+/- of F4 terrain system.

[Polak]

Generally the terrain in F4 is very well and realistically looking. It conveys feel of flying on higher altitudes amazingly convincing. I gather it has something to do with the right scale of the features on the texture. Out of many flight sims I have tried this one really shines despite certain age.

However, annoying repetition of the tiles somehow spoils the effect. Also large expanse of terrains of monogamous nature look equally bland.
Rivers and roads and rapid repetition of the same tile in the row do not look well.

Would like to exchange some insights or comments with someone who has some ideas how to cope with this and other problems.

[Snake Man]

Well as you create terrain look by usually with one terrain type and feature tile, it starts to look bit repeating. You can of course increase the variations by adding two, three or more sets of features (roads, rivers etc) and the main terrain type tile variations of course.

Then there is the dreamers of photorealistic terrains, I let them rant for you as I am not one of them.

Basically couple of variations by talented texture artist goes a long way.

[Polak]

Then there is the dreamers of photorealistic terrains, I let them rant for you as I am not one of them.

I guess, from the label "dreamers", that photorealistic terrains are not feasible. I am speaking photorealistic in the sense of say Google Earth terrain, or similar streaming system. So called photoreal in the "conventional" sense of tiled terrain is that the source of texture for each tile is ..say Google Earth. But the tiles still need to be grouped, seamles, rotational and what have you.

From some work I have done in the past for other flight simulators I know some of the basics of tiling etc.

What I am curious particulary in the Falcon case is to discuss how to make certain types of terrain kind of random and place them in between of the main types dictated by the altitude of its existance (this is I assume the main way how F4 tiles are being normally distributed) .

[Snake Man]

Dreamers as in that you need like 4096 different tiles etc

[ranger822]

Well, as a "dreamer" I feel I have speak up -- seriously I have to agree with Snakeman. Photorealistic terrain is a tough nut to crack but also as SM has said - good artistry goes a very long way. Until a new way of doing business with terrain tiles and such is created we have to use what we have. That being said - I do think that a closer approximation of photorealistic terrain can be achieved. Ripsaw is doing some nice tile work at FreeFalcon and here are some of the points I would offer:

1. Creating larger sized repeats of terrain - i.e. say 50 by 50 tiles - - and then stitching those together seemlessly - - you could conceiveably create terrain that would be most excellent if you fly say between 5-10 thousand feet and have a lot of nice detail. Obviously the higher you fly the further the horizon is and the more difficult it will be to disguise the terrain blocks that repeat.

2. To answer the tile repeats issue it should be possible to create tile sets (that don't follow CATE conventions but rather could be used as tile substitutions that could break up the repeates by making the tile appear more random.

3. Photorealistic terrain for specific items of interest can be made - such that you could photo tile something like a specific city or monument area and then blending that with the more generic tiles . . . still not idea but gets you closer.

4. Concentrate on a few terrain types and really develop them - ie. desert, mountains, or coastals and avoid the mistakes of previous tile makers of simply reversing tiles to complete the set - you really need to make each tile unique in the set and then plan/make some random tiles to throw in later by hand-tile replacement or if CATE or some other program will place/replace tiles with a random set. In MacFalcon we had a tool to randomize tiles after the initial tile work was done - - it worked off of either a single tile replacement alogrithm or using a four tile set - if it recognized the group it would replace the group with an entirely different but seemless set and could choose among a coupld of dozen choices - - time for running the tool - just a few seconds - then rebuild the terrain . . voila much improvement.

If you are really interested in terrain conce
pts - MadDog on his SAAFOPs site has a good discussion thread about his research on what is possible with Falcon code.

[Polak]

Thank you for interesting replies.

I think #4 is the real solution. Each tile should be done as a seperate entity and then "conformed" to CATE standards as far as that configuration of transition. I have analysed ODS tileset and see what you mean by simply reusing the same tile either by rotation or by flip.

How many additional generic nontransitional tiles (say desert1, desert2 etc...) can be added to each tile group? 16?

I remember that each group carries 16 , but if you over that number you can use the next group. 16 random desert tiles (as an example) could make the desert looking pretty difersified. I have heard that in order to get illusion of no repeating tiles min 20 is needed. However 16 is pretty close here.

