Manchester in two weeks
so I'm sure I'll find out more than I need to know. (Curse you, NDA, curse
you!)

I've done a blind cleanup but to no avail. "Blind" meaning contour weeding
and drawing cleanup with no regard to maintaining contour accuracy. I'm OK
with drawing cleanup but weeding is an art I've yet to master.

I'm definitely interested in the program Laurie mentioned, especially for my
pipeline groups. We resorted to a similar program on the last pipeline
project I worked on to create a 1000-meter swath of sonar points. We still
struggled but it did make the difference between impossible and submitted.

--
Mike Norton
Total CAD Systems, Inc. - Houston, Texas
"Anthony Governanti" wrote in message
news:5258544@discussion.autodesk.com...
Well, actually this file has 1.8 million points, but determining the ceiling
wasn't the issue here. In this particular post, the user wanted to get his
surface built from the 55 MB coordinate file, and was having issues with
crashing. And the actual size of this swath is roughly 6000' wide and
100,000' long.

After I experimented, I was able to get it to build without the crash, and
was able to do design work with it by using the techniques I posted. This
is not to say that it'll work for everything, but in this case, I hope it
helps him get his job done.

As far as the future, I can't say anything detailed (as you are already well
aware Mike). Are Lidar data and point clouds becoming more and more common,
yes. Are we aware of this trend, yes. When will we deal with massive
amounts of point data? I can't say, but rest assured we are looking at
solutions.

As far as the 39000+ contours, have you tied to weed the vertices down? Run
Map Cleanup and all the usual tricks? How many acres are you dealing with,
and can you break it down into separate drawings?

AG
1.8GHZ
2.0 GB Ram
128 MB Video (not a certified driver)

"Mike Norton" wrote in message
news:5258321@discussion.autodesk.com...
My math may be a bit off but assuming they shot a 500' swath we're talking
about less than a million points here. I didn't think the ceiling was that
low.

Maybe my machine just isn't up to the task (Dell Precision M70, 2.3 ghz, 2.0
gb RAM) but I just attempted to build a surface from 2' contours (39,000+)
and crashed in just under 40 seconds. (I think with a bit more processor
speed I can get it down to under 30.) Nothing else was running and I even
disabled all options for minimizing flat areas.

Don't get me wrong - I'm not at all really interested in who can come up
with the most extreme dataset but this is a definite problem in the
petroleum industry and increasingly becoming a problem with advancing
"normal" survey technology. Can you share with us the prognosis (if any) for
future releases and/or service packs?

--
Mike Norton
Total CAD Systems, Inc. - Houston, Texas
"Anthony Governanti" wrote in message
news:5257672@discussion.autodesk.com...
Ok folks, I got the LIDAR point file from "akbowbender" and was able to get
it to work pretty well. its a 55 MB text file of XYZ coordinates. here is
what I recommended:

Ok, so I was able to work with the surface and get an alignment and profile
generated.



Here are some best practice suggestions when working with large amounts of
data:

1.. Use the Point File option for building the surface
1.. Create a new surface in a new drawing
2.. Use a Styl
e that just displays the Border (no contours or triangles)
3.. Expand the surface definition, then right click on the Point File
collection
4.. Choose Add.
5.. Select the appropriate file format (if there isn't one that matches
you can define one) - in this case, I used ENZ (XYZ) space delimited
6.. Browse and select the point file
7.. Select OK
2.. Once the surface is built (this took ~10 minutes on my machine) zoom
extents an
d save the file
3.. Optional step - close Civil 3D to clear out the memory on the system,
then reopen that drawing.
4.. Change to a style that displays you Contours - I used both 25'/125'
and 5'/25; intervals, and was able to get the display to come up in about 30
seconds
5.. Create your alignment
6.. Sample for your profile
7.. Create a Profile View - use a style for the profile view that has the
Minor grids turned
off, and doesn't Clip the grids to the profile. Clipping
takes up some memory, and will cause a delay, and sometime a lock up if the
system runs out of physical memory.


