I am using gdal_translate convert USGS DEM data to geotiff ,and
generate .ive file with a geotiff texture via osgdem.But I see a flat
terrain with the texture not a real terrain.

   neke

Hi Lluís,

I spent some time looking at your building data.

For the most part, it is slow simply because it is very detailed, hence a
high polygon count.  For example, when i tried the "Barri Vell" data, it is
2860 buildings.  In Enviro, when all the buildings are in view, i get around
20 fps.  When walking around the town, only some buildings are in view, so
it is 30 fps.  For larger datasets it gets slower:

"VAlegre_carme_Pedreres", 2970 buildings, 20 fps
"Eixample fins tren", 8292 buildings, 5 fps
"Girona sud", 8631 buildings, 4.3 fps

However, i did see one fixable thing that contributes to the slowness:
rounded building corners.  Whereas a simple building has around 30
triangles, the ones with rounded corners are _very_ rounded, with 100+
sides, which means 500+ triangles.  If all the buildings were simple, then
8000+ buildings would be around 250000 triangles, but the smooth corners are
probably putting this a fair bit higher.

It would also be possible to make the buildings even more simple - 1 story
level instead of 2, which should improve fps by at least 50%.

Regarding the river, the best thing to do is probably to draw it into the
base texture, as the VTP doesn't do dynamic bodies of water (yet).

Hope that helps,
Ben

> -----Original Message-----
> From: Lluís Vicens [mailto:lluis@...]
> Sent: Tuesday, November 29, 2005 11:37 PM
> To: Ben Discoe
> Subject: Re: [vtp] Problems when running Enviro
>
> Hi Ben,
>
> Ok, as you said, I'll try to compress the vtst files and look
> for what's going on. In any case, I've put up the files into
> the file area of the mailing list (BuildingData.zip). I'll be
> so grateful if you can check this files and express your
> opinion or ideas about how to solve the problem.

> [..] as you can see, the city is crossed by a river
> but I didn't found a solution to represent the river with VTP.

> -----
> From: nekewong
> Sent: Thursday, December 01, 2005 7:48 PM
>
>    I am using gdal_translate convert USGS DEM data to geotiff
> ,and generate .ive file with a geotiff texture via osgdem.But
> I see a flat terrain with the texture not a real terrain.

Neke,

You might want to ask on the GDAL or OSG mailing lists.  The VTP reads USGS
DEM directly, so it has no need of gdal_translate to do that.  We also don't
use osgdem at all.  Finally, converting DEM to geotiff is a poor idea since
geotiff was not designed to contain elevation.

-Ben

Hi folks,

NASA has release a new version of SRTM, version 2, which has the water area
flattened to solve the noise problem.  The voids (gaps) remain.
http://www2.jpl.nasa.gov/srtm/ has more info.

It is quite a tedious process of identifying and downloading the 1-degree
tiles from the NASA website, then merging and gap-filling to cover a desired
area.  However, this is a _very_ popular use for the VTP software, driving a
lot of interest in VTBuilder for this purpose.

I have recently been working with the nice folks at geoTorrent.org, with the
idea of distributing SRTM via bittorrent which has already been gap-filled.
Bittorrent should make sense, because the size of data is huge, and the
number of people interested is large.

We were going to launch with a gap-filled copy of the Africa tiles (as
independent 1-degree .hgt.gz files, but a single torrent), when SRTM V2 was
released, causing me to rethink the approach.

The default groupings of the SRTM datasets are massive: Africa, Australia,
Eurasia, Islands, North_America, South_America.  Consider the common case of
a user who wants a single country, e.g. Pakistan.  It would not be a good
idea to require them to download the entire Eurasia.torrent (14.8 GB
uncompressed, around 6 GB zipped) just to get Pakistan from it.

In other words, the 1-degree tiles are too small, and the continents are too
large.

Perhaps the most useful thing to do would be to generate a merged,
gap-filled elevation dataset for each country in the world, then make each
of those a torrent, tracked by geoTorrent.org.  I could do that with
VTBuilder fairly easily.  The torrents themselves could either contain a
single BT file, or left un-merged as a sparse array of gap-filled 1-degree
.hgt files.

Input, ideas?

-Ben

On 12/3/05, Ben Discoe <ben@...> wrote:
> Hi folks,
>
> NASA has release a new version of SRTM, version 2, which has the water area
> flattened to solve the noise problem.  The voids (gaps) remain.
> http://www2.jpl.nasa.gov/srtm/ has more info.
>
> It is quite a tedious process of identifying and downloading the 1-degree
> tiles from the NASA website, then merging and gap-filling to cover a desired
> area.  However, this is a _very_ popular use for the VTP software, driving a
> lot of interest in VTBuilder for this purpose.
>
> I have recently been working with the nice folks at geoTorrent.org, with the
> idea of distributing SRTM via bittorrent which has already been gap-filled.
> Bittorrent should make sense, because the size of data is huge, and the
> number of people interested is large.
>
> We were going to launch with a gap-filled copy of the Africa tiles (as
> independent 1-degree .hgt.gz files, but a single torrent), when SRTM V2 was
> released, causing me to rethink the approach.
>
> The default groupings of the SRTM datasets are massive: Africa, Australia,
> Eurasia, Islands, North_America, South_America.  Consider the common case of
> a user who wants a single country, e.g. Pakistan.  It would not be a good
> idea to require them to download the entire Eurasia.torrent (14.8 GB
> uncompressed, around 6 GB zipped) just to get Pakistan from it.
>
> In other words, the 1-degree tiles are too small, and the continents are too
> large.
>
> Perhaps the most useful thing to do would be to generate a merged,
> gap-filled elevation dataset for each country in the world, then make each
> of those a torrent, tracked by geoTorrent.org.  I could do that with
> VTBuilder fairly easily.  The torrents themselves could either contain a
> single BT file, or left un-merged as a sparse array of gap-filled 1-degree
> .hgt files.
>
> Input, ideas?

