Tuesday, August 28, 2007

CFS3 Marianas

I can now reveal the primary area that I am working on (there are two more) - The Mariana Islands. To date I have the terrain mesh, ocean polygons, & shorelines working. There are minor issues such as airfield facilities distorting terrain mesh near coastlines - just a matter of playing around with airfield altitude in the global layer. As I mentioned in the previous post I have found that hand editing or hand creating coastlines works best.

Next I move on to landclass & scenery objects, that will take me a couple of weeks. Presently the terrain only has one woodland landclass just to get the terrain & ocean into CFS3.

Here are a couple of pics - one of a rather bare Apra harbour and the other two of Pagan Island, which Japan held until late in the war and used to raid Saipan.



Rob

Sunday, August 26, 2007

Update Sunday

My CFS3 terrain experiments still continue. I managed to create a group of islands at an image stride & LOD of 64 meters which gave me a buzz this morning. What amazed me was that frame rates were barely affected. The verdict is out as to whether mesh at this resolution is worthwhile. The 64m LOD took all night to compile and takes about 600mb space. A 128m LOD (at 64m stride) only takes a couple of hours to compile and is about 140mb in size.

As to coastline data, I find that hand digitising of coastlines worked best in my current theatre. Of course such is a bit unweildy in areas with hundreds of islands. The problem with the CFS3 terrain engine & tools is that if a shoreline lines up with terrain at an altitude of more than a couple of meters it produces a sliver like spike radiating from the shoreline across the ocean, like a mini tsunami if you get down close and have a look at it. I notice a few in the Solomons theatre as well as my own experiments.

Anyway on to tidy up my coastlines - there are still a few glitches. The next step will be landclass & scenery, it's all very generic at this stage. This will keep me going maybe a couple of weeks.

Monday, August 20, 2007

Partial Success

I now have a couple of CFS3 theatres partially up & running. At this stage I have only completed the basics to get a map area operational.

Currently I have to commute daily to hospital for radiotherapy, which can take up to five hours out of my day if I have lunch & do a bit of shopping. Transport from my home on the island, to the hospital only takes just over an hour, but the return trip can take 3 hours given timetable gaps in the public transport system. Thus my experiments in terrain creation are somewhat punctuated. I also hope to resume aircraft creation once I am sure I have the main issues in terrain/scenery creation under control. All this is for relaxation LOL.

I have experimented with mixed results. So far I have yet to see any benefit using the CFS3 SDK with hires mesh, though I have more experiments to carry out this week The tiff2msh tool is where things currently fall down as to resolution. I also experienced interesting results with the landclass system which I will fully explore, once I have a nice terrain mesh working.

Also problems with water occur around some of the coastlines causing spikes in the water polys. I am trying to determine whether it is a DEM problem or a problem with some of the SWBD vector data, or whether it is a CFS3 terrain SDK issue or a Global Mapper issue.

I am being a bit circumspect in writing this. I'd like to disclose a bit more, especially some of my explorations as I go along. However with rivalry situations that exist from time to time in various CFS & FS communities I'll keep quiet until I have something to show. I am doing this for relaxation and don't want to be stampeded into competitive situations, which I can see happening.

The other person creating terrains & theatres for CFS3, Ed Wilson, is quietly getting on with his creations just like me, and I have no problems with that and I admire his achievements. After all MAW did inspire me somewhat to experiment with terrain/scenery creation. There are others however, who are prone to creating competitive situations.

One thing I will say though, is that those who liked what I acheived with the CFS3 & FSX Bristol Fighter, can look forward to the same approach in the world of scenery.

However one thing at a time.

cheers

Rob

Sunday, August 12, 2007

Update

Following on my last post it appears that I am not the only person working on a Solomons theatre for CFS3 as witness the current news on the Korean Skies page.

In many ways that is good news and I am generally pleased. I don't object to anyone else making terrain for the Solomons or elsewhere. There are heaps of areas I want to create & experiment with and I could have possibly chosen an other area to start with. Anyway I wish Ed well with it.

I have learned a great deal in the last week. The CFS3 SDK tools to convert DEM files to mesh appear to work ok. They even allow the creation of rectangular as opposed to square maps. Unfortunately this is not the case with tool that converts landclass maps to lcf files. It only accepts square maps. Why the rest of the SDK allows for rectangular maps and the landclass tool does not is beyond me... you should have heard me cursing.

