Pandora Tincs - Multiplayer Shooter For Pandora (video)

[giovanni]

What I'm saying is not an information that I got from the official manual so I can be completely wrong... When you are in Texture mode you don't see the texture applied to the model but a temporary immage you give to the model to understand if the UV coordinates are right or not...
In my (short) experience happened that an object had his texture but not a temporary immage in the UV editor: I looked completely white in texture mode but didn't have problems in the rendering.

But don't worry: if that's the problem it's very easy to provide a .blend file prepared with the UV editor in the main view with that temporary image yet applied... I don't know if you have to reload it (because of the change of directory during the transport)...

I'll see what I can do....
I suggest to ask for help in the blender forum, they're very available to help open projects ...

At least we can add the texture filename manually in the header of the script... It's not too hard

P.S: Why is it so necessary to export also the textures if we can add a simple line in the .bin file?

 

[Butterman]

Even if you add a texture in to the UV Image Editor, it won't export the filename. Your other models worked fine, this one didn't.

It's something to do with the way the TEXTURE, (not the MATERIAL) is linked to the object. I think if you do it via materials, the texture part of the object won't get initialized. Then, trying to export it, it freaks out, calling the texture filename a "NoName" class.

It's important to export textures in Blender, because we support multiple textures per mesh and per face. We need to be able to set for each face which texture it's applied to, thanks to the horrible help in the Blender Artists forum (no help at all), I managed to find a filename, that faces stored, if they all store this, I can create a global list and check against it to make a material index. For some reason that's not getting initialized in this latest .blend file. It was fine in the others.

Also, darkblu, you're on the ball with the problems you're mentioning, I try to implement global specularity and I get this. (8 vertices, 6 indices per cube btw, normals are perfect in manual inspection)

Oh, I lie, I lie. Specularity looks amazing on Giovannis gun, it must just be those cubes. I'll upload a video.

 

[giovanni]

Ok...
I'm going to reload my model and see what I can do...
I'm very sorrry for this disorder... I really don't know where I made the mistake...

However I remember that the Blender Artists forum it's something different from the Blender forum (I can be wrong)...

P.S: Ok for the importance of exporting textures with the model but I still don't understand (sorry ) why we can add it manually (before the exportation) in the script itself... So it doesn't have to search it somewhere in Blender/ It can't fail with the filename...

 

[Butterman]

We need to include the filename, because, that's the only per-face handle on any texture information I can find. I can't find any global material or texture index list or anything. The only way we can know if two faces share the same texture, is if the filename is equal.

It's not your fault . It's the shitty crappy fucking Blender API documentation, it's so poor, it's disgraceful.

Anyway. Some good/sexy news.

This video HAS TO BE WATCHED.

WATCH IT NOW

CODE
http://www.youtube.com/watch?v=1L_ICWbPQ8g


Here's a still.

 

[giovanni]

THANK YOU for the video (also if Youtube doesn't want to work)

I worked on my cubes...
To say the truth I really didn't know what I was doing... However I add:

-Specular map for each cube
-Specular map in bmp (why?... bho)
-I renamed every object or texture to seem more "Blender-ish" (tex.001, cube, cube.001).

I'm not able to do more... sorry

http://digilander.libero.it/giov_lol_92/Cubes2.zip

P.S: I tried to fix the problem with the normals (I saw you solved already it... however) and I modified a little (important changes for me) the texture of my gun... New file at:

http://digilander.libero.it/giov_lol_92/PunchNormalFix.zip

Punch1 and Punch2 are the same thing, one with the normals of the upper part in a way, the other one with the normals flipped.

 

[Butterman]

That worked. What did you do differently?!?!

 

[giovanni]

:unsure: :unsure: :unsure: ..... I didn't do anything different from what I wrote... I renamed... Different specular map format (bmp)... I don't know.

try to change the name of the texture (in the program, not the filename) I suspect that that is the problem (only suggestion)...

P.S: Damn... Youtube doesn't work...

 

[Butterman]

Hmm. It works here. I'll put the video up on the drop.io, it will be .mp4 format. I think you should be able to run that.

CODE

http://drop.io/PANDORA_TINCS/asset/tincs-model-viewer-2009-03-08-10-15-17-34-mp4



The download button is there somewhere.

 

[giovanni]

Good! I like it!

Thank you very much!

I will try to make a specular map for te gun...

 

[darkblu]

Also, darkblu, you're on the ball with the problems you're mentioning, I try to implement global specularity and I get this. (8 vertices, 6 indices per cube btw, normals are perfect in manual inspection)

Oh, I lie, I lie. Specularity looks amazing on Giovannis gun, it must just be those cubes. I'll upload a video.

well, there is your problem - you cannot have a properly-shaded cube mesh of 8 vertices. again, check the smoothing groups in the original model - there should be at least 3 smoothing groups for a cube, and if those are there then check your exporter (if they arent then it is the artist who messed up).

the reason i am nagging you about that is that if it is the exporter that messes the smoothing groups you will have plenty of shading issues in the future.

 

[Butterman]

well, there is your problem - you cannot have a properly-shaded cube mesh of 8 vertices. again, check the smoothing groups in the original model - there should be at least 3 smoothing groups for a cube, and if those are there then check your exporter (if they arent then it is the artist who messed up).

the reason i am nagging you about that is that if it is the exporter that messes the smoothing groups you will have plenty of shading issues in the future.

I'm not even exporting smoothing groups?