I think tiles much smaller than a single country tile would be
friendly to most peoples download limits and allows us to pick local
areas of interest.

chris
>
> -Ben
>
>
>
>
> Yahoo! Groups Links
>
>
>
>
>
>
>

> -----
> From: chris
> Sent: Saturday, December 03, 2005 10:31 PM
> To: vtp@yahoogroups.com
>
> On 12/3/05, Ben Discoe <ben@...> wrote:
> > In other words, the 1-degree tiles are too small, and the
> > continents are too large.
> >
> > Perhaps the most useful thing to do would be to generate a merged,
> > gap-filled elevation dataset for each country in the world [..]
>
> I think tiles much smaller than a single country tile would
> be friendly to most peoples download limits and allows us to
> pick local areas of interest.

Would you propose some tile size larger than 1x1 degree, but smaller than a
country?  How large would this 'local area of interest' be?  The user might
then face the challenge of determining which tiles they need in order to
cover their desired area, and manual downloading of each tile and merging
them.  It would also reduce the distributed bandwidth advantages of
bittorrent, since the greater number of individual files, the less chance of
multiple downloaders of a single torrent.

-Ben

Quite frankly, Ben, I do not see the need to redistribute the data in
any form, appreciated though your efforts are, as always!

The 1 degreee tiles are between 45kb and 1.4Mb in size, which is not
excessive, and download speeds from NASA are acceptable. VTP has the
ability to fill voids on import, and as Chris stated, most users are
only going to want a local area of interest most of the time anyway.

I personally keep void-filled copies of the SRTMv1 data for the whole of
the UK on my local drive at all times, for use in clinet projects, where
it forms the bulk of data that I use for mid-distance "filler" terrain.
I usually resample it to 100m or lower res anyhow, and it takes minimal
time to load the tiles I want into an appropriate program (Global
Mapper, Manifold, VTBuilder, or whatever) and export a single tile of
the appropriate size.

Whilst I applaud the theory and intent behind the redistribution of
VTP-ready versions of the data via bitorrent, I am not entirely sure
that it isn't making things TOO easy for users! ;)

Happy to admit that this is just my personal take on things, but I
learned how to utilise the software I had to access the portions of SRTM
data that I need for a particular project, with the minimum of effort
and stress. Understanding how to use VTBuilder efficiently is one thing
that unprocessed SRTM data allows you to do - I think most users of the
VTP software could benefit from having to perform the void filling and
tile selection themselves.

That said, if there is anything I can do to help in your processing of
the datasets, I will happily do so. I have copies of all the SRTMv2 data
here locally and would be happy to batch process datasets to reduce
processing time on your part. Please feel free to contact me offlist
with your requirements should you choose to proceed with the data
redistribution.

Best Regards

Adam

> -----Original Message-----
> From: vtp@yahoogroups.com [mailto:vtp@yahoogroups.com] On
> Behalf Of Ben Discoe
> Sent: 04 December 2005 09:00
> To: vtp@yahoogroups.com
> Subject: RE: [vtp] How should SRTM V2 be distributed?
>
>
> > -----
> > From: chris
> > Sent: Saturday, December 03, 2005 10:31 PM
> > To: vtp@yahoogroups.com
> >
> > On 12/3/05, Ben Discoe <ben@...> wrote:
> > > In other words, the 1-degree tiles are too small, and the
> > > continents are too large.
> > >
> > > Perhaps the most useful thing to do would be to generate a merged,
> > > gap-filled elevation dataset for each country in the world [..]
> >
> > I think tiles much smaller than a single country tile would
> > be friendly to most peoples download limits and allows us to
> > pick local areas of interest.
>
> Would you propose some tile size larger than 1x1 degree, but
> smaller than a country?  How large would this 'local area of
> interest' be?  The user might then face the challenge of
> determining which tiles they need in order to cover their
> desired area, and manual downloading of each tile and merging
> them.  It would also reduce the distributed bandwidth
> advantages of bittorrent, since the greater number of
> individual files, the less chance of multiple downloaders of
> a single torrent.
>
> -Ben
>
>
>
>
> Yahoo! Groups Links
>
>
>
>
>
>
>
>

Davide,

From the correspondence i've received, tile per country is definitely a
popular choice.  Leaving smaller countries inside the tiles for larger
countries, as you say, definitely seems to be a good idea, as well as
splitting larger countries. Perhaps Italy would have a North and South.

Additionally, perhaps it would be good to put all the gap-fixed 1x1 degree
tiles on a server, then for each country, simply make a web page with links
to that country's tiles, perhaps via an image map showing the tiles as a
grid on a map of the country.

Yet another idea would be to add a feature to VTBuilder which lets the user
set the Area Tool to an area of interest, then VTBuilder will automatically
download and merge all the SRTM tiles to produce a single elevation layer
for that area.

Ultimately, serving SRTM V2 is perhaps best done by an OGC WCS (Web Coverage
Service), however i don't know of anyone who has done it yet.

