New FDAI tex 1800x900

chrival

Active member
Joined
Sep 7, 2016
Messages
25
Reaction score
37
Points
28
Location
Vaucluse
Hi
I was aproached by n72.75 about 2 weeks ago on the french forum and I volunteered to build a new hi resolution FDAI texture.
After considering a 1440x720 tex (4px/deg) I eventually opted for a 1800x900 (5px/deg)

Unfortunately I can't show you this one in NASSP since I'm unable to load it. It seems that it does not accept my file.
In fact, I'm not familiar with .dds format and while exporting it from Gimp, I'm asked for several compression options. I tried some but with no success. I still need help for that.

So I put it in the ADI MFD instead.
I cheated a little bit because this MFD does not accept a tex higher than 1024x1024. So I cropped (instead of resizing) the 1800x900 image to 1024. At least, you can have an overview.
This trickery is the reason why the rendering looks like a bit distorted.

FDAI_1800.jpg


However, I resized the 1800x900 to 1024 too and then I can show you more :

FDAI_1024_0.jpg
FDAI_1024_1.jpg


The 1° graduations still can be improved. Easy to do.
I didn't yet made the digits and 5° dashes because I want to test it in NASSP before.
As far as I understand :
  • The VC accept 4k tex and according to n72.75, a 1800x900 FDAI should no experience any trouble.
  • The FDAI on 2d panel can go up to 1024x512 and the current one is only 512x256 that is far too low. Such a small resolution cannot display the 1° graduations at all.
Besides I can't see no reason why the 1024 tex from above would not work since the result is quite acceptable on an MFD.

This texture has been entirely made from scratch.
I did my best to match the authentic design and colors, using pics from the net.
Mainly these ones:

ap15_fdai_2_450.jpg
ap15.jpg
fdai.jpg


Hope you like it.
I'll show you more as soon as I find the way to get it in NASSP.
With digits and all, LM FDAI ...
 

Attachments

  • FDAI_1024_1.jpg
    FDAI_1024_1.jpg
    109.7 KB · Views: 52
Last edited:

Lupus_Vulpes

Member
Joined
Jan 29, 2021
Messages
45
Reaction score
23
Points
23
Location
Germany
If you want to test it in NASSP, the easiest way to put it in is...
Open the FDAI_ball.dds file with regular old Windows Paint, paste the new texture in, and just save the file. It magically works. If the resolution is larger than 1024p it won't load in the 2D cockpit, but it should still render in the VC.

BTW, the detail on this looks amazing, great work man. If you have any questions, ill gladly help. ?
 
Last edited:

chrival

Active member
Joined
Sep 7, 2016
Messages
25
Reaction score
37
Points
28
Location
Vaucluse
Thanks Lupus.
I'll try your method.
But, anyway I need the specification to import in .dds format from Gimp.
 

Lupus_Vulpes

Member
Joined
Jan 29, 2021
Messages
45
Reaction score
23
Points
23
Location
Germany
Im curently at work, will take a more detailed look at it when i get home...
The problem with the 2D cockpit is that it renders everything in 1024p, and when the ball rotates the rendering bunches up the pixels to simulate movement. In my textures i mitigated this "pixelisation" effect by adding just a little bit of blur effect, so that the transison of the moving ball looks a bit smoother. Thats sadly the best you can get out of the 2D cockpit, but the real magic happens in the VC, the 3D cockpit. You acces it by pressing F8 two times. There it renders the texture in its full size, so if you load the 1800px texture in game, you wont See it in 2D, but it will render in its full size in the 3D cockpit. The solution is to have two textures, one high-res for the 3D virtual cockpit, and one 1024 p optimized for the 2D cockpit
 
Last edited:

chrival

Active member
Joined
Sep 7, 2016
Messages
25
Reaction score
37
Points
28
Location
Vaucluse
Forgot to mention that the link from previous post is the 1024 px texture.
Since I still do not know the required format for the .dds to be accepted I can't upload the 1800 for the moment.

Thanks Lupus for your help, I followed your advice, I used Paint. The only difference I noticed is that the final dds image is unflatten. But It seems that it doesn't matter.

Concerning the blur, It is already done in my tex. I even worked hard to transform some layers from cartesian to polar coordinates, since the Gimp's filter gives sometime some weird and unexpected results.
I've had to experiment dozens of different lines, brushes, hardness, sizes ...
That's why it took me two weeks to make only a very basic mock up.

