1 hour ago, CO2Neutral said:

Ortho sources are hit and miss and often low quality.  Sat imagery that is color matched, cloud removed, baked shadow removal, seasonal, and high resolution is extremely expensive, I priced out Greenland once (from UP42 at 30cm) and it was about $6M (yes million US $$).

It would be cheaper to have a globe represented by procedural trees/buildings/terrain based on accurate world data but not world imagery.  Procedural buildings/trees/terrain would solve many problems:

• Can change gradually with seasons (rather than all or nothing at a specific date)
• Isn't subject to baked single time of day shadows (shadows from sat imagery + shadows rendered by the simulator)
• Superior road/building/tree blending (freeway under pass or bridges that are not texture walls)
• Operate on the same light sources and PBR materials so have world consistency
• Can have persistent terrain changes (i.e. flood wipes out an area, or hurricane hits, earthquake )
• Can evolve the human expansion (or contraction) on the planet over time to real world accuracy
• Retain accuracy to the changing climate
• Use far less storage space

In the meantime, we live with what is cost effective and low CPU/GPU resources (lowest common denominator) ... I was shocked to discover the average GPU is a 1060 according to latest Steam survey that auto detects a users hardware.  

 

All right, I always put my shadows off when I download Ortho. What you can suggest? How we can do ground imagery suitable for our sim and at the same time not very labor intensive. 

20 hours ago, PavlinS said:

What you can suggest? How we can do ground imagery suitable for our sim and at the same time not very labor intensive. 

Don't use photographic imagery.   

We have building data (size, shape, location), road data (width, type, location), terrain type data, tree type data, etc. (that data is available and free) ... map that data to dynamic 3D rendered objects using native texturing (rather than based on an photographed image) and place those objects in the biomes/locations matching the data.  

You end up with roads that blend well, traffic that can stay on the road rather that float above or below it.  Trees that look like actual trees and not some mash of texture array.  Do away with different meshes and texture mapping that creates unrealistic blurring of a texture because it's being stretched over a bad mesh (think mountains, cliffs, etc).  Do away with the entire UV mapping of 2D image on a 3D model.

End result will be a lot less VRAM usage, most likely higher performance thanks to instancing (render variants of the same 3D model), and accurately shaded materials.

Who wants to fly a world where it represents just one snapshot in time?  The world is not a static environment, it's constantly changing.

 

@CO2Neutral I don't understand still what do you mean. Okay, let's use the vector data, make roads, make airports with it and make every single building and tree [ which we already do with geographical polygonal / lines and points data anyway]. Tell me how we can export all that without imagery? Where do we place this if there is no imagery? Even MSFS2020 did not found a way to work that problem without imagery, they use Bing and there is even Google imagery DLC mapserver used there as well. 

I am sincerely curious - how we can model the whole scenery without imagery? On what at top of we place all that data that we handcrafted. You simply said to me that we need to create handcrafted photoscenery. Right now I am doing exactly that - I am doing polygons for buildings in Jakarta area in Indonesia to create photoreal imagery with handcrafted terrain. How do I do that without the imagery? It is simply illogical what you are suggesting. To place all that data on the random scenery of P3D world without any scenery data? You do realize that all that handcrafted data is exported as autogen [ .agn] files, which is type of a texture that contains vector data that I created [ as a .shp or so-called shape file]. So, basically you are saying " create the textures, but do not place them on scenery imagery". That is like, exporting scenery without the scenery itself. 

2 hours ago, PavlinS said:

It is simply illogical what you are suggesting. To place all that data on the random scenery of P3D world without any scenery data?

Seems to be a disconnect ... I didn't say anything specific about how to do this in any existing platform (P3D or MSFS) ... you read in what I did not say (an assumption).

My comment was how to approach a different world design paradigm for some future flight simulator moving away from the issues brought on by photographic imagery (which are many).  Moving away from current platforms into next generation platforms, we need to move away from photographic images as texture bases for all objects (terrain, buildings, roads).  20-60 PT of imagery data just isn't going to be realized any time soon for a flight simulator, and doesn't need to be, there is a better way, procedural.  All these objects can be generated with material constructs and NOT photographic image based constructs (UV mapping).

