 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
High!
Everybody is talking about GoogleEarth... even in less high
tech-enthusiastic countries like Germany, the issue made it into prime
magazine stories. But as GoogleEarth aims to be a real-time navigable
representation of our planet, its level of detail is limited and far
from photorealistic. As a commercial, proprietary project, the satellite
and aerial imagery also is tagged over and over with those annoying "(C)
2005 Google" watermarks. Then, for vast parts of Earth's land mass,
GoogleEarth's satellite maps are of lower resolutions as publicly
available nowadays (for example the Landsat series).
So what about starting a PoVEarth project? There are terabytes of public
domain NASA/JPL image mosaics available, down the line at 0.5 arcseconds
/pixel, which equals about 15 m/pixel on the equator. And not only image
mosaics, but also altimetry data at a horizontal and vertical resolution
high enough to do it without vertical exaggeration (3 arcseconds per
sample, 1 metre vertical accuracy)! In my last rendering of Afghanistan,
I used one of these 1 by 1 degree tiles...
We could start with an overview representation of the Earth made of,
let's say, 5 by 5 degree image mosaic tiles and the less accurate
GTOPO30 altimetry data (15 arcseconds per sample, yet still 1 metre
vertical accuracy) and then gradually filling in with the more precise
datasets.
Quite a challenge would be the color adjustment of the tiles, as they
were photographed during several Space Shuttle Radar Topography Missions
(SRTMs) in 2000 and 2001, so frequently adjacent portions of the tiles
were shot at different daytimes and seasons, consequently showing
different lighting angles, vegetation status and snow cover (an example:
Tierra del Fuego on http://makeashorterlink.com/?H130237CB - note the
sharp boundary between mountains deep in snow and almost snow-free or
between dry reddish and lush green pampa!
At a later stage of development, we could again start filling in even
larger-scaled altimetry data derived from printed maps (for example 1 :
25.000 topographic maps) and converted from geodesic to simple
cylindrical projection, same with aerial photographs replacing satellite
images... this also would be the time to start populating the landscape
with actual single rocks, trees and buildings.
After the first "overview" modeling, each participant of the project
might start modeling one certain place, region or country more
thoroughly - whether this might be his or her home town or some other
area of personal interest. With me, this would include the Cologne area,
of course Afghanistan and some subantarctic islands...
We PoV people don't have billions of dollars behind us. But we have
thousands of gifted and enthusiastic programmers and multimedia geeks,
and, most important, no deadline! No banks and shareholders would
prevent us from embarking upon a decades- or even centuries-long
raytracing adventure which only the grandchildren of our grandchildren
live to complete. This soon could develope into one of the largest and
longest-lasting non-profit computing enterprises ever undertaken...
So, now what about PoVEarth?
See you in Khyberspace - and not only there!
Yadgar
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
I like the idea very much as I have played also somewhat with the GTOPO30,
and the NASA DEM data. My personal interest is the Iberian Peninsula by the
way (oh! maybe Deception Island too, which I visited, 20 years ago...).
It is a project with ambition. I think that I would need another PC anyway
for better data crunching (something I am thinking about seriously)... and a
lot of theoretical help too. I am really a lay person in this matter. Time
would not be a problem, though.
I am tempted to contribute...
Thomas
news:4324810e$1@news.povray.org...
> High!
>
> Everybody is talking about GoogleEarth... even in less high
> tech-enthusiastic countries like Germany, the issue made it into prime
> magazine stories. But as GoogleEarth aims to be a real-time navigable
> representation of our planet, its level of detail is limited and far
> from photorealistic. As a commercial, proprietary project, the satellite
> and aerial imagery also is tagged over and over with those annoying "(C)
> 2005 Google" watermarks. Then, for vast parts of Earth's land mass,
> GoogleEarth's satellite maps are of lower resolutions as publicly
> available nowadays (for example the Landsat series).
>
> So what about starting a PoVEarth project? There are terabytes of public
> domain NASA/JPL image mosaics available, down the line at 0.5 arcseconds
> /pixel, which equals about 15 m/pixel on the equator. And not only image
> mosaics, but also altimetry data at a horizontal and vertical resolution
> high enough to do it without vertical exaggeration (3 arcseconds per
> sample, 1 metre vertical accuracy)! In my last rendering of Afghanistan,
> I used one of these 1 by 1 degree tiles...
>
> We could start with an overview representation of the Earth made of,
> let's say, 5 by 5 degree image mosaic tiles and the less accurate
> GTOPO30 altimetry data (15 arcseconds per sample, yet still 1 metre
> vertical accuracy) and then gradually filling in with the more precise
> datasets.