You can see it clearly (1024px) :

FDAI_tex0.jpg


FDAI_texzoom0.jpg


FDAI_texzoom1.jpg


Maybe not enough after all ...
This is a resize from the original 1800, but instead of the entire image, I resized layer by layer, added few corrections to eachone to get better and more accurate result.
The best is the ennemy of the good. I'll try a basic image resizing for check. The result is generally more blurry.

On the other hand
I negligated NASSP for about 10 years so I still use the good old NASSP 7. If I key F8 two times I can't access to the VC. Strange, I thought I remembered there was a VC included in 7, no ?
Anyway I plan to upgrade to Orbiter beta/NASSP 8 today (after my nap?)
 

Lupus_Vulpes

Member
Joined
Jan 29, 2021
Messages
45
Reaction score
23
Points
23
Location
Germany
Forgot to mention that the link from previous post is the 1024 px texture.
Since I still do not know the required format for the .dds to be accepted I can't upload the 1800 for the moment.
You can paste the 1800 texture in MS Paint into the .dds and just save it, it will work as the 1024 one.
The problem is that i also don't know the required format, i tried everything, to no avail... But if you save it with paint, it works somehow.


Concerning the blur, It is already done in my tex. I even worked hard to transform some layers from cartesian to polar coordinates, since the Gimp's filter gives sometime some weird and unexpected results.
I've had to experiment dozens of different lines, brushes, hardness, sizes ...
That's why it took me two weeks to make only a very basic mock up.

Maybe not enough after all ...
This is a resize from the original 1800, but instead of the entire image, I resized layer by layer, added few corrections to eachone to get better and more accurate result.
The best is the ennemy of the good. I'll try a basic image resizing for check. The result is generally more blurry.
For the 2D cockpit, you need more. Something more like this:

FDAI-interpolation.png


I got this result by scaling the 1024p up to 4096p using NoHalo interpolation, and then scaling it back down to 1024 using "none" for interpolation.
It looks smoother in game, but since the distance between the degree marks is only 2 pixels you loose some detail in the process
The 2D cockpit is what it is. If the texture is too sharp, it looks too pixelised when the ball is moving, and if you want to smooth it out, you loose detail. It's a compromise.
But in the 3D cockpit it looks great, even for 1024p. The 1800p one will look heavenly in the VC. Here is the orginal 1024p .dds that you uploaded in the 3d VC

FDAI-3d.png


I negligated NASSP for about 10 years so I still use the good old NASSP 7. If I key F8 two times I can't access to the VC. Strange, I thought I remembered there was a VC included in 7, no ?
Anyway I plan to upgrade to Orbiter beta/NASSP 8 today (after my nap?)
Oh boy... :D
Since NASSP 7 a lot has changed, you would be amazed. :)
The developers are very diligent, with micro-updates coming out almost daily, sometimes even hours apart! Rarely have i seen some team committed to a project like this one... It's truly amazing, and all the stuff that's in here. I started playing it a year ago and it still amazes me.
 
Last edited:

chrival

Active member
Joined
Sep 7, 2016
Messages
25
Reaction score
37
Points
28
Location
Vaucluse
scaling the 1024p up to 4096p using NoHalo interpolation, and then scaling it back down to 1024 using "none" for interpolation
That sounds good. I'll give a try, at least for the vertical centerline and yaw lines that both seems to be the main responsibles.

Wow ! the VC FDAI looks great, even in 1024. I imagine the 1800.
It would be cool if dev could redo the sphere with more polys ... ;)
Can't wait to see.
 

chrival

Active member
Joined
Sep 7, 2016
Messages
25
Reaction score
37
Points
28
Location
Vaucluse
Here we are

FDAI_Ball1800.jpg


FDAI_1800_1.jpg


Tex is 1800x900.
Load with no problem in VC but blank in the 2d panel, as expected.
So the question is : shall I continue and finish it ?
In such a case, the sim must accept 2 different textures. One 1024 for 2d and one 1800x900 for VC.
Is it possible ?
 
Last edited:

jalexb88

Addon Developer
Addon Developer
Joined
Aug 11, 2008
Messages
142
Reaction score
136
Points
43
Location
Canada
It looks awesome! Yes you can finish it, with 2 versions if that works for you (1024x512 and 1800x900) Also as I said in the git post, we'll need a set of the textures without the red areas (for the LM) Thanks!
 

chrival

Active member
Joined
Sep 7, 2016
Messages
25
Reaction score
37
Points
28
Location
Vaucluse
Yes I think It's worth it.
But IMO mipmap is necessary.

