TerraScene Resolution v AutoAsm "Reduced Points" Functi...

[victorwest2 0]

I know that Terrascene enables you to adjust the resolution of the output file. So you can have 7.5m or 15m or whatever other figure you want. I also know that AutoAsm Reduced Points function allows you to reduce the number of points in lines by half, third etc for improved FS performance. My question is this. If I had an image that was, say, 7.5m resolution and then I decided to reduce this resolution to 15m would that be the same as taking the same 7.5m resolution image and then using the Reduced Points option in AutoAsm to half the poinys? Or do the features affect totally different aspects of the image?

[Guest cwright]

Hi, Victor, that's an interesting question. I've alwys used the same resolution in TerraScene so I can't speak from experience. However, bear in mind that the TS resolution value works in two dimensions. Therefore, if you reduced the TS resolution by a factor of 2 then it should be equivalent to reducing the number of points by a factor of 4 (i.e. an AutoAsm Reduce setting of 4). Best regards,. Chris

[nudata 10]

I have used from 7 to 20 as a resolution value in Terrascene. The lower value produces much larger TGAs and takes forever to process in Paint Shop. I found that 10 meters/pixel was a nice value for filesize/processing. The size reduction values within AutoASM are the most convienent to use. At 7.8 meters in Terrascene and no reduction in AutoASM the processing time in AutoASM was about 25 minutes (1.6ghz 512mb). Additionally, the memory requirements rise and you may use pageed-memory which takes a terrible toll on time. So, try 10 meters in TS and 5 as a reduction value (LWM- & VTP) in AutoASM for starters.Dick Boley @KLBE

[Guest victorwest]

Thanks for the replies.Dick just as an aside, have you used Ifranview (or similar) to convert the tga's to bmp format and MS Paint to convert the bitmaps to 256 colours? Now AutoAsm includes it's own editor any amendments can be done within the program. I guess that PSP is a pretty intensive program as regards memory use and when you're processing a large file I'm not surprised it takes forever. Off the top of my head, when I've tried 7.8m resolution, the processing to 256 colour bitmap takes a few minutes on my XP 1800+ 512mb RAM.Victor

[Guest Slacktide]

>I know that Terrascene enables you to adjust the resolution>of the output file. So you can have 7.5m or 15m or whatever>other figure you want. I also know that AutoAsm Reduced Points>function allows you to reduce the number of points in lines by>half, third etc for improved FS performance. My question is>this. If I had an image that was, say, 7.5m resolution and>then I decided to reduce this resolution to 15m would that be>the same as taking the same 7.5m resolution image and then>using the Reduced Points option in AutoAsm to half the poinys?>Or do the features affect totally different aspects of the>image? My take... By reducing the resolution of the input image, you reduce the resolution of the output scenery not only in the number of points used, but also their placement. If you keep the image resolution higher, but use the "Reduce points" option, you reduce the number of points used in the output scenery, but keep the more accurate placement of those points.FWIW, my Long Island Coastlines scenery was made with a 20m/pixel input image. I found it was just enough to get decent resolution without bringing my computer to a grinding halt when manupulatioing the bitmap.Dan

[Guest cwright]

Hi Victor, I just thought about my reply again and I think I was wrong - not unusual! What I said would be true if we were dealing with areas. But we're dealing with lines, which are effectively single-dimension objects. Consider a single vertical road. If you halve the TS resolution then you will clearly halve the number of points (of course, an area would have only one quarter the number of points). I think your original suggestion was right: halving the TS resolution setting should be equivalent to using a Reduce value of 2 in AutoAsm. Best regards, Chris

[Guest cwright]

>My take... By reducing the resolution of the input image, you>reduce the resolution of the output scenery not only in the>number of points used, but also their placement. If you keep>the image resolution higher, but use the "Reduce points">option, you reduce the number of points used in the output>scenery, but keep the more accurate placement of those>points.Dan, that's a very good point (no pun intended!) I found that, when there are a number of adjacent half degree areas, quite a large Reduce value was needed to get smooth operation in FS (the main problem was delays lasting seconds when changing views). I use Reduce values of about 5. An equivalent reduction in TS resolution would indeed introduce a loss of positional accuracy. Best regards, Chris

[nudata 10]

But, doesn't the position of the line also suffer in TS. If the road were to follow a narrow canyon a low resolution in TS could cause it to climp the walls as the old Bill Gates data did. AutoASM is not effected by this problem since you are only dealing (I think) with space between points. The road may lack definition and in the the extreme become a straight line and would climb walls at bends. However, the remaining points are positionally accurate. In effect you deal with one dimension (linear) and TS deals with two (X/Y). A serious reduction in TS resolution can be more harmful than a high reduction value in AutoASM. Probably wrong but it is early here.Dick Boley @KLBE

[Guest cwright]

Hi Dick, it may have been early but you're absolutely right! Using Reduce in AutoAsm throws away a fraction of the points but the remaining points have full accuracy. A similar reduction in TS would result in a similar reduction in points - but I would expect the positional accuracy to be reduced by a similar amount also. I use 7.5 meters/pixel in TS and reduce values of 3 (LWM) and 5 (VTP). On my Florida test area (which has *very* complex coastlines) the detail is still excellent and FS performance good (mine's a 2400+, ATI 9200, 512 Mb RAM, Win XP Home). The slight disadvantage is that the bitmaps are about 8500*8500, but with those reduce values it only takes a couple of minutes to read them in. Also, a high res bitmap will help to stop nearby polygons getting 'tangled' with each other. I would certainly agree that the best approach is to use high TS resolution and appropriate reduce settings in AutoAsm. Best regards, Chris

[nudata 10]

I used Ifranview for awhile but had an issue managing the palette since AutoASM requires specific numerical values for the color of each item. Yes, Paint Shop Pro is "bulky". I have 512mb of memory and if I operate with 7.8m/pixel in Terrascene it becomes slow(1.6ghz). I use 10m/pixel since a point diviation of 2.2m over the default 7.8m is straightened out by AutoASM. If the TS resolution is too low the roads do start to zig-zag.Since it is easy to select a very small area in TS (lock to grid off) experiment and develop the values/process/tools that best suit your tastes. I spent months as Chris went through the various versions. I was determined to stick with it since I had a "visual" committment to synthetic terrain. I had done my own "photo-real" stuff but found that it was not as clear as the synthetic textures at VFR 2000 to 5000ft AGL. I also looked at the demo of MegaScenery and quickly came to the same conclusion. At 10000ft AGL photo-real look great but too low and the it becomes a fuzzy mess. Thus AutoAAM is the only program to offer very accurate placement of LWM & VTP things. By using the AutoGen annotator you can fix favorite areas in terms of proper items within textures and using something like SBuilder, or Ground2K, you can place landclass replacements to fix that farm community that Bill Gates believes is a city. Dick Boley @KLBE