I was able to get a working drawing with these steps. Now we can further
improve performance by taking advantage of the Projects and data
referencing, but for now, I think this should help.





I've attached two screen shots - one of the contours at 5'/25' interval; and
one on th
e profile - Alignment length is 97,977 feet! Drawing size is 121
MB, but it opens it less the 30 seconds.



AG

"Anthony Governanti" wrote in message
news:5255646@discussion.autodesk.com...
I want to test this for myself. Can you send me a text file of the points
gridded at 15'?

anthony dot governanti at autodesk dot com

AG
wrote in message news:5253912@discussion.autodesk.com...
I have a LDT project for 65 miles of new road. Th
e terrain models are from
LIDAR gridded to 15'. The largest tin file is just over 300mb. Can C3D
handle this size file?

Autodesk seems to be hanging it's hat on C3D, so I guess I can't expect much
improvement in stability or new features in LDT. It can handle the large
surface files, but it does have stability problems at times. If C3D can
handle large surfaces, I may give it another try.

[Anonymous]

You've got my attention, Laurie. The program sounds ideal for my LIDAR and
SONAR shoots. My pipeline guys rarely need surface data beyond a relatively
narrow swath.

I work with a group here shooting point clouds to create as-built drawings
for offshore facilities to facilitate new construction. Do the point clouds
you mention include co-planar points.

--
Mike Norton
Total CAD Systems, Inc. - Houston, Texas
"Laurie Comerford" wrote in message
news:5258556@discussion.autodesk.com...
Hi Mike,

We are currently quality control testing a program to filter a Point Cloud
based on an offset from an alignment.

At its simplest level of operation it will create one new file with all
points within a user nominated corridor half width.

At its most complex level of operation it will copy the original data file
into 31 segments:

Points not opposite the alignment
10 chainage ranges left of the user nominated corridor half width
10 chainage ranges right of the user nominated corridor half width
10 chainage ranges within of the user nominated corridor half width

This should allow the user only to work with relevant data.

The program should be ready for release to www.civil3Dtools.com within a
week or so.

By working logically you should be able to create data sets which are
relevant to a specific area.
We have run the program on 2,800,000 points over a 140km alignment (time
under 10 min on a midrange computer)
It is currently with a client who has a 30,000,000 point cloud and I'm
awaiting his report.
While under development I used a 100,000 point cloud which was processed in
18 seconds.

The program does not lose any data. We are considering an algorithm for a
"lossy" process which would create DTM suitable for site planning - as
distinct from design where the output of the current program should be used.

The sort of thing I have in mind is to throw the data into a data base and
use SQL queries to extract cells of information and compute the centroid for
each data set.

I would not expect any real development in that field for several months due
to other commitments as CADApps.


--

Laurie Comerford
CADApps
www.cadapps.com.au
"Mike Norton" wrote in message
news:5258321@discussion.autodesk.com...
My math may be a bit off but assuming they shot a 500' swath we're talking
about less than a million points here. I didn't think the ceiling was that
low.

Maybe my machine just isn't up to the task (Dell Precision M70, 2.3 ghz, 2.0
gb RAM) but I just attempted to build a surface from 2' contours (39,000+)
and crashed in just under 40 seconds. (I think with a bit more processor
speed I can get it down to under 30.) Nothing else was running and I even
disabled all options for minimizing flat areas.

Don't get me wrong - I'm not at all really interested in who can come up
with the most extreme dataset but this is a definite problem in the
petroleum industry and increasingly becoming a problem with advancing
"normal" survey technology. Can you share with us the prognosis (if any) for
future releases and/or service packs?

--
Mike Norton
Total CAD Systems, Inc. - Houston, Texas
"Anthony Governanti" wrote in message
news:5257672@discussion.autodesk.com...
Ok folks, I got the LIDAR point file from "akbowbender" and was able to get
it to work pretty well. its a 55 MB text file of XYZ coordinates. here is
what I recommended:

Ok, so I was able to work with the surface and get an alignment and profile
generated.



