CFS3 Terrain Creation - is anyone still around?

[gecko]

Ted
I am presently running 30x30 srtm files creating mesh files. 17920 columns and rows/pixel dimension. 1720 km theater.

Chose not to run make mesh on my tower computer.
108 hours so far on a desktop, which I would consider slow. Averaging 2 files created per minute. Presently about halfway through 512. Have 256 and 128 files to go having already created 4096, 2048, and 1024.

Would this influence the resolution of the lcf file? I have heard of one person creating a much higher resolution lcf for WOFF, and I think that's a huge step forward in terms of the overall appearance of the terrain. Having landclasses in roughly 1 km x 1 km blocks seems a bit dated.


For water, cfs3h2o.dds is the key. It is a direct map of the entire theater, but it is flipped vertically. Note that it is not rotated 180 degrees, but flipped. It was very tedious to map out the shorelines pixel by pixel, and I got a huge help from a guy over at the WOFF forum who did probably 75% of that part. I think it would be a lot easier if you're able to export your water shapes to a texture sheet somehow. I used a 4096x4096 map, but I think higher might allow you to put in some decent reef detail. According to Ankor, even though the file size would be very large, the way it is done in CFS3 is efficient and will not cause a huge performance hit. I'd love to see what could be done in 16384 resolution! I would think it might be possible to paint an individual body of water or island coastline on a smaller sheet and then put the different pieces into a single large map at the appropriate places. Or are you suggesting that it might be possible to get CFS3 to recognize more than one texture each covering a section of the map.

The 16 h2od_XX.dds files are not used for color anymore, but now are used only for creating the bump map for the waves.

EDIT: here's a link to the water colors I did for the stock map: https://www.sim-outhouse.com/sohforu...ors-Europe-zip

[gecko]

Also note that water color is heavily affected by the color of the sky and clouds as it is in real life. I am currently working on bringing big improvements in this area to CFS3.

[rbp71854]

Or are you suggesting that it might be possible to get CFS3 to recognize more than one texture each covering a section of the map.

The 16 h2od_XX.dds files are not used for color anymore, but now are used only for creating the bump map for the waves.

I am at a loss when it comes to landclass, it was a dead stop for me when doing RS as I do not have adobe photoshop, but have coral. I am hoping in the future someone will come to my rescue on png redux when the time comes to put together the lcf.

Yes Dan my thought is that there may be a method to have separate dds files for ocean, coral ocean, and lakes. Presently we process the lake and ocean water as two pol files as they are areas and at the same time. I think they use the same dds file.

Rivers being vectors are processed with railroads and roads.
Rivers have their own separate texture sheet from ocean water.

Do you know specifically which dds file is for the ocean? Update figured it was the h2o.dds.

What are the 16 h20d_xx files you mention?



Thank you for the info.

In further thinking I have not tried running an area vector in the the vector data process normally used for line type vectors. hmmm maybe maybe not.

[sdsbolt]

Ted

What files did you download?

EarthExplorer/nasa lpdaac collection/modis landcover ver6.1
I used the 2002/2003 data as its about the resolution I wanted but newer data is much sharper

[NachtPiloten]

I was going to make a couple of theaters with different values for the last two numbers and see if the differences were noticeable. Glad to see that I was not the only one making a 30 by 30 theater. Once we figure this out this will mean so much to the old game.

There is one more data point that interests me in the ers2tiff process and it is this one ers2tiff *.ers N38.1875 W28.1875 256 1024 1974 CFS3Europe -393216 -393216 3 3. As you know when you change (value in red Stride Check as per the CFS3 Theater Scratchpad) this to a lower value it alters the dimensions displayed for the theater. But for the life of me, I really do not know what this value does as far as creating msh files. I question this value since it does alter the theater size and with a 1024 value the dimensions in km is not equal to the size of the theater. Lowering it to 256 in some cases gets the size to be similar to what the theater is but not exact. Steve mentioned to be that he believes that a lower value sharpens up the scenery. What is interesting the scratchpad has a value of 2 as the smallest values that can be used. What I am getting to understand (or not) is this, if you want all these numbers to work, we may need to select our theaters in km and find a stride check that gives us a value that agrees with the formula in the new theater kit supplied by Odin. We do as rbp does a use the larger values for the last two numbers.....This may explain the size of the cfs3 theater being smaller than the max allowable and the original lcf being 1536 pixels. Oh the mystery.... I am going to spend some time in the Theater scratchpad to see what I can learn. Anyone know who Robert Bruce is, his name appears as you scroll over some cells....

