Griff's normalmapped ship remakes

[Cody]

edit: hmm, it looks like it's something in my decal applying function....
edit2: specifically this

Code: Select all

decal_TexCoord *= mat2(decal_c, decal_s, -decal_s, decal_c);

I'll hold the driver roll-back for a while, in case you figure something out.

[Griff]

ah, OK I think I have a fix, it doesn't like taking the 'decal_TexCoord' variable, doing the mat2 sums on it then writing the results back directly (the *= in the following line in the shader)

Code: Select all

   decal_TexCoord *= mat2(decal_c, decal_s, -decal_s, decal_c); 

but it seems to be happy if you take an extra step and do the sums on another variable and then write the result back to the original variable in this case called 'DecalRotTemp' :-

Code: Select all

   vec2 DecalRotTemp = decal_TexCoord * mat2(decal_c, decal_s, -decal_s, decal_c);
   decal_TexCoord = DecalRotTemp; 

I'll do a bit more testing then upload some fixes, but if you want to try the fix yourself, in the ship shaders search for

Code: Select all

   decal_TexCoord *= mat2(decal_c, decal_s, -decal_s, decal_c); 

it appears once, in the 'decaliser' bit of the code and only in the main ship shader (not in the engines or gun subentity shader if the oxp has them) and replace with the following two lines:

Code: Select all

   vec2 DecalRotTemp = decal_TexCoord * mat2(decal_c, decal_s, -decal_s, decal_c);
   decal_TexCoord = DecalRotTemp; 

[Cody]

Hmm... that doesn't seem to work, I'm afraid.

[Gimi]

ah, OK I think I have a fix, it doesn't like taking the 'decal_TexCoord' variable, doing the mat2 sums on it then writing the results back directly (the *= in the following line in the shader)

Code: Select all

   decal_TexCoord *= mat2(decal_c, decal_s, -decal_s, decal_c); 

but it seems to be happy if you take an extra step and do the sums on another variable and then write the result back to the original variable in this case called
'DecalRotTemp'

Code: Select all

   vec2 DecalRotTemp = decal_TexCoord * mat2(decal_c, decal_s, -decal_s, decal_c);
   decal_TexCoord = DecalRotTemp; 

I'll do a bit more testing then upload some fixes, but if you want to try the fix yourself, in the ship shaders search for

Code: Select all

   decal_TexCoord *= mat2(decal_c, decal_s, -decal_s, decal_c); 

it appears once, in the 'decaliser' bit of the code and only in the main ship shader (not in the engines or gun subentity shader if the oxp has them) and replace with the following two lines:

Code: Select all

   vec2 DecalRotTemp = decal_TexCoord * mat2(decal_c, decal_s, -decal_s, decal_c);
   decal_TexCoord = DecalRotTemp; 

Ok, tried your fix on my multidecal_cobraIII_player.oxp and it worked fine on that one.
Haven't tried going through the rest...

Edit: Patched the griff_cobra_mk3_player.fragment shader as well and it worked fine here.

@Griff, are you working on a fix for the resource set?
@Cody, what file did you try to patch?

[Cody]

Ok, tried your fix on my multidecal_cobraIII_player.oxp and it worked fine on that one.

Oh, I didn't think to try my late-model multi-decal Cobra III. <opens text editor>

[Gimi]

Ok, tried your fix on my multidecal_cobraIII_player.oxp and it worked fine on that one.

Oh, I didn't think to try my late-model multi-decal Cobra III. <opens text editor>

<Grabs a beer and waits for Cody to post results...>

[Cody]

<Grabs a beer and waits for Cody to post results...>

The man from Del Monte, he say Yes! I had to edit several fragments in the late-model OXP, but it looks fine now, with no errors.

I wonder why it wouldn't work with Griff's newer Cobra though? User error, probably!


I note that those new drivers have a Shader Cache thingy - something I've not seen before in the Nvidia control panel.