Here are some best practice suggestions when working with large amounts of
data:

1.. Use the Point File option for building the surface
1.. Create a new surface in a new drawing
2.. Use a Styl
e that just displays the Border (no contours or triangles)
3.. Expand the surface definition, then right click on the Point File
collection
4.. Choose Add.
5.. Select the appropriate file format (if there isn't one that matches
you can define one) - in this case, I used ENZ (XYZ) space delimited
6.. Browse and select the point file
7.. Select OK
2.. Once the surface is built (this took ~10 minutes on my machine) zoom
extents an
d save the file
3.. Optional step - close Civil 3D to clear out the memory on the system,
then reopen that drawing.
4.. Change to a style that displays you Contours - I used both 25'/125'
and 5'/25; intervals, and was able to get the display to come up in about 30
seconds
5.. Create your alignment
6.. Sample for your profile
7.. Create a Profile View - use a style for the profile view that has the
Minor grids turned
off, and doesn't Clip the grids to the profile. Clipping
takes up some memory, and will cause a delay, and sometime a lock up if the
system runs out of physical memory.


I was able to get a working drawing with these steps. Now we can further
improve performance by taking advantage of the Projects and data
referencing, but for now, I think this should help.





I've attached two screen shots - one of the contours at 5'/25' interval; and
one on th
e profile - Alignment length is 97,977 feet! Drawing size is 121
MB, but it opens it less the 30 seconds.



AG

"Anthony Governanti" wrote in message
news:5255646@discussion.autodesk.com...
I want to test this for myself. Can you send me a text file of the points
gridded at 15'?

anthony dot governanti at autodesk dot com

AG
wrote in message news:5253912@discussion.autodesk.com...
I have a LDT project for 65 miles of new road. Th
e terrain models are from
LIDAR gridded to 15'. The largest tin file is just over 300mb. Can C3D
handle this size file?

Autodesk seems to be hanging it's hat on C3D, so I guess I can't expect much
improvement in stability or new features in LDT. It can handle the large
surface files, but it does have stability problems at times. If C3D can
handle large surfaces, I may give it another try.

[lctilley]

I would be very interested in what you have done to break up your site into more manageable areas. We have a 215 lot subdivision with a pretty large surface and it has given us fits from beginning to, well I was going to say end, but we are not quiet there yet. I was considering breaking apart the surface for just the areas of the streets and infrastructure. Maybe even a separate small surface for each street even. Streets are approximately 12 to 18 feet ling with the entire development have 6.

How are you handling it?

Also, can anyone tell me if in 2007 they have allowed us to reference 2 or more alignments with structures, for labeling purposes at least?

[Anonymous]

"Dana Breig Probert" wrote in message
news:5258662@discussion.autodesk.com...
Another question:

perhaps the topography (contours) are accurate for 100 scale drawings
But we are building TIN from those contours, which in itself is not an
optimal exercise.
And if after weeding, the contours don't change their locations or even have
any noticeable change at all, aren't we still following that 100 scale
drawing accuracy?
-------------------------
What happens is that the 2nd generation contours we created are as accurate
as originals in some areas, and little less accurate in oter areas. Just
depends on how we go about weeding.
I've seen dozens of weeded contours and topos, and weeded dozens myself.
One of more common 'overweeding' issues I've seen is when ridges or drainage
features end up getting chopped off. After weeding, and re-tinning, I've
seen some of those feature disappear almost completely.

I wish tere was selective weeding. I haven't tried it myself yet, ideally, I
would like to have a broad brus pen, much like a masking tool in photoshop
where I can quickly isolate parts of contours where my weeding factors will
be more conservative, where I know I will have edge conditions, or
anticipate drainage connections or something else of little mopre
significance.
Then there are other areas of a site where I would gladly apply much higher
weeding factors, since from experience I know it won't affect me. (Come to
think of it, I should give it some more organized thought and writ a little
paper on this. There's so many different what if's that come to mind I'd
like to elaboprate on at the moment, I'm not sure which to turn to).

