a little hint,

due to a slight change in the materials for alpha31

opacity 99

won't lead to semi-transparent material

if you like such you will have to set it to

opacity 98

---

some more experiences,

shinyness should never be below 1 (=0), else it leads to unwanted extreme shaded surfaces (either lit or fully shaded, depending on view angle) 

if you like something totally flat shaded 1 is okay and use aditionally a black area in the specularity bitmap for such parts, then they turn out well.

take also a little care when you use patterns, materials below a shinyness of 10 look "washed out" in conjunction with pattern colors,

except you would like this specifically.

while i'm not sure if 200 makes sense, because i can't see no difference to 100

patterns have a "secret" (i have no idea if that will stay in this way)

imagine them as if one of the used patterns overlays the other,

if you use only one i recommend to use the first available in the greyscale, turns out to be the red one actually in the game.

this will be the first "layer". "green" tops "red" and "blue" tops or covers "green".

also i found out that the mip-mapping doesn't do the pattern material good, but there is a little trick to workaround that.

that's of course first, details are really bad for it (and i do mean bad, it's a bit sad, one could say best is to use it all only for whole objects/parts, thus you get around this issue).

and if your machine allows it, use a medium to high mip-mapping level (8x, 16x) pioneer anyway looks rather bad with lower settings.

especially above a certain angle textures smear else to much.

but also try to choose a grey for the pattern which is closest possible to the change for the next color, a bit complicated to explain for me.

that means i.e. "white 128" (of 256) seperates the "red" from the "green" "layer", it's on the edge you could say.

if you choose a rather low white (i.e. "white 48") the pattern vanishes with distance and the borders turn to the underlaying color, in the case for the first pattern to the color the material has.

if it's the second layer the first layer gets visible which really looks bad on distance.

but if you set it rather high to the edge i.e. "white 120" that won't happen neither it will tend to fade to the next (very close) layer.

this method can be used for all three layers.

but also be aware that there is always a tiny border to see from the overlaying to the underlaying layers,

like i said best is to use the first ("red"), any else didn't turn out well, except you cover the whole object/part with it.

or it might be you like a sort of rainbow, then it's ok to use "blue", the border will be then red,green,blue. 😉

---

i hope the "rubber coated" look will disappear once, it's not very good to use it at all actually.

the random material, or what should turn once to this, decolorizes the texture completely

like it would be a only low indexed greyscale or b&w

i guess this and that it's somehow "overlayed" makes it look like a "rubber coating",

i really don't like how it looks, i used it so far...

Can the new model system handle tailsitters and/or variable geometry, yet? I've got the urge to play around in Blender a bit...

Can the new model system handle tailsitters and/or variable geometry, yet? I've got the urge to play around in Blender a bit...

That's got nothing to do with the model system, really. The game would need to understand the alternative orientation... I think it would be a fun thing to have and I'd work on it if I had time.

BTW, Blender 2.66 seems to have broken collada animation export - waiting for blender devs to confirm and/or fix https://github.com/pioneerspacesim/pioneer/issues/2123

2.65 and other old versions are available at http://download.blender.org/release/Blender2.65/

Thanks for the tip. I guess I'll keep it at 2.65 for now.

What is the absolute highest tri count we want for ships to be included in master? The wiki says aim for under 10,000; is that a hard limit or is there some flexibility? My comp can run models more detailed than that no problem but I know the goal is to be able to run Pioneer with low-spec systems easily.

I added the note about 10,000 triangles, if you're talking about here http://pioneerwiki.com/wiki/Optimising_geometry

I've seen that number quoted a few places (the bug tracker I think, perhaps the mailing lists too), possibly by kko / Luomu, but have subsequently realised they may have meant 10,000 polys, which is quite a difference. Hopefully someone can clarify.

It seems it's much more important to get the LODs right, e.g. this page on the old model system.

There's no established polygon budget, but don't go crazy. It's more important to have detail levels set up correctly so you don't use a 10K mesh for something that's only visible as a dot.

http://pioneerwiki.com/wiki/Blender_Model_Export