Regarding gap filling, to the best of my knowledge SRTMFill and VTBuilder do
the same simple thing: each unknown heixel is set to the average of the
surrounding known heixels, repeatedly until there are no more unknowns.
Guessing unknown data is the kind of problem which can be addressed by more
and more complex approaches, but they are not necessarily any more accurate
results.

-Ben

> -----Original Message-----
> From: davide_borelli
> Sent: Sunday, December 04, 2005 1:11 AM
>
> --- In vtp@yahoogroups.com, "Ben Discoe" <ben@v...> wrote:
> > How large would this 'local area of interest' be?  The
> > user might then face the challenge of determining which
> > tiles they need in order to cover their desired area, and
> > manual downloading of each tile and merging them.  It would
> > also reduce the distributed bandwidth advantages of
> > bittorrent [..]
>
> Well, the first thing I did when I saw the SRTM V2 was
> starting to make the gap-filled elevation dataset for Italy.
> I think that a tile for every country should be the best
> idea. Maybe it's possible to split larger countries (for
> example China, Brasil, etc...) in two or more tiles, and
> include smaller ones in the tile of the nearest "big"
> country (I think that tiles for Vatican City, San Marino
> should be in the "Italy tile", Andorra in the "Spain Tile",
> Liechtenstein in the "Switzerland Tile" and so on).
>
> Another thing is: which is the best option to fill the gaps,
> i mean the value in the "Set unknown areas" function of
> VTBuilder? Or should I use SRTMFill?
>
> Thanks
> Davide

In my experience, far better compression ratios can be achieved with software
like WinRar or 7-Zip than using standard  ZIP/GZIP files. Both support multiple
operating systems. WinRar has an easier to use interface but 7-zip offers
superior compression.

http://www.rarlabs.com/download.htm
http://www.7-zip.org/download.html

> Additionally, perhaps it would be good to put all the gap-fixed 1x1 degree
> tiles on a server, then for each country, simply make a web page with
> links
> to that country's tiles, perhaps via an image map showing the tiles as a
> grid on a map of the country.
>
> Yet another idea would be to add a feature to VTBuilder which lets the
> user
> set the Area Tool to an area of interest, then VTBuilder will
> automatically
> download and merge all the SRTM tiles to produce a single elevation layer
> for that area.
>
> Ultimately, serving SRTM V2 is perhaps best done by an OGC WCS (Web
> Coverage
> Service), however i don't know of anyone who has done it yet.

Ben, all,

CGIAR (http://srtm.csi.cgiar.org/SELECTION/inputCoord.asp) has a nice
interface for selecting&downloading SRTM data.

Vasile

The CGIAR site has a good interface but the download speeds I'm getting are
about 3 k/sec and I have a T-1. Torrent files would make this go a lot faster!


-----Original Message-----
From: vtp@yahoogroups.com [mailto:vtp@yahoogroups.com]On Behalf Of
Vasile Craciunescu
Sent: Monday, December 05, 2005 11:13 AM
To: vtp@yahoogroups.com
Subject: Re: [vtp] RE: How should SRTM V2 be distributed?


> Additionally, perhaps it would be good to put all the gap-fixed 1x1 degree
> tiles on a server, then for each country, simply make a web page with
> links
> to that country's tiles, perhaps via an image map showing the tiles as a
> grid on a map of the country.
>
> Yet another idea would be to add a feature to VTBuilder which lets the
> user
> set the Area Tool to an area of interest, then VTBuilder will
> automatically
> download and merge all the SRTM tiles to produce a single elevation layer
> for that area.
>
> Ultimately, serving SRTM V2 is perhaps best done by an OGC WCS (Web
> Coverage
> Service), however i don't know of anyone who has done it yet.

Ben, all,

CGIAR (http://srtm.csi.cgiar.org/SELECTION/inputCoord.asp) has a nice
interface for selecting&downloading SRTM data.

Vasile



Yahoo! Groups Links

Ben Discoe wrote:
> Regarding gap filling, to the best of my knowledge SRTMFill and VTBuilder do
> the same simple thing: each unknown heixel is set to the average of the
> surrounding known heixels, repeatedly until there are no more unknowns.

    Yep. No rocket science in SRTMFill. ;)

> -Ben

--
Chris 'Xenon' Hanson aka Eric Hammil | http://www.3DNature.com/ eric at logrus
   "I set the wheels in motion, turn up all the machines, activate the programs,
    and run behind the scenes. I set the clouds in motion, turn up light and
sound,
    activate the window, and watch the world go 'round." -Prime Mover, Rush.

I agree :). Nice idea but a slow server makes CGIAR almost unusable for
large datasets. The interface design remains a strong point. That kind of
options, plus a "select by country" option and a fast way of distributing
should please everyone.

Vasile


> The CGIAR site has a good interface but the download speeds I'm getting
> are about 3 k/sec and I have a T-1. Torrent files would make this go a lot
> faster!
>
>
> -----Original Message-----
> From: vtp@yahoogroups.com [mailto:vtp@yahoogroups.com]On Behalf Of
> Vasile Craciunescu
> Sent: Monday, December 05, 2005 11:13 AM
> To: vtp@yahoogroups.com
> Subject: Re: [vtp] RE: How should SRTM V2 be distributed?
>
>
>> Additionally, perhaps it would be good to put all the gap-fixed 1x1
>> degree
>> tiles on a server, then for each country, simply make a web page with
>> links
>> to that country's tiles, perhaps via an image map showing the tiles as a
>> grid on a map of the country.
>>
>> Yet another idea would be to add a feature to VTBuilder which lets the
>> user
>> set the Area Tool to an area of interest, then VTBuilder will
>> automatically
>> download and merge all the SRTM tiles to produce a single elevation
>> layer
>> for that area.
>>
>> Ultimately, serving SRTM V2 is perhaps best done by an OGC WCS (Web
>> Coverage
>> Service), however i don't know of anyone who has done it yet.
>
> Ben, all,
>
> CGIAR (http://srtm.csi.cgiar.org/SELECTION/inputCoord.asp) has a nice
> interface for selecting&downloading SRTM data.
>
> Vasile
>
>
>
> Yahoo! Groups Links
>
>
>
>
>
>
>
>
>
>
>
>
> Yahoo! Groups Links
>
>
>
>
>
>
>