I am currently working on water polygons derived from SWBD data. Unfortunately the data contains shapes for many of the islands instead of empty space. I have to convert many small islands to holes in the water by hand....painstaking & I still don't know how the CFS3 water tools will cope with such hi-resolution data......more cussing on the way perhaps? LOL

Anyway back to square one. I will still persevere with the Solomons area as my learning & test area given that I have come so far and carried out much research. I will at least get it into the sim. It will be interesting to see how it compares to the Korean Skies version. Whether I will ever release it..... who knows? I have a great many theatres large & small in many parts of the world from 1914 to 1960's that I would like to attempt. Of course given the contrary way the world operates MS might surprise us all and announce a new CFS.... that would be a long time coming if they did... plenty of time to create for CFS3 ;)

cheers

Rob

Monday, August 6, 2007

Ready to start Solomon Islands theatre

This week I am ready to start moving things into CFS3. My main goal is to get things working using as much of the original CFS3 installation as possible. Later on I can experiment with higher resolution data. However the CFS3 scenery system is difficult enough as it is without trying to improve things from the outset.

The region I am working on is centered at S7.5 , E152.5 and is planned to cover a similar area to CFS3 Europe.

I have been obtaining the appropriate data and checking it out and assembling in my primary GIS application Global Mapper. The CFS3 terrain SDK is certainly making sense now. However I still have to check out all the SDK tools. There's much work in all this & I don't want to overdo things. I also had a nasty flu virus last week which delayed matters. Hopefully I am on the mend now. Radiotherapy begins this week so I'll have less time for creative endeavours.

The data is split into 3 main types -
  • topological DEM data - to produce terrain mesh.
  • vector data - for coastlines, waterbodies, roads & rivers.
  • land coverage data - for the landclass system.
The landclass side of things is still WIP at the moment. I want to take a look at the existing CFS3 landclass system & textures before moving ahead with mine. I also have to find a way to convert from USGS data to CFS3 data & line it up on the map. I will add a few new landclass types to CFS3s existing types, but not too many as that will entail a lot of additional work. I will be refering to period maps too, as land use & coverage in the Solomons area differs today from that of the 1940's http://www.lib.utexas.edu/maps/historical/pacific_islands_1943_1945.html Much of the landclass info will be hand edited in a paint program.

As to DEM data, I had originally intended to use the most recent available 3 arc second data. As this is 10 times higher resolution than the data used for CFS3, I am not sure that the CFS3 engine could cope with it over a region as large as the CFS3 European theatre. One has a number of choices : -
  • use that data and then use the numbers as in CFS3 & SDK examples. This will leave any resampling downwards to the CFS3 tools. The results of such would not be known. A lot of work just to experiment.
  • use SRTM30 data. This is 1km data similar to the resolution used in CFS3. However this data is more recent and accurate than the data used in CFS3.
  • use the 3 arc second data & down sample the DEM data in Global Mapper. This is what I have decided on as I will be using the most recent data available.
I intend to further experiment with hi resolution data later and would love to do this also with textures, however for now, using the above approach I will be able to create a theatre the same size as the CFS3 Europe map, using the same figures in the xml files and SDK tool examples.

Vector data was a big chore last week, especially coastline data. Rivers & roads are fairly easy. For this exercise I am using vmap0 data, available from NGA via their rasterroam site. It is easy to extract this data using Global Mapper (the Swiss Knife of GIS applications.). Water body vector data is easy too. I am using SWBD data derived from SRTM data. It is good resolution, though perhaps stepped along the coast lines. I'll have to see how it looks or whether it is noticeable in the theatre.

Even though we have this data we still need vectors for the coastline, this is where the ordeal began (along with the flu), involving many hours of trial and error. I found the vmap0 & NOAA coastline data to be too lo-res & didn't line up with some of the smaller islands. The SWBD data is in area format. The coastline data has to be in lines. Global mapper's convert area to lines function caused some of the smaller islands to disappear, probably due to flaws in the data and we are still left with the problem of deleting the lines comprising the tile borders, which also deletes parts of the coastline in Global Mapper. I needed to find some other way of editing the vector data and extracting and converting to line shape data. The solution - export the vector data to dxf format & import & edit in 3DS Max (this may work too in gmax, though I haven't tried). This was somewhat unwieldy, time consuming and involved much trial & error. Eventually I got there, imported the edited DXF into Global Mapper and then re-exported as a line vector shapefile, now ready for the CFS3 sdk tools. The SDK shoreline tool is reputedly finicky & again much experimentation will be required to find the correct resolution.

A few links

DEM data is obtainable from a number of sources :-
Other data sources :-
Hopefully I'll have a rudimentry theatre up and running in the next 10 days or so.

Rob