[Griff]

I wonder why it wouldn't work with Griff's newer Cobra though? User error, probably!

no I don't think it's your fault, I think I've got quite a few different versions of these oxp's around the place in various states of updated-ness which have even more shader errors in than the one that's just shown up now in the new Nvidia driver!
I've started uploading fixed versions into the two folders over at box Core shipset for Oolite v1.77 (and higher) and Core Shipset for Oolite v1.79 (Trunk) - https://app.box.com/shared/rpkms25cc9
updated OXP's are:-

Code: Select all

Griff_Worm_v1.02.1.zip
Griff_Adder_v1.04.1.zip
Griff_Anaconda_v1.04.1.zip
Griff_Asp_v1.04.1.zip
Griff_Boa_Mk2_v1.04.1.zip
Griff_Boa_v1.04.1.zip
Griff_Cobra_Mk1_v1.04.1.zip
Griff_Cobra_Mk3_v1.04.1.zip
Griff_Ferdelance_v1.04.1.zip
Griff_Gecko_v1.04.1.zip
Griff_Krait_v1.04.1.zip
Griff_Mamba_v1.04.1.zip
Griff_Mamba2013_v1.04.1.zip
Griff_Python_v1.04.1.zip
Griff_Shuttle_v1.02.1.zip
Griff_Sidewinder_v1.04.1.zip
Griff_Transporter_v1.03.1.zip
Griff_Viper_v1.02.1.zip

(all the ones with decals in basically). I haven't updated the bundle OXP over at Cmdr Cheyd's deephorizons site, i'll get that sorted soon and the other standalone oxp's too - bug, gnat, new viper remake etc

[Gimi]

<Grabs a beer and waits for Cody to post results...>

The man from Del Monte, he say Yes! I had to edit several fragments in the late-model OXP, but it looks fine now, with no errors.

Yes, great, we are getting somewhere.

I wonder why it wouldn't work with Griff's newer Cobra though? User error, probably!

Which OXP exactly?

I note that those new drivers have a Shader Cache thingy - something I've not seen before in the Nvidia control panel.

I haven't really dug into the nVidia settings, so never noticed any changes. I do notice a better performance with these drivers though, so I'm reluctant to roll back...

[Cody]

Shader Cache - reduce CPU usage by saving compiled shaders to a disk cache.

I've enabled it anyway.

[DGill]

I seem to have a problem with Griff_Ferdelance_v1.04.1oxp

shipdata has :

"griff_ferdelance2013-NPC" =
{
like_ship = "oolite_template_ferdelance";

but oolite_template_ferdelance does not seem to be defined in the oxp - is there another oxp I need to install?

[Cody]

I think that's normal for Griff's ships (and others?) - it should work as is.

[DGill]

I think that's normal for Griff's ships (and others?) - it should work as is.

Ok, thanks. It must be something I've done as after installing the ferdelance the log file gives:

14:11:07.097 [shipData.merge.failed]: ***** ERROR: one or more shipdata.plist entries have like_ship references that cannot be resolved

I'll delete the file and see if I can spot what is happening.

[cim]

"oolite_template_ferdelance" is defined in the core Oolite files ... but only in 1.79, not in 1.77.1.

[DGill]

"oolite_template_ferdelance" is defined in the core Oolite files ... but only in 1.79, not in 1.77.1.

I'm using development version 1.79.0.5742-140517-f739348

I installed Griff's core shipset files for v1.79(trunk) updated may 28

All ships seem to work ok except ferdalance

13:56:24.147 [files.notFound]: GLSL ERROR: failed to find fragment program griff_ferdelance_new2013-PLAYER.fragment.
13:56:24.154 [files.notFound]: ----- WARNING: Could not find texture file "Hull".
13:56:24.169 [files.notFound]: GLSL ERROR: failed to find fragment program griff_ferdelance_new2013-PLAYER.fragment.
13:56:24.176 [files.notFound]: ----- WARNING: Could not find texture file "Hull".
13:56:24.192 [files.notFound]: GLSL ERROR: failed to find fragment program griff_ferdelance_new2013-PLAYER.fragment.
13:56:24.199 [files.notFound]: ----- WARNING: Could not find texture file "Hull".
13:56:24.214 [files.notFound]: GLSL ERROR: failed to find fragment program griff_ferdelance_new2013-PLAYER.fragment.
13:56:24.221 [files.notFound]: ----- WARNING: Could not find texture file "Hull".
13:56:24.237 [files.notFound]: GLSL ERROR: failed to find fragment program griff_ferdelance_new2013-PLAYER.fragment.
13:56:24.244 [files.notFound]: ----- WARNING: Could not find texture file "Hull".