--- In vtp@yahoogroups.com, "David Rambo" <drambo@u...> wrote:
>
>
> In my experience, far better compression ratios can be achieved with
software like WinRar or 7-Zip than using standard  ZIP/GZIP files.
Both support multiple operating systems. WinRar has an easier to use
interface but 7-zip offers superior compression.
>
> http://www.rarlabs.com/download.htm
> http://www.7-zip.org/download.html
>

Also, winimp is a very good program which has two algorithms, one of
them for text compression. I tested it and it's even better than
winrar for some files. There's also Unimp which is a free portable C
program for extracting .IMP files, and I tested it on Mac OsX and it
works fine!

http://www.technelysium.com.au/unimp.html
http://www.technelysium.com.au/winimp.html

> -----
> From: David Rambo
> Sent: Monday, December 05, 2005 4:02 AM
>
> In my experience, far better compression ratios can be
> achieved with software like WinRar or 7-Zip than using
> standard  ZIP/GZIP files.

This is very true.  One solution i found is bzip2 (http://www.bzip.org/).
It is roughly as good as Rar or 7zip, and it is available a small, fast,
easy to use open-source library.  In my testing, i see results like this:

	 38 SRTM files:        104.5 MB
	 compressed with Gzip:  42.0 MB (40.2% of original)
	 compressed with 7zip:  30.7 MB (29.4%)
	 compressed with RAR:   28.7 MB (27.5%)
	 compressed with Bzip2: 26.8 MB (25.6%)

I have added support to VTBuilder for reading .bz2 files on import from HGT.

The 7zip project looks nice, but i encountered issues with it including very
slow compression and bugs in their Windows archive manager.  To its credit,
7zip is capable of reading and writing archives in .bz2 format.

-Ben

Davide,

thanks for telling us about WinImp (its quite fast and efficient and offers a
good GUI). I did a little analysis below comparing several major compression
formats. Since bandwidth is precious, I would think the software offering the
highest compression would be best:

I did some tests on 3 arc second SRTM2 data for Hawaii (from
http://seamless.usgs.gov). I downloaded the data in ArcGrid format compressed in
a ZIP file. The zip file was 1,993 KB which decompressed to 15,549 KB. I
recompressed the data using several formats with the highest level of
compression selected. Here are the results:

software - file type    - compressed size - time - % of original
gzip     - .gz (gzip)   - 1,894           - 14   - 12.2 %
7-zip    - .zip (zip)   - 1,718           - 35   - 11.0 %
WinRar   - .rar (rar)   - 1,388           - 14   - 8.9 %
WinImp   - .imp (imp)   - 1,246           - 12   - 8.0 %
7-zip    - .7z (7-zip)  - 1,132           - 27   - 7.3 %
WinUHA   - .uha (uharc) -   994           - 36   - 6.4 %
BZip2    - .bz2 (bz2)   -   908           - 17   - 5.8 %

I was completely blown away by BZip2. At 17 seconds and 5.8 % of the original
file size, I was impressed. I'm going to play around with it some more tomorrow.
Being a Windows user, primarily, my only gripe would be the lack of a flexible
GUI.

-Dave R.

-----Original Message-----
From: vtp@yahoogroups.com [mailto:vtp@yahoogroups.com]On Behalf Of
davide_borelli
Sent: Monday, December 05, 2005 2:14 PM
To: vtp@yahoogroups.com
Subject: [vtp] Re: How should SRTM V2 be distributed?


--- In vtp@yahoogroups.com, "David Rambo" <drambo@u...> wrote:
>
>
> In my experience, far better compression ratios can be achieved with
software like WinRar or 7-Zip than using standard  ZIP/GZIP files.
Both support multiple operating systems. WinRar has an easier to use
interface but 7-zip offers superior compression.
>
> http://www.rarlabs.com/download.htm
> http://www.7-zip.org/download.html
>

Also, winimp is a very good program which has two algorithms, one of
them for text compression. I tested it and it's even better than
winrar for some files. There's also Unimp which is a free portable C
program for extracting .IMP files, and I tested it on Mac OsX and it
works fine!

http://www.technelysium.com.au/unimp.html
http://www.technelysium.com.au/winimp.html







Yahoo! Groups Links

> -----
> From: Vasile Craciunescu
>
> CGIAR (http://srtm.csi.cgiar.org/SELECTION/inputCoord.asp)
> has a nice interface for selecting&downloading SRTM data.
>
> -----
> From: David Rambo
>
> The CGIAR site has a good interface but the download speeds
> I'm getting are about 3 k/sec and I have a T-1. Torrent files
> would make this go a lot faster!

I'm getting around 12 K/s from CGIAR today, which is unfortunate; from
NASA's servers i get SRTM at 400 K/s.