>
> Quite a challenge would be the color adjustment of the tiles, as they
> were photographed during several Space Shuttle Radar Topography Missions
> (SRTMs) in 2000 and 2001, so frequently adjacent portions of the tiles
> were shot at different daytimes and seasons, consequently showing
> different lighting angles, vegetation status and snow cover (an example:
> Tierra del Fuego on http://makeashorterlink.com/?H130237CB - note the
> sharp boundary between mountains deep in snow and almost snow-free or
> between dry reddish and lush green pampa!
>
> At a later stage of development, we could again start filling in even
> larger-scaled altimetry data derived from printed maps (for example 1 :
> 25.000 topographic maps) and converted from geodesic to simple
> cylindrical projection, same with aerial photographs replacing satellite
> images... this also would be the time to start populating the landscape
> with actual single rocks, trees and buildings.
>
> After the first "overview" modeling, each participant of the project
> might start modeling one certain place, region or country more
> thoroughly - whether this might be his or her home town or some other
> area of personal interest. With me, this would include the Cologne area,
> of course Afghanistan and some subantarctic islands...
>
> We PoV people don't have billions of dollars behind us. But we have
> thousands of gifted and enthusiastic programmers and multimedia geeks,
> and, most important, no deadline! No banks and shareholders would
> prevent us from embarking upon a decades- or even centuries-long
> raytracing adventure which only the grandchildren of our grandchildren
> live to complete. This soon could develope into one of the largest and
> longest-lasting non-profit computing enterprises ever undertaken...
>
> So, now what about PoVEarth?
>
> See you in Khyberspace - and not only there!
>
> Yadgar
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Thomas de Groot wrote:
> I like the idea very much as I have played also somewhat with the GTOPO30,
> and the NASA DEM data.
Do you mean by the latter the SRTM 3-arcsecond GeoTIFFs? If you had not
heard about them until now, the URL is
ftp://ftp.glcf.umiacs.umd.edu/glcf/SRTM/Degree_Tiles !
My personal interest is the Iberian Peninsula by the
> way
Entire Spain, Andorra and Portugal? A formidable task, comparable in
size to my dreams of virtual Afghanistan!
(oh! maybe Deception Island too, which I visited, 20 years ago...).
Cool, you went skinny-dipping in Antarctica! Have you also been to South
Georgia on that cruise? That's one of my favourite places outside
Afghanistan...
> It is a project with ambition. I think that I would need another PC anyway
> for better data crunching (something I am thinking about seriously)...
What machine do you own at the moment? Mine is an AMD Athlon 64 2800+,
running at 1.8 GHz, 512 megs of RAM... probably I'll upgrade to 1 gig
somewhen in the coming months!
and a
> lot of theoretical help too.
Yes, math is an issue... I wrote a PoV template for generating curved
meshes (i. e. meshes that follow Earth surface's curvature) from ASCII
height matrizes derived from the SRTM GeoTIFFs (the latter done with
3DEM), but I'm currently totally unaware how to write an algorithm for
optimizing the number of height samples to be taken into account for
mesh calculation depending on the camera viewpoint...
Then we also should have a program that allows filling in the data gaps
in the original SRTM data - 3DEM has such a routine, but it does not
work with gaps bordering the edges of a one-degree tile!
Also, we should make use of a multi-bezier patch utility written some
time ago by Tor Olav Kristensen to get a real smooth surface (which at a
later stage can be "roughened up" in a controlled way to get rocky
mountainsides etc.)
I am really a lay person in this matter. Time
> would not be a problem, though.
> I am tempted to contribute...
Welcome aboard! Let's beat the drum together for fellow combatants!
See you in Khyberspace!
Yadgar
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
news:43270edb$1@news.povray.org...
> Thomas de Groot wrote:
> > I like the idea very much as I have played also somewhat with the
GTOPO30,
> > and the NASA DEM data.
>
> Do you mean by the latter the SRTM 3-arcsecond GeoTIFFs? If you had not
> heard about them until now, the URL is
> ftp://ftp.glcf.umiacs.umd.edu/glcf/SRTM/Degree_Tiles !
>
No, I have not experimented with those. I looked up my data (1998-1999 I
guess): I ordered a couple of data CD's from the USGS, and they are
GTOPO30s. Not sure now about thos NASA DEMs I mentioned...
> Entire Spain, Andorra and Portugal? A formidable task, comparable in
> size to my dreams of virtual Afghanistan!
>
Hm yes... It was for the geological structural features of the peninsula as
an overview, and to see how much you could zoom in to them. Also the coastal
area of Portugal/Galicia was part of my more detailed interests, but there
the resolution was too bad.
