Page 11 of 21

Posted: Fri Aug 10, 2007 11:28 am
by TGHC
Griff wrote:
i seriously need to scale down the 1024x1024 texture and merge some of the effects maps together, i totally didn't realise we could use the alpha channel as well *studies photoshop help files to find out how to use the alphachanel* i think i might remove the specular exponent maps and just give it a fixed value in the shader, i'm not really using it much and it's taking up space.
i'll change the pow calculation to the other example you posted in the shaders outpost thread ahruman, the quadratic curve, you did warn that the pow one was expensive to use, i forgot to change it when saving out the final fragment shaders from rendermonkey.
Dohhhh, obviously! :wink:

Posted: Fri Aug 10, 2007 3:20 pm
by JensAyton
From a shader perspective, an alpha channel is like any other channel. However, prior to 1.69 (or possibly 1.68), Oolite would premultiply textures by their alpha channels when loading them under OS X (I don’t know about other platforms). Also, prior to 1.69, the alpha channel was kinda-sorta used when rendering without shaders; it is now completely ignored. If you want an OXP to work without shaders in Oolite 1.68 and earlier, don’t use the alpha channel on the texture that’s used for non-shader rendering.

To complicate matters, GLSLEditorSample (which I use to edit and test shaders under OS X) also premultiplies textures when it loads them. This behaviour comes from Cocoa/OpenStep. It probably doesn’t happen in RenderMonkey, but it might.

Posted: Fri Aug 10, 2007 6:22 pm
by Griff
here's a slightly less system demanding version of the ship
http://www.box.net/shared/4rm3dqsclr 1.3MB
I've halved most of the texture map sizes and managed to merge a few of the maps together using the baffling alpha channel. Turrets should be working too with the ones on the back not upside down anymore.
Ahruman the "#ifdef OO_REDUCED_COMPLEXITY" around your shader pulse function, does it always choose the reduced complexity version at the moment? i'm convinced on my system that it does.

Posted: Fri Aug 10, 2007 6:57 pm
by Cmdr Wyvern
Griff wrote:
that's odd, the turrets should be working, they're not in the screen shot i posted a few posts back (the quaternions were giving me problems) but they should be in the version i put up for download... there was a problem with me not giving the turrets a 'thrust' value in their shipdata.plist, ahruman pointed out that this stopped them from being able to track targets but i thought i'd fixed that. there's no turrets on the front and underside of the ship so it can't get you if you hide there, try popping around the back and fire a few lasers into it's engines.
A look into Behemoth.oxp shows turrets usually have a thrust value of 1.
Cranking it to 1.5 might give them faster tracking - if it has an effect at all - but I suspect it'll make their gunnery skills fairly sloppy.

Posted: Sat Aug 11, 2007 12:08 pm
by JensAyton
Griff wrote:
Ahruman the "#ifdef OO_REDUCED_COMPLEXITY" around your shader pulse function, does it always choose the reduced complexity version at the moment? i'm convinced on my system that it does.
Currently OO_REDUCED_COMPLEXITY is never defined, so the reduced-complexity code is never used. It’s a planned future feature, like OO_LIGHT_0_FIX. :-)

Posted: Tue Dec 30, 2008 4:22 pm
by Griff
Stupidly-high-poly-cobra-iii-mangleifification.oxp sneak peak :D
Image
How's the pilot going to make Lavian TreeGrub smoothies when the food blender's mounted outside on the hull?

Griff Cobra3 works with oolite from v1.72 and onwards
http://www.box.net/shared/buguehovmt

Cobra3 with normal mapping (only works with the 'build it yourself' trunk version of oolite)
http://www.box.net/shared/eltugqkkk0

Posted: Tue Dec 30, 2008 4:47 pm
by DaddyHoggy
:shock: :shock: :D :lol: :shock: :D

OMG - it's beautiful!

Posted: Tue Dec 30, 2008 4:52 pm
by Captain Hesperus
DaddyHoggy wrote:
:shock: :shock: :D :lol: :shock: :D

OMG - it's beautiful!
Indeed. Once this is textured, Elite IV will be such a let down....

Captain Hesperus

Posted: Tue Dec 30, 2008 7:37 pm
by another_commander
It's amazing - just as expected. My only observation and kind request would be to make the blender a little bit less tall.

Posted: Tue Dec 30, 2008 8:03 pm
by DaddyHoggy
@Griff - just noticed the "blender" as A_C calls it - I'm afraid I'm going to have to report you to the authorities as it looks a lot like a classified project I used to work on - no seriously it does - can't tell you what though - otherwise I'd have to shoot you... :D

Posted: Tue Dec 30, 2008 8:05 pm
by DaddyHoggy
PS - How high is stupidly high - when it comes to Polys...

Posted: Tue Dec 30, 2008 8:40 pm
by Griff
it's 2383 polys! I Haven't loaded the model into Oolite yet to see how it plays, hopefully it will be ok, otherwise i'll cut out a lot of the bevelled polys around edges of some of the shapes - i've put these in to see how they work with the lighting, i'm hoping they'll help oolite smooth out the lighting across the model really nicely

Good idea about the scale of the food blender, now you've pointed it out it's way too tall, funny how i hadn't noticed that before, i haven't started UV unwrapping that part of the model yet so it can be changed really easily

Posted: Tue Dec 30, 2008 11:40 pm
by Svengali
Great model Griff!!!

Hope that it won't slow down Oolite - I like the current speed so much :-)

Posted: Wed Dec 31, 2008 7:09 am
by Ark
I do not believe that this thread justifies the title of "Ruining the classics" any more. Now it is more close to "Enchancing the classics"
This model is not only a remarkable ship but also a valid replacement for the original cobra

Posted: Wed Dec 31, 2008 8:06 am
by another_commander
Griff wrote:
it's 2383 polys! I Haven't loaded the model into Oolite yet to see how it plays, hopefully it will be ok, otherwise i'll cut out a lot of the bevelled polys around edges of some of the shapes - i've put these in to see how they work with the lighting, i'm hoping they'll help oolite smooth out the lighting across the model really nicely
I think that some clever use of normal / parallax maps here could reduce the poly count requirements considerably, no?