fs2k2 and beyond: The Future of Scenery Development

[shermank 170]

I am no techie, have no abilities to design and make scenery, aircraft...etc...merely an interested and long time user with an enormous respect for those who designed the original program and those who have given us the great add ons.My, question, with the above stated, is What is the Future of Scenery Development? Even the best that there is has a cartoonish quality. This is not a criticism, but an observation of what I see as the present level of technology. Yes, there are some "photorealistic" sceneries. But, even those somehow do not look 100% real, as if we were flying in a photograph, or more accurately in a motion picture. When do you think the technology will be available on an affordable level to reach this next level of realism? Thanks for any thoughts.Sherman KaplanHighland Park ILN3534 on Vatsim.DC 626 on dc3airways

[John_Cillis 2,406]

The short answer: 2-3 years.... But there is one huge issue--that of data storage. Photoreal scenery is a real hog of data. A simple 400 sq. mile area I did contained over 20 megs of textures, and still the outcome wasn't sharp enough to get over the cartoonish quality you mention.Yet great strides have been made.... Fly around FS2002 with some high cirrus, around the deserts at dusk, and it looks very real.... Same with the basin and range country in Nevada. We still need work with clouds, which can add much to realism. And work needs to take place with shadows, lighting, to add more depth, which helps remove the cartoonish feel.I don't think FS2004 will meet these goals, since I don't think systems will be quite up to the task during its design phase. In 2-3 years, hopefully we'll have the improvements in I/O speed to make our wishes reality. In tandem with that, I suspect we'll see new players in the sim arena, as I believe new graphics cards will have easier programming support....

[barryward12345 0]

Also related to this question of scenery development is another one -- when can we expect the world scenery to be equal -- ie not as "Americanocentric" !! :) Flying around Australia currently looks nothing like "flying around Australia) :(Barry

[Guest]

Hey folks,If you are flying in the UK you might be intersted in this...http://www.visualflight.co.uk/photoscenery/faq.htmhttp://www.visualflight.co.uk/photoscenery/faq.htmEnjoy

[michal 44]

My personal opinion - it will take another 8-10 years to move us to something that you could "confuse" with real-life scenery. And I think it will most likely be a synthetic scenery that is indeed very good rather than a real-life scenery where every pothole, every tree, every creek, every house, every outhouse is represented in its right position. And I agree with others - we can certainly much improve on the weather - and this should be happening relatively soon.Michael J.

[Guest]

If you remember that even full motion simulators don't do that, and they have a lot more computer power than we do, you'll see one of the main reasons.Another is indeed storage.Getting photorealistic textures that stay that way from groundlevel up to the edge of space (someone will always complain that it doesn't look right at his favourite altitude if you do anything else) requires a huge amount of storage and consequently a huge amount of processing power to determine which texture to use at what altitude (mipmapping merely combines several texture files at different resolutions for ease of archiving and loading).Ideally, you'd have a 1cm (though 10cm might work well) mesh worldwide with the true colour for each season and environmental condition embedded in that data.There would be no more need for textures, as all the data were inside the mesh itself.An algorithm could then calculate the smallest size to display in any colour (thus, how many datapoints fit a single pixel on screen) and calculate the average of those datapoints and output it to the video.I'm not quite sure what you'd need, but I do know you'd need a lot more power than can be expected to be available to your home computer soon (not just CPU speed, but RAM demands would soar). It probably could be done, but only for static scenes (the problem is generating it all quickly enough, which means 15-20 times a second at least).Add to that that you'd want multiple screens each running at a very high resolution and high framerates (everyone seems obsessed with getting hundreds of fps), and the power required goes up even more.

[Guest Leesw]

If Moore's law continues as it has then we should see about a 100 times increase in processor power, video processor power, and memory in the next 10 years (e.g. FS2014). That should be enough to provide Autogen'ed scenery detail down to the level of street curbs, lamp posts, picket fences, runway lamp casings, and some kind of Autogen'ed ground textures that have fine 3D detail like clumps of dirt and grass that actually has some thickness and looks reasonably realistic close-up.These things might not be important to guys who spend most of their flying time in the stratosphere, but it would enormously enhance the 'immersion' level for people who fly low and slow, and even when just taxiing at the airport. The Autogen'ed buildings and trees helped considerably in FS2002, but the ground and neighbourhood streets still look like flat, fuzzy blobs from below 5000 ft AGL.

[michal 44]

Assuming we are today at 4 m resolution and where we should be going is roughly 5 cm resolution (you won't be able to see individual grass but lamp posts and picket fences would be there - a simple calculation shows it is roughly a 5000 times increase in data content. I am afraid we will have to wait a very long time for something like that ....100 time increase would give us about 40 cm resolution - perhaps not enough for potholes or lamp posts but it would be spectacular nevertheless ... Airports could always be done in higher resolution. Michael J.

[Guest Leesw]

I'm not suggesting rendering every blade of grass for 100 mile radius, but there are ways that the ground can be made to have the appearance of depth and not be so darned flat and out of focus. Even if it only means having the generic photo-real textures in steps of resolution that kick in at different distances. So when you are looking out the cockpit taxiing on a grass strip you can actually see blades of grass, with a small change in terrain levels at the edges of the grass texture, and maybe a few rendered weeds to enhance the depth perception.MS should consider making a general purpose scenery modelling engine and database that can be used on all of their (future) outdoor games, from RTS games to driving games to flight sims, that would allow them to maximise their development efforts. Instead of having to have every game team develop their own stuff. Have one team that creates one big model of the world in high detail and each game division could create whatever localized specific details are needed for each game. Why not have the entire highway system accurately modelled, then you could then drive all across North America or Europe in Midtown Madness games (or maybe a large scale Cannonball Run type game) or fly VFR over the same terrain in FS2014, or take the train (e.g. Train Sim 2014).Or have CFS7 integrated with ground and naval war games and Rainbow Six type games. The Earth model could be back-dateable so you could have truely large scale Age of Empires type games. The potential would be enormous. You could go anywhere in the world in almost any outdoor game.

[michal 44]

>photo-real textures in steps of resolution that kick in at >different distances. I know what you are talking about. B-52 radar simulation I worked on long time ago at Singer-Link used precisely this technique. The ground was stored at a few different resolutions and depending on the range distance selected different mapping files would be accessed. However I am not sure how well it would work for the scenery. Such 'step' resoultion changes might not look too good. I think completely new technology is required that would provide seamless resolution change. I don't think we will see anything of this kind in the next 5-10 years. For this to happen truly revolutionary ideas in storage, bandwith, video-display, CPU-memory are needed.Michael J.