The biome/eco data, road placement data, building data are simply data descriptions of the objects (length, height, width, class, type, location , etc.) but is NOT image data.  You would have only a contour/outline data that defines the globe with appropriate attributes to identify what type of objects should be rendered (i.e. grass terrain, dirt terrain, etc.).  The render pipeline constructs these objects just like any other 3D graphics pipeline, only its using materials on these objects not image based textures.  This gives far more control over the lighting that can be applied to materials and provide for a much more realistic environment.  I also allows for objects to weather in real-time.  You have proper seasonal gradients and not just one day it's green and the next day it's brown.

Perhaps it's easier if you think of this similar to P3D or XP or MSFS land class titles only no photographic images and the tiles would actually make sense and not be disjointed.  The world gets pre-rendered at a primitive level then as one fly the world is generated and cached over time and if world events change the cached scenery/world is updated just at those in LOD to a view.

Fundamental shift on how to render the world and solve all the problems I pointed out.

 

Quote, "Who wants to fly a world where it represents just one snapshot in time"?  

Yes please, a one time image to me is far more realistic than the Lego brick alterative that has never even existed.

Phil

12 hours ago, CO2Neutral said:

Who wants to fly a world where it represents just one snapshot in time?  The world is not a static environment, it's constantly changing.

It's a flight simulator, not the real world and by definition, it must be a snapshot in time.
The only way to get round this, is to go outside and do something real, instead of simulated.
As this option is closed to most of us in respect of flying an aircraft, the simulator is the next best thing.

9 hours ago, CO2Neutral said:

My comment was how to approach a different world design paradigm for some future flight simulator moving away from the issues brought on by photographic imagery (which are many).

Your comment lacks sense. We are discussing why not developers to create project together here. You are discussing the " fundamental flaw of creation of scenery for flight simulator". Too deep my dear. Myself don't understand you. Nether what you're saying. So stop shifting  the subject into something that is not. 

9 hours ago, Reader said:

it must be a snapshot in time.

Why?  I'll disagree, hence what I described in how to make it a truly living world and not a disjointed snapshot of various time periods.  There is no technical barrier to moving away from static photographic images to native materials.  In fact, this is the likely progression forward to the next flight simulator platform for the many limitations is described above using statice photographic images.

7 hours ago, PavlinS said:

We are discussing why not developers to create project together here. You are discussing the " fundamental flaw of creation of scenery for flight simulator".

Why do you feel the need to control the flow of conversation?  Anyway, technically the P3D visual engine can be replaced entirely, hence P3D's support for CIGI (read up on it in the SDK/PDK here).  So yes, the community could implement what I was suggesting if a group of us wanted to take on the task (which BTW commercial companies have already done).

I've been reading along a bit here, and would like to briefly share my thoughts:

Your approach, @CO2Neutral, makes perfect sense - Sort of "rebuilding" the world, based on the data we already have.

I think this is a good way to fix a lot of problems (large amounts of data, shadowing, blending roads, etc, etc...).
This will also look better - IF an engine would be able to reproduce in detail the light scattering, color rendering, all different surfaces, calculate all shadows (keyword raymarching reflections) in real time, take into account all parameters (weather, temperature, sun position, atmospheric thickness...) AND IF worldwide data is available for this - in the centimeter resolution range. Not only the ground surface, also the atmosphere. 

And that does not work until now. No PC in the world can do it, no remote sensing satellite can deliver this data, no engine can represent the world with such performance.

On a small scale, it may work great - but not yet on a global scale.

Therefore, one simply has to say that the use of orthophotos (i.e. georectified satellite and aerial images) is still by far the best solution for "us simpilots" - and probably also in the medium term.

Edited September 8, 20232 yr by MatthiasKNU

1 hour ago, CO2Neutral said:

Why do you feel the need to control the flow of conversation? 