What about that tool distributing those tiles during autotiling process? Is it available, or only for Mac?

[ranger822]

I would be happy to send you the program - unfortunately, learning how to use it would probably be rather difficult. Also, you need to have MacFalcon installed for it to activate. But, I found I can swap out files. One limitation is MacFalcon was limited to 64K Fartiles. So, the code, if ever obtained and ported would need to be updated. Similarly there is a 4096 tile limit for MacFalcon. Finally, it uses OS-IX (the previous OS for Macintosh prior to OS-X. I am not sure if it will run on the new G5 Macintosh machines or if they will run Classic OS-IX at all.

To my knowledge I am the only one to actually use the program in its final form before MacFalcon Developer Team broke up, since I was the last terrain guy for MacFalcon. I have some detailed notes on how to use MacTac (the program) but haven't touched it in over two years. MacTac is actually several programs rolled into one; TerrainView, Tacedit, Cate, several of the perl script functions and many of the theatermaker functions. Also, the workscreen allowed up to 1920x1200 resolution so I could work on my HD Cinema Display.

The program creator - Rick Prior was MacFalcon's counterpart to Juian Onions (Codec). Rick wrote the orignial LOD Editor from which Fred Balding ported to PC Falcon and the rest of course is history now.

One option is to try to contact Rick -(I think he might still be around but has moved onto other things now) - see if he will send the code for MacTac. I tried to arrange a code transfer from Rick to Fred Balding - Rick was willing but would only share it with Fred. Unfortunately, Fred wasn't interested at the time so the entire plan fell apart. If there was a way to port MacTac to PC we would be able to accomplish quite a bit of upgrading of the program. Similarly, MacFalcon had another program by Rick Prior called MacBrowse. This was the same as F4Browser - but again better in many ways. Incidently, Mac LOD Editor had a much enhanced graphic package which allows the same 1920x1200 screen resolution for Model work. Those who were part of the old MacFalcon know exactly what I am talking about here. If MacBrowse could be ported to PC - - we would then likely be able to make substantial improvements because we would then have the code support back in the noncommerical Falcon developers instead of resting with AF Executives (i.e. Julian and company). Talk about dreaming - these are definitely some very old dreams I had.

[ranger822]

I didnt' answer you question about the tile sets at 16. I am not sure but the quality of terrain isn't so much a function of the number of tiles use as it is multifactorial - good artwork makes a huge difference, but the placement of random tiles in additional sets also helps. Sometimes, terrain benefits form very small variations in the colors - when you have larger color separations I think it makes the terrain too easy to spot in repeats - whereas if you have a desert tile which very subtle shading differences it looks more generic until you get closer to the ground - then begin to appreciate the subtle differences.

Twister made some interesting terrain improvements in Peru Theater - - he applied some of the ideas of larger tile groups I think to better represent desert and mountainous areas . . .Mountains are a big challenge - you almost need to have an entire mountain range maybe 40-60 Km long - with unique tiles. Then transition on either side to your usual tiles. We experimented with phototiles in ITO - Death119 did some interesting work - what I appreciated most was that in order to make mountains look convincing you have to be able to run from horizon to horizon with unique tiles -- matching those up with SRTM however may provie quite a challenge. More research needs to be done with this to see what sort of effective mountains can be created with conventional tiles.

Adding tile sets shouldn't be a problem - updating the bin file and then adding the tile sets. I am not sure about tile limits now for PC - - maybe someone else can chime in on that.

[Polak]

To make terrain tileset looking convincing is tricky. I agree that keeping all tiles in small variations of color is the ticket to making it more real looking than not.

As an example here is an article from MSFS creators with sample of few textures choice. IMHO they compliment each other and their colors are not far off.
All is very strange kind of balance : if you to close the effect is bland , too far apart and suddenly it does not look good.

http://www.fsinsider.com/developers/Pag ... rrain.aspx

Thanking you for very interesting discussion and exchange, I would also like to share this article below. Its called "Pattern Reduction" and it was written by the same guy as article in the link #1. Jason Waskey MSFS developer and terrain specialists. Here:

http://www.gamasutra.com/features/gdcar ... waskey.doc

I hope we still continue this discussion....