[NachtPiloten]

Robert Bruce - I fear his cancer diagnosis was terminal as his last activity was in 2008. Found this though Virtual Terrains (scene-explore.blogspot.com) Really detailed reading. Wish his attached files could be found.

[NachtPiloten]

It seems that to get the math to work in CFS3, picking the size of the theater in km is key to have the dimensions to sync with the math in the tiff2ers. In an earlier post the original CFS3 theater is not 1835, but 1572 or so in km. If you enter this into the spreadsheet main tab and you enter in the original CFS3 data (position of theater etc you reproduce the data that is in the tiff2ers code, that is the dim data ers2tiff *.ers N38.1875 W28.1875 256 1024 1974 CFS3Europe -393216 -393216 3 3. Double these and you get 786432, which is the data in the formulas Robert Bruce developed. This goes along with my earlier posts on the fact that the original CFS3 is 14 degrees N/S and 21 degrees E/W and to make it square you can change this in Global mapper. When you do this you will get data in meters which I hope is in sync with all of this. Bottom line, I think I am getting the math and how to enter the stride check data (the 1024 number). Will need to take more time than I can give it this week but may get it yet. Please look at the terrain scratchpad.

[rbp71854]

Before I began running the 30x30 srtm Dems, I ran some Worldwide - GTOPO30 Format Dems with Resolution: ~900 m, 30 arc-sec

With the gotopo the columns and rows were 1792 and ended up with 7 7 in the ERS2tiff command line. Notice the resolution of the gotopo. I believe the 30x30 maybe 90 meter resolution. therefore the 17920 columns and rows and 70 70.

Presently over 6 days running mesh, still in 510. Probably got another week. Just hope no loss of power next week with eclipse and thunderstorms forecasted all next week.

[gecko]

I am at a loss when it comes to landclass, it was a dead stop for me when doing RS as I do not have adobe photoshop, but have coral. I am hoping in the future someone will come to my rescue on png redux when the time comes to put together the lcf.

Yes Dan my thought is that there may be a method to have separate dds files for ocean, coral ocean, and lakes. Presently we process the lake and ocean water as two pol files as they are areas and at the same time. I think they use the same dds file.

Rivers being vectors are processed with railroads and roads.
Rivers have their own separate texture sheet from ocean water.

Do you know specifically which dds file is for the ocean? Update figured it was the h2o.dds.

What are the 16 h20d_xx files you mention?



Thank you for the info.

In further thinking I have not tried running an area vector in the the vector data process normally used for line type vectors. hmmm maybe maybe not.

There are 16 h20d_xx.dds in the effects/fxtextures folder. In stock CFS3 they controlled the wave animation and influenced water color. With Ankor's shaders they no longer influence color, which is now entirely determined by cfs3h2o.dds and environment reflections from the sky.

I have been curious if the river vectors could be repurposed. For instance on the stock map, many large and important rivers are represented by the river vector and appear only as a minor stream. On the one hand a more limited use of the river vector for only those truly minor streams might be appropriate, or the river texture could possibly be repainted as an alternate shoreline without surf and the vector placed along the shoreline. That way we could still have ocean surf for ocean shorelines, but also a lake shore/river bank textures for calmer bodies of water.

I have experimented with using the alpha channel to mix the local color of the water with the shore texture instead of trying to paint water the onto the shoreline. I then added a white wave foam texture on top for the surf. It then fits with any location, color-wise, but it would be great to be able to limit surf to ocean shorelines while still having a shoreline texture for the rest of the bodies of water.

[MajorMagee]

From my AnKor's Shaders Developer Notes.