> (oh! maybe Deception Island too, which I visited, 20 years ago...).
>
> Cool, you went skinny-dipping in Antarctica! Have you also been to South
> Georgia on that cruise? That's one of my favourite places outside
> Afghanistan...
Yes, in 1985, I went on a scientific cruise on board 'Polarstern' to the
Bransfield Strait and went down as far south as Anvers Island approximately.
Only time on land was Deception Island though. Was impressive. Not been to
South Georgia.
Went back through the Strait of Magellan in Patagonia, to Punta Arenas. Such
impressive landscape... the end of the world.
>
>
> What machine do you own at the moment? Mine is an AMD Athlon 64 2800+,
> running at 1.8 GHz, 512 megs of RAM... probably I'll upgrade to 1 gig
> somewhen in the coming months!
>
Old Compaq Pentium3 133Mhz...
I am thinking about an Athlon 64 machine indeed.
>
> Welcome aboard! Let's beat the drum together for fellow combatants!
>
Right!
Thomas
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
High!
> No, I have not experimented with those.
You really should... their resolution is 5 times higher than that of
GTOPO30! As they are only in GeoTIFF available, you'll need a conversion
software such as 3DEM ( http://www.visualizationsoftware.com/3dem.html )
to transform them into ASCII matrizes, which can be read in from PoV-Ray!
> Yes, in 1985, I went on a scientific cruise on board 'Polarstern'
Did you research back then? Or did the 'Polarstern' occasionally pick up
(well-paying) tourists when going to Antarctica?
> Went back through the Strait of Magellan in Patagonia, to Punta Arenas. Such
> impressive landscape... the end of the world.
Must be awesome - evergreen subantarctic rainforest side by side with
glaciers ending at sea level...
> Old Compaq Pentium3 133Mhz...
A Pentium at 133 MHz? Or a Pentium 3 at 1333 MHz?
>>Welcome aboard! Let's beat the drum together for fellow combatants!
>
> Right!
Do you know of other PoVers how could like to join us? There are some
people on the PoV scene who already digged quite into Earth
visualization, such as Christoph Hormann... but I remember him being not
very willing to share his experience :-((
See you in Khyberspace!
Yadgar
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Yadgar wrote:
> There are some
> people on the PoV scene who already digged quite into Earth
> visualization, such as Christoph Hormann... but I remember him being not
> very willing to share his experience :-((
What?
I don't remember ever having turned down someone asking questions
related to terrain rendering. On the contrary there is quite some
information available on my website even without asking. What i won't
do even if i am asked is doing others homework.
Your posting here does not only describe something extremely ambitious,
it also quite strongly indicates you did not truely realize how
ambitious if meant seriously. This is emphasized by the lot of
inaccuracies that jump into the eye when reading the text like:
- confusing of 'resolution' with 'accuracy'.
- the SRTM did not produce any optical data so there is no necessity for
color processing, the link you posted is to an image generated from
landsat imagery.
Those things might not seem important to you but when planning a project
like this i think it would first of all be important to have an idea
about the basic facts like what data sources actually exist.
You should realize though that detailed large scale earth surface
rendering (meaning high resolution data with less than 100m grid spacing
over larger areas) is one of the toughest tasks concerning data
handling. Working with ascii data file formats and mesh files will not
lead you anywhere. For a render like on
http://news.povray.org/povray.binaries.images/thread/%3Ccllr5b%24v86%241%40chho.imagico.de%3E/
to be suited for high resolution printing (the one that is posted there
was not yet sufficient in this respect) you have to process several GB
of (binary) data for example. (the bicubic interpolation patch
mentioned in this posting is meanwhile available in MegaPOV BTW)
This is not meant to discourage anyone to do earth renders - on the
contrary i have seen very nice examples of small scale renders like for
example
http://news.povray.org/povray.binaries.images/thread/<421a4678%40news.povray.org>/
and it would be great to see more of such. People working on such
concrete earth rendering project can be sure to get suggestions and tips
from me if i have them.
Christoph
--
POV-Ray tutorials, include files, Landscape of the week:
http://www.tu-bs.de/~y0013390/ (Last updated 24 Jul. 2005)
MegaPOV with mechanics simulation: http://megapov.inetart.net/
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
High!
Christoph Hormann schrieb:
> Your posting here does not only describe something extremely ambitious,
> it also quite strongly indicates you did not truely realize how
> ambitious if meant seriously. This is emphasized by the lot of
> inaccuracies that jump into the eye when reading the text like:
>
> - confusing of 'resolution' with 'accuracy'.