I see that CGIAR has arranged gap-filled SRTM V1 into 5x5 degree tiles,
which is an interesting compromise between the original 1x1 degrees tiles
and country/regional scope.  Their zipped GeoTIFFs, signed 16-bit integer
seem to actually work pretty well as a distribution format.  On a faster
server, with SRTM V2, this would be nice.

-Ben

> -----Original Message-----
> From: davide_borelli
> Sent: Monday, December 05, 2005 9:14 AM
>
> Also, winimp is a very good program which has two algorithms,
> one of them for text compression. I tested it and it's even
> better than winrar for some files.

Here's my testing results expanded a bit with winimp.  Interestingly,
winimp's algorithm "1" is the default, but it's not good for this SRTM data,
whereas their algorithm "2" is as good as bzip2.

	 38 SRTM files:  104.5 MB
	 WinZip 9 normal  42.5 MB (40.7% of original)
	 Gzip:            42.0 MB (40.2%)
	 WinImp algo1     39.3 MB (37.6%)
	 7zip:            30.7 MB (29.4%)
	 RAR:             28.7 MB (27.5%)
	 Bzip2:           26.8 MB (25.6%)
	 WinImp algo2     26.7 MB (25.5%)

-Ben

David R., thanks for the detailed interesting testing results.  They agree
quite well with my own tests.

> -----
> From: David Rambo
>
> I did some tests on 3 arc second SRTM2 data for Hawaii (from
> http://seamless.usgs.gov). I downloaded the data in ArcGrid
> format compressed in a ZIP file.

I presume this is an Arc Binary Grid, not Ascii Grid?  Testing on binary is
desirable as it's going to be more efficient overall.

> software - file type    - compressed size - time - % of original
> BZip2    - .bz2 (bz2)   -   908           - 17   - 5.8 %
>
> I was completely blown away by BZip2.

Me too.  That's how i was inspired to jump right in and added .bz2 support
to VTBuilder/vtdata :)

> Being a Windows user,
> primarily, my only gripe would be the lack of a flexible GUI.

For us Windows folks, i found that 7zip comes with a flexible GUI, and it
can read and write .bz2 files.  I tried it with Explorer integration and it
works nicely.  The only other drawback to bzip2 (and gzip) is that they only
contain a single file.

-Ben

Ben,

I was happy I could contribute.

Yes, I used a binary Arc grid.

Today I tested 7-zip's ability to create '.bz2' files (I didn't notice at first
7-Zip could do this because I hadn't created the TAR file first). The original
'.bz2' file I created yesterday was 908 KB and took about 17 seconds to produce.
The '.bz2' file I created today with 7-Zip is 899 KB but took 1 minute and 55
seconds to create!!! That's 98 seconds longer and about 6.8 times as long! (I
double-checked). I decided to check some larger files. I created a 243,835 KB
TAR file containg 2 almost equally sized grids in MapInfo binary numeric and
classified grid formats (.grc/.grd) that came from an antenna propagation model.
Below are the results which show the same pattern:

software - file type    - compressed size - time (m:s) - % of original
7-Zip    - .bz2 (bz2)   -   25,565        - 14:43   - 10.48 %
BZip2    - .bz2 (bz2)   -   25,612        -  2:43   - 10.50 %

For anyones's reference, I used the Win32 Dos based version of Bzip2 software
from http://www.bzip.org/downloads.html. I think for my own purpose I might try
using 7-Zip's GUI to create TAR files of directories/files and then use Bzip2 to
compress them to .tar.bz2

-Dave R.

--- In vtp@yahoogroups.com, "Ben Discoe" <ben@v...> wrote:
>
> > -----Original Message-----
> > From: davide_borelli
> > Sent: Monday, December 05, 2005 9:14 AM
> >
> > Also, winimp is a very good program which has two algorithms,
> > one of them for text compression. I tested it and it's even
> > better than winrar for some files.
>
> Here's my testing results expanded a bit with winimp. Interestingly,
> winimp's algorithm "1" is the default, but it's not good for this SRTM data,
> whereas their algorithm "2" is as good as bzip2.
>
> 38 SRTM files: 104.5 MB
> WinZip 9 normal 42.5 MB (40.7% of original)
> Gzip: 42.0 MB (40.2%)
> WinImp algo1 39.3 MB (37.6%)
> 7zip: 30.7 MB (29.4%)
> RAR: 28.7 MB (27.5%)
> Bzip2: 26.8 MB (25.6%)
> WinImp algo2 26.7 MB (25.5%)
>
> -Ben
>
Wow!
It seems Winimp is the best of them!
When I wrote about the algorithm for text, I was talking about
algorithm 2. You can change the default algorithm in the Options-
Configuration menu, if you want.

From the Winimp help:
"Algorithm 1 (Fast/Normal/Best) is a general-purpose compression
algorithm which works on a wide variety of file types. The Fast
option sacrifices compression for higher speed, and Best sacrifices
speed for higher compression.

Algorithm 2 is a text compression algorithm. It compresses most text
files much better than Algorithm 1, but is worse on many other file
types. It is also not very suitable for small text files. This
algorithm may work very slowly on certain blocks of data containing
large repeated sections, and WinImp switches to Algorithm 1 if this
occurs."

Anyway, I don't know anything about IMP compression speed, but
there's also a command-line version of the program at
http://www.technelysium.com.au/impdl.html (I write it 'cause David
Rambo wrote that the Dos based version of Bzip2 was 98 seconds faster
than the 7-zip's GUI!)
and maybe, with the source C code of Unimp
(http://www.technelysium.com.au/unimp.html) it would be easy to
add .imp support to VTBuilder!