*** 5. Water Mods ***

CFS3 default water animation files are located in the effects\fxtextures folder and are named h2od_01.dds through h2od_16.dds


With AnKors shaders the color for these files is rendered using the cfs3h2o.dds file for the local water color information. The cfs3h2o.dds file is a vertically inverted rendering of the entire theater so each pixel's tint covers a fairly large area of sea surface shading. Edits to cfs3h2o.dds can create local effects like muddy river water exiting a delta into the sea, sandy shoals and coral reefs.


The BaseOpacity parameter controls how easily you can see through the water animation to the texture underneath. This can easily go wrong if the underlying textures don't look like a sea bed.


The reflectiveness of the clouds and sky on the surface is controlled by the ReflectLevel parameter. The sun sparkle on the surface is controlled by the SpecularPower and SpecilarNormDivisor parameters.


The AddFoam parameter in SeaWater.fx is used to determine the grey scale level that will be interpreted as foam. The animations files need to have their brightness level distribution adjusted to provide discrete areas for foam patches to appear and dissapear as the wave crest passes by. The boundry of the textures need to wrap the opposide edges to avoid having seams appear. A somewhat randomized pattern of foam patches helps hide repeating patterns as the sea surface is tiled with these textures. Using low numbers creates an interesting effet that will make the sea surface appear to be covered in sea ice, high numbers will remove all of the foam patterns.


An optional set of water animation files set up to display the sea foam are provided with this package, but you may use the game's originals if you prefer without any conflict other than the difference in wave pattern and foam appearance.


It turns out that the alpha channel in the effects\fxtextures\h20d_##.dds files has to be pretty finely textured, and the color variations have to be subtle (no harsh shade transitions) for it to bump map realistically.


The CFS3 Default Water Detail Texture Size is 5 and the SeaWater.fx settings in this package are preset to match. Using a different texture size will change the appearance of the wave dimensions unless you adjust the WaveBump and WaveScale values in SeaWater.fx to match.

[sdsbolt]

the river texture could possibly be repainted as an alternate shoreline without surf and the vector placed along the shoreline

yeah rivers are a bit of a problem but your idea may work. Is there a river shoreline with alpha channel to be borrowed from somewhere?

[rbp71854]

In the past I have always processed the lake water and ocean water together. In the future it may be possible to do them separately. In RS Sino major rivers were also processed with the lakes and ocean as area vectors.

In GlobalMapper I have a shoreline layer and a coastline layer. The coastline layer vector define the ocean, harbors, and large rivers.(its a continuous line) The shoreline layer defines those areas that you want to have surf. (It doesn't have to be a continuous line around your water features.)

An interesting thing I found while searching the files I found a texture called reef which I have a vague memory that I was going to add a reef in certain areas of Pearl by means of the vector process by re-doing railroads, roads, rivers, etc.

Thanks Andy for that info. I wish I knew more about Ankor's add on. It would be ideal to integrate that into the build, or maybe best to add after. I don't know I just don't have the capability right now to understand it and include it.

River line vector textures have the shoreline as part of the river texture. It should be possible that a shoreline texture for lakes and major rivers can be processed in the vector processing with rr, roads, etc. Just use the vector line(shoreline) of the lake and river area vectors.

[sdsbolt]

yeah rivers are a bit of a problem but your idea may work. Is there a river shoreline with alpha channel to be borrowed from somewhere?

oh wait, I remember the water mask will end up on top of this new vector shoreline creating a sharp water edge unlike the existing shoreline which lays on top of the water mask. Maybe someone knows a way around this??

[rbp71854]

hmmm. Maybe if the vector is placed landside the water line on a second line. Will have to experiment on how much away from the water line and the width of the vector. Not sure how to generate that second line easily without having to do it one point at a time.....yet

oh wait, I remember the water mask will end up on top of this new vector shoreline creating a sharp water edge unlike the existing shoreline which lays on top of the water mask. Maybe someone knows a way around this??

[rbp71854]

Oh my word! gold! I can't believe I find this global mapper workfile on a hard drive which was part of a broken bootable raid and sitting in a drawer.

There are some missing elements to the file such as the topos, and dems used to create RS China.