There's areas of uniform slopes where I they could be reasonably
approximated with 3 planes defined with 10 ponts total, but due to the way
contours are, there are 1000+ points defining them.
------------------------
Shouldn't out "sharpened pencil" be measured from the actual ground suface
or actual TIN as generated by aerial mapping folks and not the TIN we make
from contours?
If so, how do we do that?
---------------------------------------
Sorry, I probably didn't make myself all that clear. The pencil rule of
thumb is to be applied to acceptable deviations after the original aerial
topo was flown.

It comes from the old days of doing design on paper. If you're doing a
conceptual grading design on a subdivision of some 300 acres, it is most
likely to fit on one 30"x42" sheet of paper, at about 1"=100' scale. You'll
sketch out your design with a pencil... Contours are very likely to be at 5'
for minors, and 25' for majors.
Take 0.5 or a 0.7mm pencil, put it next to the engineering scale, at
1"=100', what's the pencil with in feet (on that map), about 3-5 feet.

Add to that a bit of fuzzy vision or some paper stretch, and you have an
anticipated fudge factor of about 10 feet on 100-scale drawings. Much more
then that, in the hand drafting days may have been considered more on the
sloppy side.
The bigger the fudge factor, the more chances you have to paint yourself in
a corner down the road.
---------------------------------------
I don't mean to be devil's advocate, i just want to make sure that there
isn't unnecessary data being hauled around.

If I weeded contours and it make ANY visible difference in those contours,
it would be different. But the experiment I did, weeding to 1' didn't even
make the contours look a lick different.

Now the TIN of course had changed. But those 1:100 contours were identical.
Plotted lines nor pencil lines have vertices.
--------------------------------------

Exactly!
Now that the CAD is our pencil and out calculator, the way to figure out
what's an acceptable deviation is going to be little different. However, the
old style rules of thumb can pretty easily converted to CAD numbers if we
ask a few questions before weeding.

For example, the fact that the topo is available in 1-foot contours tells me
they were most likely flown with the intention of 40-scale final design.
But, to thrown in the monkey wrench, I've noticed that more and more often
1-foot contours end up being requested long before they are actually needed.

So, first question to ask is, what is the stage of the project, are we in
very preliminary stages, are we doing conceptual design, tentative map or
preliminary engineering, or preparing final construction documentation.

If I were given 1 foot contours and told we're doing preliminary design on a
500 acre site, on maps that will be plotted at 1"=200 scale, I'd drop the
1-footers, and turn them into 5' and 25's (saving into a new drawing,
preserving an original) before doing much of anything else.

Then I'd look at the topo and se if I need to draw in any significant
breaklines.

Then I apply the weeding factors, using the pencil rule of thumb. I like to
use the distance/angle based weeding. I run it several times, increasing the
factors, and examining the potential problem areas after each run. When I
start pushing the limit of the pencil thickness deviations, I stop weeding.
On something that will be plotted at 1"=200' scale, that would mean I may be
comfortable with a horizontal movement of 7-10 feet in most extreme places.
I might accept little more on some sharp turns in the middle of nowhere.

The reason I use the rule of pencil is that now that the design is done
numerically, the accuracy we'd lose by scaling and measuring off the paper
product is gone, and it can be applied to simplifying the contour data to
reduce the processing overhead.

There's other considerations too. You have more room for slop if you're
laying out a road in open undeveloped land, vs. trying to fit it through an
urban populated area, or running along another structure.

[Anonymous]

I appreciate the insight!! I am so glad to be finally getting my head
around this- and I know many of my clients feel the same way.

I think I am going to try an experiement....

I am going to take a really great tin and contour it at a tight interval.

Then explode it, and rebuild the surface from those contours

And do a comparision

I think that is the real issue that I can't get my head around is- are we
really building an accurate surface from contours- even with a million
vertices?