Sorry to be so insistent but to do that I absolutely need the required compression/format of the dds file.

Lupus said it tried all the possibilities and never got it. He gave me the way it did to, at least getting a dds file using windows paint. But that way it's impossible to generate mipmap.
On an other hand, Is it possible to improve the sphere mesh as the current one has too low poly ? Segment are 15°, It seems to me, far to much.
Not a master of 3D, but I think I could do it, along with mapping.

Aniway I'll get back to work right now.
It could be done in a few days.
The LM FDAI is very similar, except the eight ball you mentionned, but also the centerline and a few other details. Once the CM is finshed the LM will follow in the wake.

Of course my priority is a soon release but I will continue to improve some annoying things.
I mean the pitch lines ondulations near the pole (red ball), from around 45° yaw and above. You can see it in the pics above.

Edit : Thanks to all of you guys for the like(y) you gave to me. I don't use social media so I'm not familiar with liking but it warms the heart. ?
 
Last edited:

jalexb88

Addon Developer
Addon Developer
Joined
Aug 11, 2008
Messages
142
Reaction score
136
Points
43
Location
Canada
I think the DDS file for the FDAI is somehow a different type of DDS then the others...not sure right now what the exact compression/format the file is other than being 24 bit.
Once I get back to my main PC I will take a look at it more closely. About the higher definition ball, do you use Blender? Its quite easy to create one in there and you can subdivide to whatever amount you think is best.
 

chrival

Active member
Joined
Sep 7, 2016
Messages
25
Reaction score
37
Points
28
Location
Vaucluse
By the past I've made some basic 3D objects for a flightsim using gmax. A simple sphere should not be that hard.
I start thinking that the pitch lines distorsion partly are due to the mesh itself. Too rough.
To draw the pitch lines I used the Gimp's filter Polar coordinates. As the result was not satisfying, I've suspected something wrong with it, I even checked with a spreadsheet where I computed the coordinates by myself but, at the cost of just a few manual retouching, the filter works well. The fact is it's made for a perfect true sphere. I forgot its not the case. It's just a flat poly assembly.

Meanwhile, things advance :

FDAI_1800_2.jpg
 

jalexb88

Addon Developer
Addon Developer
Joined
Aug 11, 2008
Messages
142
Reaction score
136
Points
43
Location
Canada
Looks good! I think it might be easiest if I just make a new sphere myself in Blender and post an updated VC mesh with it on here. Im on the road right now, but should be able to have it within a few days.
 

chrival

Active member
Joined
Sep 7, 2016
Messages
25
Reaction score
37
Points
28
Location
Vaucluse
Well, CM FDAI is done :)

Here's the 1800x900 :

FDAI_CM_1800_0.jpg

FDAI_CM_1800_1.jpg

FDAI_CM_1800_2.jpg


Of course, it appears as blank in the 2D panel
After resizing to 1024x512 and additional blur (thanks to Lupus for the method) :

FDAI_CM_1024_0.jpg


But the VC FDAI is less accurate than the 1800x900 :

FDAI_CM_1024_1.jpg


The attached file contains the set of both dds, 1800 and 1024.
Simply unzip wherever you want, copy the choosen FDAI_Ball.dds and paste in your '...\Textures\ProjectApollo' folder.

Hopefully, dev team will provide the possibility to have 2 separate tex, one 1024 for the 2D panel, one 1800 for the VC.

Although the result is quite acceptable I'm not fully happy with it.
The pitch lines distortion, in particular near the eightball gives me hives.
IMO there's 2 reasons for that :

First : The mesh itself seems to have a too low level of polys, or lack of smoothing or something like that.
One can easily notice the break at the edge of each poly

Second : The texture itself.
In fact, its a equirectangular projection of the ball, just like a planisphere, where yaw lines are the latitudes and pitch lines are the meridians.
As the latitudes are parallel lines, yaws lines are not a problem
But the pitch lines are, since meridians are not parallel on the sphere.
The lines have to be transformed from cartesian to polar to be drawned properly.
That's the reason why they look like trumpets on the equirectangular image.
During processing a blur is added automatically by the filter. But it's not completely effective.
Looking at it closer, I notice some breaks between 2 adjacent pixels.
Correction is not that easy since the blur itself is "polar".
Btw, I can't make efficient tests as long as the mesh is so imprecise.

Hope you will like it all the same.

Edit : LM FDAI is coming very soon
 

Attachments

  • CM FDAI.zip
    337.6 KB · Views: 56
Top