Anyway, I moved all of my code over to the fragment shader (who thought per-pixel shading was that easy?), so I'm per-pixel now, it's hawt, and that artifacting isn't present. I'll get a picture. Unless you can spot something off, I think it's fine.

I lie, I can see it still. *sigh*

 

[Butterman]

Exciting times

I've finished my lights for now. I've turned my directional lights into point ones, so that they change, constantly, linearly and quadratically, depending on light source distance. Really, really cool effect at the extremes. Will probably use extreme values for flash lights.

So, now, my list is getting smaller, I need to get specular maps working. Then move all of this code into TINCS from the model viewer. Set up light storage in my world system. Move all of the current assets, (yes, the steampunk gun and muzz's random character model), into the new JMDL format. Then write up tutorials for modelers on using it properly.

After that, we should have some sexy looking lit up maps. I would then have to implement ray-triangle/ray-quad collision into the server. Map collision and proper shooting mechanisms hit with one stone.

So, all of that should be done and polished in about two weeks. Then we can start doing some play testing!!!

 

[darkblu]

I'm not even exporting smoothing groups?

you dont need to export them - in the modeler though they define your normals, and need to be considered at export time. basically, smoothing groups define the vertex-sharing/discontinuities in your mesh, and by merit of that - the calculation of normals.

here is one minimalistic tutorial on smoothing groups in blender:
http:\\ruzzy2006.deviantart.com/art/Blender-Smoothing-groups-tut-46039478

QUOTE

Anyway, I moved all of my code over to the fragment shader (who thought per-pixel shading was that easy?), so I'm per-pixel now, it's hawt, and that artifacting isn't present. I'll get a picture. Unless you can spot something off, I think it's fine.

I lie, I can see it still. *sigh*

per-vertex or per-pixel does not matter if your mesh does not have correct normals.

 

[Bramrash]

Butterman, the progress that you are making is immense, you are really fast. Can't wait to play this on my pandora.

 

[Mithrildor]

Wow, that vid , looked so awesome. Awesome butterman, keep up the great work.

 

[GPSchnyder]

I love the Lights in the Video. Really love it.

 

[Butterman]

This is strange.

Is light jumping? I can't see how my calculations are wrong :S.

I just figured it out, angle-wise, that point is ok to receive light. Whereas, the normal on the back of the closest tile is pointing away from the light. Then my light distance calculation is only a multiplier, so, the multiplier is still being used. If any of that makes sense?

 

[darkblu]

This is strange.

Is light jumping? I can't see how my calculations are wrong :S.

I just figured it out, angle-wise, that point is ok to receive light. Whereas, the normal on the back of the closest tile is pointing away from the light. Then my light distance calculation is only a multiplier, so, the multiplier is still being used. If any of that makes sense?

start from fixing your normals (smoothing groups), and then work on from there. trying to figure out why some normals-based shading is broken when your normals are off the mark is pointless.

in general, doing graphics is a matter of producing pictures that meet your expectations. for that you need:
(a) mathematically-sound expectations (what should my picture look like, and why)
(b) input assets meeting your math prerequisite conditions (directions, positions, scale, mesh connectivity, texture depth and resolution, required precision of scalars, etc).

for instance, from your picture above i can tell you right away that from the point of view of your shading algorithm you "tile" is not a tile (as in: a cuboidal entity with six flat faces and sharp edges among them) but rather is a continuous surface without edges, and hence that is what your shading outputs. apparently you did not expect that.

 

[Butterman]

I really don't get what's wrong with the normals to begin with man. You're telling me that I need to make sure smoothing groups are set up properly? I'm not a modeler, I have no idea how to do such a thing. I can't spend more than 30 seconds on a modeling program without freaking out at the thing. So, the problem is that I don't have a model with proper smoothing groups? Well, that's nothing to do with code so I'm gonna have to wait until I receive a model that does.

 

[darkblu]

isnt you mesh format text-based? if so you dont need an artist to make you a properly-connected cube. here is one for you:

CODE

// 24 vertices, given as position & normal per line, like that:
// pos.x pos.y pos.z nor.x nor.y nor.z
50 50 50 1 0 0
50 -50 50 1 0 0
50 -50 -50 1 0 0
50 50 -50 1 0 0
-50 -50 -50 -1 0 0
-50 -50 50 -1 0 0
-50 50 50 -1 0 0
-50 50 -50 -1 0 0
50 50 50 0 1 0
50 50 -50 0 1 0
-50 50 -50 0 1 0
-50 50 50 0 1 0
-50 -50 -50 0 -1 0
50 -50 -50 0 -1 0
50 -50 50 0 -1 0
-50 -50 50 0 -1 0
50 50 50 0 0 1
-50 50 50 0 0 1
-50 -50 50 0 0 1
50 -50 50 0 0 1
-50 -50 -50 0 0 -1
-50 50 -50 0 0 -1
50 50 -50 0 0 -1
50 -50 -50 0 0 -1

// 12 tri indices, one triangle per line
0 1 2
0 2 3
4 5 6
4 6 7
8 9 10
8 10 11
12 13 14
12 14 15
16 17 18
16 18 19
20 21 22
20 22 23



note that there are 24 vertices (6 faces x 4 verts), every subsequent 4 of which share the same normal so there are 6 unique normals used (+x, -x, +y, -y, +z and -z). that configuration gives you the 12 edges of discontinuity along the surface, like a proper cuboid would have.