--
Dana Breig Probert
http://civil3drocks.blogspot.com/
CADapult Ltd
Empowering Design With Innovative Solutions
www.cadapult.net

[Anonymous]

Well, I'm glad it helped some.
I wasn't sure if I was being helpful or not.
I figure I'd share the thought process I tend to go through, and you pick
and choose what makes sense... or look at it and thing, gawd she's being
full of herself 😉

As to are we building an accurate surface with a million points.... I should
bring your attention to the precision vs. accuracy mind-bender.
A surface with a million points may be very precise, but may not be accurate
at all... Just depends. I'm not sure if this is taught in engineering
classes. In surveying schools, it's drilled into you a lot, we spent two
semesters on error theory, accuracy, precision and significant figures
(google search hint... precision, accuracy, error theory, significant
figures)... It's a question that seems so simple on the surface (pun not
intended), but can boggle the mind.

Here's a few links you might find interesting, which explore the concept:
http://honolulu.hawaii.edu/distance/sci122/SciLab/L5/accprec.html
http://en.wikipedia.org/wiki/Accuracy
http://mathforum.org/library/drmath/view/61947.html

http://www.cs.cmu.edu/afs/cs/project/anim/ph/paper/multi97/release/heckbert/simp_sl.pdf#search='surface%20simplification%20algorithms'
http://graphics.cs.uiuc.edu/~garland/papers/simp.pdf#search='surface%20simplification%20algorithms'
http://www.cs.cmu.edu/afs/cs/project/pscico/pscico/src/surfsimplify/README.html

Actually, if you look at the gaming industry, they've advanced more then
engineers in the surface simplification efforts without wiping out the
significant feature, and automating the process... I like to take a peak and
see what they're up to on occasion...

I think you're a smart enough cookie to draw your own conclusions 😄

[Anonymous]

Actually, this may be a good item for a C3D wish list, for, I dunno, 2015
release...
Adjustable level of detail surfaces.
http://csdl2.computer.org/persagen/DLAbsToc.jsp?resourcePath=/dl/mags/cg/&toc=comp/mags/cg/2001/03/g3toc.xml&DOI=10.1109/38.920624

How come video games can have it, and we can't? 😞

"Dana Breig Probert" wrote in message
news:5259551@discussion.autodesk.com...
I appreciate the insight!! I am so glad to be finally getting my head
around this- and I know many of my clients feel the same way.

I think I am going to try an experiement....

I am going to take a really great tin and contour it at a tight interval.

Then explode it, and rebuild the surface from those contours

And do a comparision

I think that is the real issue that I can't get my head around is- are we
really building an accurate surface from contours- even with a million
vertices?

--
Dana Breig Probert
http://civil3drocks.blogspot.com/
CADapult Ltd
Empowering Design With Innovative Solutions
www.cadapult.net

[Anonymous]

It is the accuracy vs. precision thing exactly that is killing me on this
issue.

I often feel like we are being overly precise with inaccurate data.

Thanks again. I am know a lot of people are probably bored to tears by this
thread, but it has helped me huge. Your resources are great. I have been
through all of my site textbooks and found nothing nearly as satisfying as
the links you posted.

Do you have any surveying topography texts that you can recommend?

--
Dana Breig Probert
http://civil3drocks.blogspot.com/
CADapult Ltd
Empowering Design With Innovative Solutions
www.cadapult.net
----------------------------------------------

[Anonymous]

Well, I'll give you few of my favorite links for survey and mapping and
related, you can browse through and pick and choose what you find
applicable...

http://www.acsm.net/
http://www.acsm.net/alta.html
http://www.acsm.net/cgi-local/SoftCart.exe/online-store/scstore/p-C171.html?L+scstore+cccg4265ff45be45+1161015915
http://en.wikipedia.org/wiki/Photogrammetry
http://www.asprs.org/fall2006/index.htm
http://rockyweb.cr.usgs.gov/nmpstds/acrodocs/nmas/NMAS647.PDF

