[KARMA]

i have a strange situation here, i'll try to be as much precise as possible, i'm posting here because it seem system related

my ststem is a pII 400+ voodoo2 (a "little" outdated:p)

i have a model with some strange clipping, like if there are intercepting polys, but there aren't any. I also tried to fully triangulize it (splitting it in two subobjects) and i still have the same clippings

the clipping is somehow related to size: i usually test a 20x mesh in game during stabilization, and at this size the clipping was really minimal, and became weird scaling the model

i discovered that this clipping happens only if i  play in d3d mode, if i play in 3dfxglide mode the clipping disappear
Hades tested the mesh with his d3d card (that i presume little old) and he had the same clippings
I discovered also that 3dfxglide is much, much, much more stable: i built my early models with interceptions because i never saw problems, the edges generated by the interceptions (so "virtual" edges) were stable and defined, but when i switched to d3d recently i saw for the first time clippings due to interceptions

The strange thing is that i see those clippings (in the model w/o interceptions) even in truespace if i use opengl or directx to view the rendered model in real time

After some thoughts i noticed that those clippings are located mostly around convex parts and parallel faces very near to each other (this may explain why the clipping got increased when i scaled down the mesh: close faces got closer)

Clipping is not the only problem, it also is like if in game some uvmaps got little screwed
 here are some pics: in these series the first one is taken in game by Hades, the second by me with modelview






 


 

as you can see it is like if some uvmaps are little wrong, but it isn't: both with modelview and the internal render of lithunwrap they are perfect, they got screwed only in game in d3d mode, if I switch to 3dfxglide again the textures are fine

here is also the model http://digilander.iol.it/AlessandroFerrarese/varie/aw.zip

Can someone tell me what the hell is happening and how/if i fix this?

[DTP]

you may have some flipped normals.

I took one look in model view at your POF model, and it seems you got flipped normals.

Trusespace4 will still render normals even when they are flipped.

in modelview, that does not render these flipped normals when viewed from the oppersite side, you will get Holes in the model

and fs2 uses a similar system of rendering, but with out taking a closer look at your model in a prog like 3dmax, i cant tell for sure.

okay, taking a look converting to 3ds then importing into max.

[Solatar]

I just got a Radeon 9000 Pro 128Mb 3d card. once that's installed, I'll see if there are clipping issues there as well.

[DTP]

I have taken a closer look.

Seems to me in order to achieve a low poly-count, that you cut some corners.

And that’s fine as long as you work with the model in the scale it should be, but i know that is a ***** to do in TS, so naturally you scaled it down, when you finished the model.

But, did you do this by using PCS scaling option,
Or did you do it manually, in Truespace.

Scaling is really only practical, when you work with a mesh that is closed so that all polygons will be relative scale of the original.

Scaling can give some pretty horrific, misalignments.

As for the UV mapping, I’m sorry, I think that is editor/game misalignment thing. so if you got uvmappings, that is not at the position where they should be, then you got to make some educated guessing, and misalign the maps in the UVMAP-editor in the first place, so that the game projects them correctly.

[KARMA]

Originally posted by DTP


But, did you do this by using PCS scaling option,
Or did you do it manually, in Truespace.

Scaling is really only practical, when you work with a mesh that is closed so that all polygons will be relative scale of the original.

Scaling can give some pretty horrific, misalignments.

I scaled it both with PCS and manually, with same result
And i think that if the vertices relative position is the problem, there should be a lot of "holes" (non stable faces)
I also already verified that the clipping is not caused by nontriangulated faces: i tested a full triangulized model too and even in this case the clipping is present and got increased when i scale down the model

As for the UV mapping, I’m sorry, I think that is editor/game misalignment thing. so if you got uvmappings, that is not at the position where they should be, then you got to make some educated guessing, and misalign the maps in the UVMAP-editor in the first place, so that the game projects them correctly.

i will probably do this way, but i've never heard about game uv misalignment when the uvmaps are correctly aligned with any other rendering engine tested (lithunwrap, truespace, 3dexploration, modelview)
And more important, the uvmaps are PERFECTLY aligned in 3dfx glide mode and are misaligned only in direct3d

what i really can't understand is why the model is PERFECT in glide mode and weird both for clipping and uvmaps alignment ONLY in direct3d (and as said i've already seen that in glide mode even intercepting polys are well rendered)

it seem to me that in this case those clipping problems happens around convex elements, like if the game have some problems rendering those elements (maybe it can't count if it has to render one face or the face below, with an effect like when there are interceptions)

what i'd like to know is if it is a known problem or not (i've never heard of clipping probs not related to triangulization when you don't use intercepting polys)

the strange thing is why it has these problems with D3D and not with glide? it is something related to D3D and GLIDE specs or it is the game that simply works better with glide than with d3d?

btw thanks for attention

[Solatar]

Okay karma, I couldn't install that Radeon (no AGP slot) so I installed an nVidia GeForce4 64mb. Same clipping issues.

[DTP]

Originally posted by KARMA



I scaled it both with PCS and manually, with same result
And i think that if the vertices relative position is the problem, there should be a lot of "holes" (non stable faces)
I also already verified that the clipping is not caused by nontriangulated faces: i tested a full triangulized model too and even in this case the clipping is present and got increased when i scale down the model

i will probably do this way, but i've never heard about game uv misalignment when the uvmaps are correctly aligned with any other rendering engine tested (lithunwrap, truespace, 3dexploration, modelview)
And more important, the uvmaps are PERFECTLY aligned in 3dfx glide mode and are misaligned only in direct3d

what i really can't understand is why the model is PERFECT in glide mode and weird both for clipping and uvmaps alignment ONLY in direct3d (and as said i've already seen that in glide mode even intercepting polys are well rendered)

it seem to me that in this case those clipping problems happens around convex elements, like if the game have some problems rendering those elements (maybe it can't count if it has to render one face or the face below, with an effect like when there are interceptions)

what i'd like to know is if it is a known problem or not (i've never heard of clipping probs not related to triangulization when you don't use intercepting polys)

the strange thing is why it has these problems with D3D and not with glide? it is something related to D3D and GLIDE specs or it is the game that simply works better with glide than with d3d?

btw thanks for attention

No problem, regarding Glide, well Glide and OpenGl will always be much more predictable than D3D will ever be. Maybe when an OPEN GL version of FS2_open is released, well then, you can test it on this. Not sure if phreak has finished development, but I think he has pretty much finished.

D3D has multiple issues, and that’s because it has to support a wide variety of GFX cards, while the 3dfx company who also produce the cards developed Glide, they would always be able to test their drivers on the used chip-sets.

While I take it, that not all "scenarios" are being tested for issues with D3D, because there is so many “scenarios” to test.

Think of a Video-Card as a human from anywhere in the world
Think of the registers and buffers and RAM these video cards as the human’s language

Now Think of D3D as a translator, because that is what it is. and sometimes slight miss-translations causes a slight misunderstanding.

It might be a D3D translation error.