FMSXYZ New Version V1.1.0

[Perry]

Ok one thing at a time.

Firstly inserting a sky texure.
If you have not done so, first make a new folder where you will assemble your new scenery.
Yes you need to make 16 bitmaps each 256 x 256 of your panorama.
I might be able to help you with that if you send it to me. You want to take a stab at it then you will need a good paint program that will allow you to scale your scene first. Break it into chunks of 256 x 256.
After you have finished that, open FMSXYZ and create your new scene.

From your new scene dialog box, I suppose you chose a panorama scene option.
Three tabbed choices are listed below.
Sky,Height Dist , & Landscape

Choose Sky. You will have a single box below it. Click inside the box. You will have a file requester where you may search through your directory for a sky texture. Any might do for now, borrow 1 from another scene. You can easily change it later.
It should be a 256 x 256 sky map, but I believe it will work with 1024 x 1024 also.

All picture boxes work this way in FMSXYZ. Just click inside to change it.

After loading a skytexture, go on to Lanscape. It is not necessary to load a Height Dist Map when you have a panorama. Again click inside the boxes. You will get a file requester which will allow you to load in scenery.

After Naming your scene click OK and you will be able to save it.
This will be the core of your scene where you can start adding objects to your scenery.

You should be able to fly it at any time from this point on.
Let me know when you have caught up to this point.


Perry

[TreeHugger]

[Perry]

[lcsaba]

[TreeHugger]

[Perry]

[Perry]

[TreeHugger]

[lcsaba]

[lcsaba]

I would like to report on two things and ask two questions.
1. I have made a test with FMSXYZ. I converted the pictures of St_ Oswald by FMSXYZ to a .scn file. Edited, to have decimal points. The result is a little bit different than the downloaded file but it is OK.
2. On the basis of this knowledge I have made my first attempt, to create my airfield from scratch.
There are 2*8 256*256 .bmp files. And the resulting scn is working. But the outcome as a whole is terrible. There are trivial errors, the edges of pictures are not mapping, even it is true in vertical direction as well. I have no idea what is the reason of that, (I am trying to fix the three pod to be vertical) but hopefully I will find a good photo taking method.

First question: My problem is that with a panorama picture the distance information for the viewer is some knowledge of the sizes of various objects. For instance it is not nice if a model flying in front of a remote building is smaller than expected.

I guess that the FMS is calculating the size of the model as getting far or aproaching. The size of the model someway should match to the vision of the panorama picture.

All of the landscapes transformed from Reflex are at the QUADS definition 1000 ... -1000 sized, because, I guess, the distance between the pilote and the remote surface of the panorama objectes is nearly 1 km. In my case the diameter of the panorama picture is nearly 500 m. Threfore the remote objects are too big, when FMS displays them.

Can I set somewhere this fact using FMSXYZ, I guess that edit the QUADS at the scn fike is not correct?

Question 2. I have studied using 4H_ Camp and St Ostwald, what happens when the model is flying behind obstacles.

I think the absolutely perfect solution would be quite difficult. If at 4H_Camp.. the TRANSPARENT command is deleted the crash objects can be seen. The solution is correct, there is a collision when the model is going far from the standing point of the pilote, to the direcrtion of crash object, and if approaching from the other direction.

The problem is, that if the object is TRANSPARENT the model can be seen if it is behind the object. The reason is, that the vision of the object is on the surface of the panorama picture, meanwhile the position of the TRANSPARENT crash texture, is calculated.

My question is about the Reflex solution. I have made experiments with the demo version of it and have a feeling, that if the model is behind the object (tree, building) it disappears.

Is it true, and what is the solution.

As far as FMS is concerned I know a possible solution. The picture of the object, (a good example is a man standing in the front of the StOswald scenery, and now the model is flying across the body of him (because thre is no crash object defined)) should be a visible texture over the transparent texture of crash object, like the trees and building at the FMS landscapes.

SORRY being so long.

[TreeHugger]

Hi,

I'll try to address the issues one by one:

1) The result looks weird, because the St_ Oswald scenery is made to be mapped onto a sphere while FMSXYZ uses cylindrical mapping. I know for sure, because I made the mapping for this scenery initially to be used with "Aero Club Bad Oldesloe" (and the author "borrowed" it without asking...).

There is a mapping difference between cylinders and spheres that becomes ever bigger the further from the horizon you are. You'll find a lot of valuable info at wikipedia, in the mapping section. Look up "equirectangular projection" and "mercator projection" to find out why this matters.

2) Eight textures of 256px is by far too small. Try with 8 * 1024px, gives a horizontal resolution of 8k pixel, that's OK. To get everything matching, you need to create a seamless panorama first (e.g. with autostitch). This is not a trivial task, you'll need to consult some tutorials to get it correctly.

If the visual impression of the model size or some landscape object is wrong, then you see a mapping error of the pano. You can rely on the camera calculation of FMS, it's works correctly in 3D-scenes. If dimensions appear wrong, your pano is either stretched (mapped incorrectly) or it has a horizon problem (wrong height or tilted).

The size of the pano scene has nothing to do with this. When you are in the center of a sphere or cube, it will look the same no matter whether its 500m or 1000m in diameter. Double distance and double size equal out each other. The big size is just to make sure you don't hit it normally.

The whole collision object thing is a different matter alltogether. It's too early to bother until the rest is working properly.

Good luck!

edit:
Another pano scenery for you to study: "Modellflugverein Große Heide 73" or MFV73. I made the FMS scenery, the pano image already existed for Reflex. No collision points because there is simply nothing you could crash into. You'll need to convert all .jpgs to .bmp format.



Regards

[lcsaba]

Thanks,

it is a great help for me.

There was an misstyping in my message, my pictures are 1024*1024 as it is required by FMSXYZ, therefore this is OK.

You wrote:
: your pano is either stretched (mapped incorrectly) or it has a horizon problem (wrong height or tilted).

This can be my problem.

Does this means that the upper 8 pictures should be taken with a horizontal camera?
I made on intent tilted to have enough room for the lower 8 ones.
How to position the camera for the upper and lower 8 shots?

The lower ones should match to the upper ones.

I will study the dowloaded fields.
Yesterday I tryed to download from RC-sim.de a hall, to see how the pictures are composed if the distances are small, but failed, because a very low dowload speed. I will try again and the Grose Heide as well.


Regards


[/quote]

[TreeHugger]

Hi,

eight images horizontally are usually not enough to get a decent resolution. This tutorial about pano-stitching explains the whole process:


It's available in german only, you may need a translation tool:
or

Regards

[lcsaba]

Hello,

I have made some experiments and some progress.
I have made 2*8 pictures with a horizontal camera, and transformed them to have 16 pieces of 1024*1024 pixel pictures.

Using FMSXYZ the landscape scn file has been produced, but as You said a cylinder, in fact a prism with 8 faces is not a good resolution.

I have selected a very primitive method. I have taken the mvf73 landscape, which is by the way a mystery for me, but at least there are 8 pictures taken horizontally. I have replaced the photos with my ones and got a quite resonable result.

Now I have a new challange. The ground level is defined somehow, the shadow of the model can be seen on the ground, and mfv73 is correct, in this respect, because the collision with ground seems to be at the ground level of the photo. This is not generally true in the case of other transformed reflex landscapes.

My trouble is now that the collision on my landscape happens approximately at 1.5 meter height. By the way this was my camera position.
You might say that I have to study the literature.

OK.

Regards

[Perry]

The next version of FMSXYZ will allow you to adjust this, but for now you might have to do it manually.

You can adjust the last data in each line of your pano + 1.5 which I believe will yield the result you want.

I will have to experiment with this.

Perry