[Non-text portions of this message have been removed]

Hi Ben,

Thanks for your help. I'll try to 'simplify' the buildings layer with
the aim to improve the scene. Thanks again for your time.

Lluís


En/na Ben Discoe ha escrit:

>Hi Lluís,
>
>I spent some time looking at your building data.
>
>For the most part, it is slow simply because it is very detailed, hence a
>high polygon count.  For example, when i tried the "Barri Vell" data, it is
>2860 buildings.  In Enviro, when all the buildings are in view, i get around
>20 fps.  When walking around the town, only some buildings are in view, so
>it is 30 fps.  For larger datasets it gets slower:
>
>"VAlegre_carme_Pedreres", 2970 buildings, 20 fps
>"Eixample fins tren", 8292 buildings, 5 fps
>"Girona sud", 8631 buildings, 4.3 fps
>
>However, i did see one fixable thing that contributes to the slowness:
>rounded building corners.  Whereas a simple building has around 30
>triangles, the ones with rounded corners are _very_ rounded, with 100+
>sides, which means 500+ triangles.  If all the buildings were simple, then
>8000+ buildings would be around 250000 triangles, but the smooth corners are
>probably putting this a fair bit higher.
>
>It would also be possible to make the buildings even more simple - 1 story
>level instead of 2, which should improve fps by at least 50%.
>
>Regarding the river, the best thing to do is probably to draw it into the
>base texture, as the VTP doesn't do dynamic bodies of water (yet).
>
>Hope that helps,
>Ben
>
>
>
>>-----Original Message-----
>>From: Lluís Vicens [mailto:lluis@...]
>>Sent: Tuesday, November 29, 2005 11:37 PM
>>To: Ben Discoe
>>Subject: Re: [vtp] Problems when running Enviro
>>
>>Hi Ben,
>>
>>Ok, as you said, I'll try to compress the vtst files and look
>>for what's going on. In any case, I've put up the files into
>>the file area of the mailing list (BuildingData.zip). I'll be
>>so grateful if you can check this files and express your
>>opinion or ideas about how to solve the problem.
>>
>>
>
>
>
>>[..] as you can see, the city is crossed by a river
>>but I didn't found a solution to represent the river with VTP.
>>
>>
>
>
>
>
>Yahoo! Groups Links
>
>
>
>
>
>
>
>
>
>

Hello,
I downloaded all the files for Italy, I filled the gaps and I
compressed them with Winimp algorithm 2.
The result is:

84 uncompressed .hgt files: 242323368 bytes
1 Italy.imp file: 41993452 bytes

that's circa 17.3%

In these files are included the Countries of Vatican City, Republic of
San Marino, Principality of Monaco and Republic of Malta (...and
Italy, of course!)

I hope these informations can be useful in some way!

Davide

Folks,

I finally found recovery software that was able to read most of the content
of my failed hard drive, and spent around 8 hours yesterday merging the
recovered files into the 6-month backup.  I think the debacle is finally
past.

For those curious about the details, i put a writeup at
http://vterrain.org/Implementation/Misc/drive_trouble.html

-Ben

Ben Discoe wrote:
> Folks,
>
> I finally found recovery software that was able to read most of the content
> of my failed hard drive, and spent around 8 hours yesterday merging the
> recovered files into the 6-month backup.  I think the debacle is finally
> past.
>
> For those curious about the details, i put a writeup at
> http://vterrain.org/Implementation/Misc/drive_trouble.html
>
> -Ben

great to hear! :))))))))))))))

i can imagine you do more backups in the future. ;)

some companies really need a kick in the ass and better not to believe a
marketing site like cnet.

best regards

Ben,

sorry to hear about your trouble. I had a similar issue before and I then I
became a "backupaholic." I would suggesting using an external USB backup drive
to mirror your internal drive on a daily basis (updating the mirrored file with
any changes you made but perhaps without mirroring deletions or etc...).
Microsoft's SyncToy should work well for this and it can be used with the task
scheduler. It's also been my experience that when a drive starts reporting bad
sectors that it's time to get a new drive. Furthermore, I had 3 separate Maxtor
drives fail on me in 3 separate machines over a course of 2 years so I stay away
from them!

-Dave

-----Original Message-----
From: vtp@yahoogroups.com [mailto:vtp@yahoogroups.com]On Behalf Of
hannes_b@...
Sent: Wednesday, December 07, 2005 8:33 PM
To: vtp@yahoogroups.com
Subject: Re: [vtp] Drive trouble story


Ben Discoe wrote:
> Folks,
>
> I finally found recovery software that was able to read most of the content
> of my failed hard drive, and spent around 8 hours yesterday merging the
> recovered files into the 6-month backup.  I think the debacle is finally
> past.
>
> For those curious about the details, i put a writeup at
> http://vterrain.org/Implementation/Misc/drive_trouble.html
>
> -Ben

great to hear! :))))))))))))))

i can imagine you do more backups in the future. ;)

some companies really need a kick in the ass and better not to believe a
marketing site like cnet.

best regards



Yahoo! Groups Links

Not that this is quite on topic, especially for my first post to the list,
but here's a second vote for hard drive mirroring as backup method,
especially when handling large volumes of data.

Having just bought about fifty low-cost 200+gb IDE hard drives this year of
various brands, I can report pretty confidently that Hitachi / IBM have
proven VERY reliable, and Seagate, Maxtor, and Western Digital are least
reliable.