The file includes streams, rivers, lakes, interior islands, shorrelines, ocean water, airfield and installations gl entries, etc. There is enough elements there that one could use to it further development if necessary in the future.

[sdsbolt]

that could save you a lot of work Bob

[rbp71854]

Yes Steve, China is basically finished, however it might come in handing if at some point if one wanted to add additional roads, railroad track, etc.

I don't know how this 30x30 1 arc mesh is going to work out for the PNG. Its still being processed after over 14 1/2 days. Still in 0256 resolution with 0128 mesh files still to go.

I have started to look at alternative dems just in case. If required it will mean redoing most of the shapefiles already finished.

So in the mean time I am working on squadron and aircraft lists with airfield locations.

[sdsbolt]

I don't know how this 30x30 1 arc mesh is going to work out for the PNG.

you should be able to use the same DEM and run ers2tiff with a lower resolution . I'm sure that will run a lot faster. Of course your terrain will have less definition.

[sdsbolt]

I don't know how this 30x30 1 arc mesh is going to work out for the PNG.

also why 30x30? I just create 1 big tiff file. I might create 2x2 or 4x4 for a big theater if needed.

[rbp71854]

I did create a large tiff file by exporting the dems then exporting a bil file from the tiff. I did this so I could make sure that the non-land area was at zero elevation.

This is the ers2tiff that I am running:

ers2tiff *.ers S7.53375 E149.9688 256 1024 4441 PNG -917504 -917504 70 70

You are suggesting I change the resolution by decreasing the 1024 to 512?

If so it will be interesting to compare the quality of the terrain.

Thanks for the suggestions.

you should be able to use the same DEM and run ers2tiff with a lower resolution . I'm sure that will run a lot faster. Of course your terrain will have less definition.

[sdsbolt]

ok, its looks like you are already using 1024 resolution. I wouldn't go lower than that.

i would try ers2tiff *.ers S7.53375 E149.9688 1792 1024 4441 PNG -917504 -917504 1 1

if that works well you can get better resolution with ers2tiff *.ers S7.53375 E149.9688 3584 512 4441 PNG -917504 -917504 1 1

or you can go even higher at 256 or 128

Steve

[gecko]

yeah rivers are a bit of a problem but your idea may work. Is there a river shoreline with alpha channel to be borrowed from somewhere?

This is an old concept WIP I tried out a while ago. Not great, but it gives the idea.

cfs3h2os.zip

[rbp71854]

ok, its looks like you are already using 1024 resolution. I wouldn't go lower than that.

i would try ers2tiff *.ers S7.53375 E149.9688 1792 1024 4441 PNG -917504 -917504 1 1

if that works well you can get better resolution with ers2tiff *.ers S7.53375 E149.9688 3584 512 4441 PNG -917504 -917504 1 1

or you can go even higher at 256 or 128

Steve

Thanks Steve for helping me out of this predicament. You nudged me in the right direction.

Presently running ers2tiff *.ers S7.53375 E149.9688 1792 1024 4441 png -917504 -917504 7 7

Processed through the night and most of this day, had to stop; before completing as I realized that my command line had capital png and my theater is set up with lower case. Also note that the theater needs to be divided by 256 so that the last 2 integers have to be 7 in this theater.


Thanks again.
Bob

[sdsbolt]

Also note that the theater needs to be divided by 256 so that the last 2 integers have to be 7 in this theater.

then I assume I am not understanding CFS3 terrain math. My thinking is 1792*1024*1=1835008 . 1835008/256=7168

[rbp71854]

The revised png theater is 1792km. I was going on the scratchpad.xls calcs. I know the srtm 30x30 .bil file had 17920 columns and rows which gave the 70 70 as the last 2 numbers in the ers2tiff command line. With the 1792 number I divided by 256 to get the 7 7 numbers, with 7 x 7 tiff files. We will see when I get the mesh into the theater.

I did try to run the command lines that you typed and received out of bounds error with the ers2tiff.

then I assume I am not understanding CFS3 terrain math. My thinking is 1792*1024*1=1835008 . 1835008/256=7168