I don't control anything, if my contribution does not like your taste this is not the place to bring it up. From what I am seeing you are suggesting that we should work " other way" - tell me, how to work other way and achieve the same results without following the same methodology? That's simply impossible. We want to make scenery that contains buildings, routes and trees, right? For P3D? Then, we need to create Autogen data that represents them. Like the autogen packages of @MatthiasKNU - he handcrafts the data, then exports it with ScenProc, creating his own script to do the job. Same as me, I do polygons on Bora Bora:

 

Then I open the ScenProc, create a script:

 

then Scenproc chew on the data I imported and creates what I want to create with it:

Also FYI every single message I publish here is approved from Moderator. So I don't wanna control the conversation. I am just contributing by the basiss of what I think, the fact that you don't like it does not mean I am " Trying to control the flow of conversation" as you blame me for. 

This is the most simple way to create autogen for scenery. As you can see that autogen sticks to the place flawlessly. So I don't understand why you have a problem with the satellite imagery, when we can create scenery for the place to cover the imagery. 

Hey Guys,

Just to jump in, there are additonal options for selected scenery.

So say you've been a long user of Orbx and have their scenery enhancements for the land class types, trees and buildings.

Looks Ok but at altitude it's quite repetitive and lets face it, a photoreal set of base textures would be nicer, which is why Orbx were producing their TrueEarth series and why other developers are also performing similar.

In between all that sits a layer of alternate scenery, although not of a high resolution.

And there is a developer who has managed to utilise the alternate to create scenery that blends with the underlying base, where the base is blended to photoreal for distance view.

He calls it "Enroute" scenery and it does enhance whatever base scenery you're using quite well and for quite large areas at an inexpensive cost.

https://secure.simmarket.com/gibson-sceneries-for-flight-simulator.mhtml

Here is Victoria Australia where my Orbx base is showing and blends in to the photoreal thus providing a more realistic view out to the distance while still showing the Orbx base (AU2, Terraflora v2 and HD buildings).

This one shows more of the circle of base scenery and the distance blending.

At altitude in the Northern Territory Australia, from the cockpit window the base scenery is typically not in the circle of visibility.

Cheers

6 hours ago, Rogen said:

He calls it "Enroute" scenery and it does enhance whatever base scenery you're using quite well and for quite large areas at an inexpensive cost.

You might had found something important, but I pref. to create scenery the old fashion way. It is the most easiest way to combine Ortho4XP for P3D Imagery and meshes with creation of polygons data and send it in ScenProc. That by my opinion is the easiest way to create somewhere photoreal scenery and this is just a single cost. I plan on distributing my scenery for free, because what I am trying to do is on amateur level still and I do not have the audacity to sell it, but this guy offers too small amount of data. 

My personal opinion is that Asia should be the concentration of the P3D world, because this is the only place around the globe in P3D where only airports are produced. It needs terrain, which is why 6 out of my 7 sceneries for the world trip are located there. It is massive amount of work and if we all gain some recognition and develop for there that will be awesome, because Asia is left behind and really needs terrain the most of it. 

Edited September 9, 20232 yr by PavlinS

15 hours ago, PavlinS said:

You might had found something important, but I pref. to create scenery the old fashion way.

That's fine, my post was more about inovative methods some people are using to produce commercially viable products.

15 hours ago, PavlinS said:

My personal opinion is that Asia should be the concentration of the P3D world

Yes it's long been a neglected area, Gibson Sceneries does include China though.

Cheers

8 hours ago, Rogen said:

Yes it's long been a neglected area, Gibson Sceneries does include China though.

I seen the scenery. No objects at all. And a scenery without objects, my personal respects, but it is not a scenery to me. Also FYI one of my projects is handcrafted scenery for whole China. it will take years, but I can promise you, the final result will be better than this one. 

Edited September 10, 20232 yr by PavlinS

13 hours ago, PavlinS said:

I seen the scenery. No objects at all. And a scenery without objects, my personal respects, but it is not a scenery to me. Also FYI one of my projects is handcrafted scenery for whole China. it will take years, but I can promise you, the final result will be better than this one. 

Yes of course a scenery fully decked out with objects is the ideal, but as you say that'll take years.

This particular scenery intent is to complement and enhance the default base or Orbx scenery, so the underlaying scenery (with objects) shows through for low level flight and at high level the photoreal is what is visible. Yes it's not perfect but is a useful compromise at an affordable price.

It's a somewhat unique concept, plus is also commercially available now.

Cheers