(background: My little business has 5 Windows graphics workstations with 1TB
RAID5 4-drive arrays running intensely about 50% of the day and night. (No
it's not porn! :)  For backup, we simply buy piles of cheap IDE drives, RAR
finished projects (16GB each) onto the drives, disconnect them and put them
on the shelf. This is my most cost-effective option, if you factor hourly
wages I pay my people to handle backup media.   We typically pay $70 for a
250GB IDE drive - not as cheap as DVDs, true, but a hell of a lot simpler
and faster to verify integrity.  Restoring is really quick, too.)

-Josh  "Never underestimate the bandwidth of a station wagon full of hard
drives" White


> -----Original Message-----
> From: vtp@yahoogroups.com [mailto:vtp@yahoogroups.com] On
> Behalf Of David Rambo
> Sent: Thursday, December 08, 2005 7:54 AM
> To: vtp@yahoogroups.com
> Subject: RE: [vtp] Drive trouble story
>
>
> Ben,
>
> sorry to hear about your trouble. I had a similar issue
> before and I then I became a "backupaholic." I would
> suggesting using an external USB backup drive to mirror your
> internal drive on a daily basis (updating the mirrored file
> with any changes you made but perhaps without mirroring
> deletions or etc...). Microsoft's SyncToy should work well
> for this and it can be used with the task scheduler. It's
> also been my experience that when a drive starts reporting
> bad sectors that it's time to get a new drive. Furthermore, I
> had 3 separate Maxtor drives fail on me in 3 separate
> machines over a course of 2 years so I stay away from them!
>
> -Dave
>
> -----Original Message-----
> From: vtp@yahoogroups.com [mailto:vtp@yahoogroups.com]On
> Behalf Of hannes_b@...
> Sent: Wednesday, December 07, 2005 8:33 PM
> To: vtp@yahoogroups.com
> Subject: Re: [vtp] Drive trouble story
>
>
> Ben Discoe wrote:
> > Folks,
> >
> > I finally found recovery software that was able to read most of the
> > content of my failed hard drive, and spent around 8 hours yesterday
> > merging the recovered files into the 6-month backup.  I think the
> > debacle is finally past.
> >
> > For those curious about the details, i put a writeup at
> > http://vterrain.org/Implementation/Misc/drive_trouble.html
> >
> > -Ben
>
> great to hear! :))))))))))))))
>
> i can imagine you do more backups in the future. ;)
>
> some companies really need a kick in the ass and better not
> to believe a marketing site like cnet.
>
> best regards
>
>
>
> Yahoo! Groups Links
>
>
>
>
>
>
>
> Yahoo! Groups Links
>
>
>
>
>
>
>

A bit off topic too, but I really like your website Josh! Very minimalistic and
easy to backup too ; )

Fredrik

       Scalo AB | www.scalo.se
       Stigbergsliden 5 / 7 tr | SE-414 63 Göteborg
       Direct: +46 (0)31 15 05 10 | Mobile: +46 (0)708 49 98 00 | E-mail:
fredrik@...

       Scalo levererar digital tredimensionell visualisering och VR till
beställare inom Flygplats,
       Byggnad, Landskap, Geo, Väg, Miljö, Hamn, Gata, VVS, El och VA.


   ----- Original Message -----
   From: Josh White
   To: vtp@yahoogroups.com
   Sent: Thursday, December 08, 2005 17:25
   Subject: RE: [vtp] Drive trouble story


   Not that this is quite on topic, especially for my first post to the list,
   but here's a second vote for hard drive mirroring as backup method,
   especially when handling large volumes of data.

   Having just bought about fifty low-cost 200+gb IDE hard drives this year of
   various brands, I can report pretty confidently that Hitachi / IBM have
   proven VERY reliable, and Seagate, Maxtor, and Western Digital are least
   reliable.

   (background: My little business has 5 Windows graphics workstations with 1TB
   RAID5 4-drive arrays running intensely about 50% of the day and night. (No
   it's not porn! :)  For backup, we simply buy piles of cheap IDE drives, RAR
   finished projects (16GB each) onto the drives, disconnect them and put them
   on the shelf. This is my most cost-effective option, if you factor hourly
   wages I pay my people to handle backup media.   We typically pay $70 for a
   250GB IDE drive - not as cheap as DVDs, true, but a hell of a lot simpler
   and faster to verify integrity.  Restoring is really quick, too.)

   -Josh  "Never underestimate the bandwidth of a station wagon full of hard
   drives" White


   > -----Original Message-----
   > From: vtp@yahoogroups.com [mailto:vtp@yahoogroups.com] On
   > Behalf Of David Rambo
   > Sent: Thursday, December 08, 2005 7:54 AM
   > To: vtp@yahoogroups.com
   > Subject: RE: [vtp] Drive trouble story
   >
   >
   > Ben,
   >
   > sorry to hear about your trouble. I had a similar issue
   > before and I then I became a "backupaholic." I would
   > suggesting using an external USB backup drive to mirror your
   > internal drive on a daily basis (updating the mirrored file
   > with any changes you made but perhaps without mirroring
   > deletions or etc...). Microsoft's SyncToy should work well
   > for this and it can be used with the task scheduler. It's
   > also been my experience that when a drive starts reporting
   > bad sectors that it's time to get a new drive. Furthermore, I
   > had 3 separate Maxtor drives fail on me in 3 separate
   > machines over a course of 2 years so I stay away from them!
   >
   > -Dave
   >
   > -----Original Message-----
   > From: vtp@yahoogroups.com [mailto:vtp@yahoogroups.com]On
   > Behalf Of hannes_b@...
   > Sent: Wednesday, December 07, 2005 8:33 PM
   > To: vtp@yahoogroups.com
   > Subject: Re: [vtp] Drive trouble story
   >
   >
   > Ben Discoe wrote:
   > > Folks,
   > >
   > > I finally found recovery software that was able to read most of the
   > > content of my failed hard drive, and spent around 8 hours yesterday
   > > merging the recovered files into the 6-month backup.  I think the
   > > debacle is finally past.
   > >
   > > For those curious about the details, i put a writeup at
   > > http://vterrain.org/Implementation/Misc/drive_trouble.html
   > >
   > > -Ben
   >
   > great to hear! :))))))))))))))
   >
   > i can imagine you do more backups in the future. ;)
   >
   > some companies really need a kick in the ass and better not
   > to believe a marketing site like cnet.
   >
   > best regards
   >
   >
   >
   > Yahoo! Groups Links
   >
   >
   >
   >
   >
   >
   >
   > Yahoo! Groups Links
   >
   >
   >
   >
   >
   >
   >