> - the SRTM did not produce any optical data so there is no necessity for
> color processing, the link you posted is to an image generated from
> landsat imagery.
No, the latter I did not assume - I'm quite aware that SRTM is only
altimetry data! Obviously you misunderstood me... the statement about
color processing was related to the Landsat sets!
> Those things might not seem important to you but when planning a project
> like this i think it would first of all be important to have an idea
> about the basic facts like what data sources actually exist.
Yes, I had not yet dived into the layer documentations of the Landsat
images... perhaps they give some hints on actual ground cover and, thus,
an idea how the "basic pigment" of a certain landscape might look like...
> You should realize though that detailed large scale earth surface
> rendering (meaning high resolution data with less than 100m grid spacing
> over larger areas) is one of the toughest tasks concerning data
> handling. Working with ascii data file formats and mesh files will not
> lead you anywhere.
Yes, I can imagine that... any bundle of ASCII-data which consists of
more than a few SRTM tiles would simply be too large to handle - unless
I content myself exclusively with low-altitude flights or even
"pedestrian views" where the horizon is not that far...
For a render like on
>
>
http://news.povray.org/povray.binaries.images/thread/%3Ccllr5b%24v86%241%40chho.imagico.de%3E/
>
>
> to be suited for high resolution printing (the one that is posted there
> was not yet sufficient in this respect) you have to process several GB
> of (binary) data for example. (the bicubic interpolation patch
> mentioned in this posting is meanwhile available in MegaPOV BTW)
What is "bicubic interpolation"? Does it relate to image texture files
or to altimetry data?
> This is not meant to discourage anyone to do earth renders - on the
> contrary i have seen very nice examples of small scale renders like for
> example
>
> http://news.povray.org/povray.binaries.images/thread/<421a4678%40news.povray.org>/
Cute... but the things Thomas and I would like to go for would be far
more realistic...
> and it would be great to see more of such. People working on such
> concrete earth rendering project can be sure to get suggestions and tips
> from me if i have them.
What do you think about the whole (yet still embryonic, I admit) concect
of PoVEarth?
See you in Khyberspace!
Yadgar
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Yadgar wrote:
>
> What is "bicubic interpolation"?
See docs:
http://megapov.inetart.net/manual-1.2.1/patterns.html#bicubic_image
and sample scene:
http://megapov.inetart.net/demo/interpolate_iso.pov
>
> What do you think about the whole (yet still embryonic, I admit) concect
> of PoVEarth?
I think i made that quite clear.
Christoph
--
POV-Ray tutorials, include files, Landscape of the week:
http://www.tu-bs.de/~y0013390/ (Last updated 24 Jul. 2005)
MegaPOV with mechanics simulation: http://megapov.inetart.net/
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
High!
> See docs:
>
> http://megapov.inetart.net/manual-1.2.1/patterns.html#bicubic_image
As a (yet) math layman, I would like to know what's the advantage
compared to traditional POV method 2 and 4...
> http://megapov.inetart.net/demo/interpolate_iso.pov
We'll try it out!
See you in Khyberspace!
Yadgar
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"Yadgar" <yaz### [at] gmx de> schreef in bericht
news:432### [at] gmxde...
>
> You really should... their resolution is 5 times higher than that of
> GTOPO30! As they are only in GeoTIFF available, you'll need a conversion
> software such as 3DEM ( http://www.visualizationsoftware.com/3dem.html )
> to transform them into ASCII matrizes, which can be read in from PoV-Ray!
>
Yes I am sure, and I will.
>
>
> Did you research back then? Or did the 'Polarstern' occasionally pick up
> (well-paying) tourists when going to Antarctica?
>
No, I was a researcher on board. Although the true story was a little bit
more complicated than that, as there was also a political reason for me and
my colleague to be on board: the (active) participation of the Netherlands
in the Antarctic Treaty. So, it turned out that we mainly helped others with
their research, taking the night shifts for example, and every kind of
things. On board research is a round the clock business you know, as
shiptime is very expensive.
> > Went back through the Strait of Magellan in Patagonia, to Punta Arenas.
Such
> > impressive landscape... the end of the world.
>
>
> > Old Compaq Pentium3 133Mhz...
>
> A Pentium at 133 MHz? Or a Pentium 3 at 1333 MHz?
Ouch!! Been dreaming!! 1 GHz is what it is!
>
> Do you know of other PoVers how could like to join us?
Those who feel interested will answer. However, As I said, it is very
ambitious and probably more of a multi-generational scope. Not sure if it is
viable project. If I remember well, Christoph Hormann did some awesome
renders using this data. Ah! But then we speak about a master :-)
Thomas
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
|
|