This paper contains the California department of transportation mapping
accuracy standards. this means that everyone that does work for them needs
to adhere to it. Many of California local agencies (cities) have adopted
very similar standards, and is what us 'consultant folks' get to live by if
we want our projects to get permitted for building.
http://www.dot.ca.gov/hq/esc/geometronics/LSITWorkbook/09.pdf#search='caltrans%20contour%20accuracy%20standards'

Lot of local agencies like county surveyors and similar have the liberty to
set those standards. It's not a nationally uniform standard, but to my
knowledge, it doesn't vary a whole lot. The American congress on surveying
and mapping and American Society for Photogrammetry and Remote Sensing are
the entities that are very involved in setting those standards, along with
USGS and NGS, and similar governmental agencies.

http://www.richlandmaps.com/redirect.asp?htmlpage=www.richlandmaps.com/standards/gps.html

Heck, here's a link to my yahoo searches: Links posted above are just the
most known entities (to me) that come up in the searches.
http://search.yahoo.com/search?p=photogrammetry+accuracy+standards&ei=UTF-8&fr=FP-tab-web-t400&x=wrt
http://search.yahoo.com/search?p=national+mapping+accuracy+standards&ei=UTF-8&fr=FP-tab-web-t400&x=wrt
http://search.yahoo.com/search?p=national+survey+and+mapping+accuracy+standards&ei=UTF-8&fr=FP-tab-web-t400&x=wrt
http://search.yahoo.com/search?p=contour+accuracy+standards&ei=UTF-8&fr=FP-tab-web-t400&x=wrt
http://search.yahoo.com/search?p=caltrans+contour+accuracy+standards&ei=UTF-8&fr=FP-tab-web-t400&x=wrt
http://search.yahoo.com/search?p=county+contour+accuracy+standards&ei=UTF-8&fr=FP-tab-web-t400&x=wrt

This is a fun one, you may know about it already:
http://www.vterrain.org/


"Dana Breig Probert" wrote in message
news:5259592@discussion.autodesk.com...
It is the accuracy vs. precision thing exactly that is killing me on this
issue.

I often feel like we are being overly precise with inaccurate data.

Thanks again. I am know a lot of people are probably bored to tears by this
thread, but it has helped me huge. Your resources are great. I have been
through all of my site textbooks and found nothing nearly as satisfying as
the links you posted.

Do you have any surveying topography texts that you can recommend?

--
Dana Breig Probert
http://civil3drocks.blogspot.com/
CADapult Ltd
Empowering Design With Innovative Solutions
www.cadapult.net
----------------------------------------------

[Anonymous]

Hi Dana,

The easiest way to understand where you can go wrong with surfaces built
from contours is to draw a simple section using parameters like:

Distance Level
0 20.001
20 19.10
20.1 8.0

Plot the location of where a 10 contour and a 20 contour would occur on this
section and compare the location of level 19.9 predicted by the section and
the location predicted by contour interpolation.

This is somewhat extreme compared with likely events in real life models,
but you will see how the errors arise.

--

Laurie Comerford
CADApps
www.cadapps.com.au

"Dana Breig Probert" wrote in message
news:5259551@discussion.autodesk.com...
I appreciate the insight!! I am so glad to be finally getting my head
around this- and I know many of my clients feel the same way.

I think I am going to try an experiement....

I am going to take a really great tin and contour it at a tight interval.

Then explode it, and rebuild the surface from those contours

And do a comparision

I think that is the real issue that I can't get my head around is- are we
really building an accurate surface from contours- even with a million
vertices?

--
Dana Breig Probert
http://civil3drocks.blogspot.com/
CADapult Ltd
Empowering Design With Innovative Solutions
www.cadapult.net

[Anonymous]

Picture please? I think I missed a step, Laurie.

--
James Wedding, P.E.
Engineered Efficiency, Inc.
Civil 3D 2007
XP Tablet, SP2, 2GHz, 2G
www.eng-eff.com
www.civil3d.com