   Yahoo! Groups Links









[Non-text portions of this message have been removed]

Oops, my main index page, I kind of forgot about it!  Ben harasses me often
about my utter disregard for updating my websites. Well deserved harassment,
obviously. I'm guessing you were looking for something like my backstory:
www.joshwhiteconsulting.com/portfolio is the best I have online right now,
but it does not mention that I started a 2D motion graphics studio
(www.wulumulu.com, www.wulumulu.com/reel.wmv). I'm moving back to 3D
somewhat at the moment, doing some architectural visualization for a golf
course developer... Whee!

Anyway. Shhh! No more offtopic!

-Josh


> -----Original Message-----
> From: Josh White [mailto:josh@...]
> Sent: Thursday, December 08, 2005 8:25 AM
> To: 'vtp@yahoogroups.com'
> Subject: RE: [vtp] Drive trouble story
>
>
> Not that this is quite on topic, especially for my first post
> to the list, but here's a second vote for hard drive
> mirroring as backup method, especially when handling large
> volumes of data.
>
> Having just bought about fifty low-cost 200+gb IDE hard
> drives this year of various brands, I can report pretty
> confidently that Hitachi / IBM have proven VERY reliable, and
> Seagate, Maxtor, and Western Digital are least reliable.
>
> (background: My little business has 5 Windows graphics
> workstations with 1TB RAID5 4-drive arrays running intensely
> about 50% of the day and night. (No it's not porn! :)  For
> backup, we simply buy piles of cheap IDE drives, RAR finished
> projects (16GB each) onto the drives, disconnect them and put
> them on the shelf. This is my most cost-effective option, if
> you factor hourly wages I pay my people to handle backup
> media.   We typically pay $70 for a 250GB IDE drive - not as
> cheap as DVDs, true, but a hell of a lot simpler and faster
> to verify integrity.  Restoring is really quick, too.)
>
> -Josh  "Never underestimate the bandwidth of a station wagon
> full of hard drives" White
>
>
> > -----Original Message-----
> > From: vtp@yahoogroups.com [mailto:vtp@yahoogroups.com] On
> > Behalf Of David Rambo
> > Sent: Thursday, December 08, 2005 7:54 AM
> > To: vtp@yahoogroups.com
> > Subject: RE: [vtp] Drive trouble story
> >
> >
> > Ben,
> >
> > sorry to hear about your trouble. I had a similar issue
> > before and I then I became a "backupaholic." I would
> > suggesting using an external USB backup drive to mirror your
> > internal drive on a daily basis (updating the mirrored file
> > with any changes you made but perhaps without mirroring
> > deletions or etc...). Microsoft's SyncToy should work well
> > for this and it can be used with the task scheduler. It's
> > also been my experience that when a drive starts reporting
> > bad sectors that it's time to get a new drive. Furthermore, I
> > had 3 separate Maxtor drives fail on me in 3 separate
> > machines over a course of 2 years so I stay away from them!
> >
> > -Dave
> >
> > -----Original Message-----
> > From: vtp@yahoogroups.com [mailto:vtp@yahoogroups.com]On
> > Behalf Of hannes_b@...
> > Sent: Wednesday, December 07, 2005 8:33 PM
> > To: vtp@yahoogroups.com
> > Subject: Re: [vtp] Drive trouble story
> >
> >
> > Ben Discoe wrote:
> > > Folks,
> > >
> > > I finally found recovery software that was able to read
> most of the
> > > content of my failed hard drive, and spent around 8 hours
> yesterday
> > > merging the recovered files into the 6-month backup.  I think the
> > > debacle is finally past.
> > >
> > > For those curious about the details, i put a writeup at
> > > http://vterrain.org/Implementation/Misc/drive_trouble.html
> > >
> > > -Ben
> >
> > great to hear! :))))))))))))))
> >
> > i can imagine you do more backups in the future. ;)
> >
> > some companies really need a kick in the ass and better not
> > to believe a marketing site like cnet.
> >
> > best regards
> >
> >
> >
> > Yahoo! Groups Links
> >
> >
> >
> >
> >
> >
> >
> > Yahoo! Groups Links
> >
> >
> >
> >
> >
> >
> >
>

Hi all,
Is anyone getting good-looking, high density tree cover in their models.
  I'm particularly interested in modeling forested landscapes and would
love to learn tips/tricks from anyone else doing so successfully.  It's
been a while since I've worked with veg data, but am getting back into
it now.

Tyler