Tweaks to Frob highlight

[rebb]

Well, it seems that indeed converting the diffuse to a .dds file removes the green tint - the reason eludes me tho, it's probably something really odd like this 5 level directory limit, gah !

 

I've read in one of the DDS related threads tho that it's not 100% sure that the game will use .dds files, depending on a quality-setting of Doom3.

Is this still true or did i stumble over a really old thread and there's better knowledge about this matter by now ?

 

Or does the game always use the .dds version if no .tga is available ?

[Springheel]

We're removing the .tga versions after they're converted, so that won't be an issue. Still, I'm very curious about why changing the image format would make a difference here.

[New Horizon]

Or does the game always use the .dds version if no .tga is available ?

 

If the game was set to ultra, it would only search for TGA, but we're not using the ultra setting in the menu and we also just need to make sure that useallformats is set to 1 in the doomconfig

[Tels]

We're removing the .tga versions after they're converted, so that won't be an issue. Still, I'm very curious about why changing the image format would make a difference here.

 

Me too, and here is another riddle. I placed a decal for "wattle", and it shows like this:

 

 

The z-fight is probably incorrect placement by me, but the "transparent" section of the wattle (clearly visible in the TGA) should be transparent, instead if shows white, even in DR.

 

I tried adding DECAL_MACRO to the material definition, but that didn't help. Can somebody please have a look and figure out whats going on?

[Springheel]

That decal worked fine for me last time I used it.

[greebo]

What's the status on this? Who is waiting for what? What can I do to help?

[Tels]

What's the status on this? Who is waiting for what? What can I do to help?

 

I am waiting for reb to solve the "green tint" issue Oh, and still can't get the "wattle decal" to work, it is not half-transparent as it should be.

[rebb]

From what i've seen, the green tint can be fixed by converting certain diffusemaps to .dds .

 

I must admit i've become a bit cautious since the green-tint issue, hoping that there aren't more such issues popping up when using conditional vfp-stages in materials.

 

So i guess the missing stuff right now would be converting green-tint materials to .dds, adding some functionality to the frob.vfp like RGB colors etc, and some fixes to the stage-replace script, as i've found a material where the fragmentMap-assignment was left blank for some reason.

 

What's the general consensus on this, do we want to pull this through or stay with a purely material-based frob-highlight ?

 

If we pull this through i think it might be a good idea to put something like an extra "/materials_newfrob" on the SVN that contains a snapshot of all materials with vfp-frob.

 

That way more people can try them out by renaming the old materials folder and renaming this one to "/materials". Unless there's some better way than that of course .

 

@Tels : Does the wattle-material even do that when removing the vfp-frob stage ?

[Tels]

From what i've seen, the green tint can be fixed by converting certain diffusemaps to .dds .

 

I must admit i've become a bit cautious since the green-tint issue, hoping that there aren't more such issues popping up when using conditional vfp-stages in materials.

 

So i guess the missing stuff right now would be converting green-tint materials to .dds,

 

We should really convert anything to DDS, anyway. Should we do this first?

 

adding some functionality to the frob.vfp like RGB colors etc, and some fixes to the stage-replace script, as i've found a material where the fragmentMap-assignment was left blank for some reason.

 

Could you give me an example? Or we just fix it manually.

 

What's the general consensus on this, do we want to pull this through or stay with a purely material-based frob-highlight ?

 

If we pull this through i think it might be a good idea to put something like an extra "/materials_newfrob" on the SVN that contains a snapshot of all materials with vfp-frob.

 

That way more people can try them out by renaming the old materials folder and renaming this one to "/materials". Unless there's some better way than that of course .

 

Might be easiest, or we create another branch for that?

 

@Tels : Does the wattle-material even do that when removing the vfp-frob stage ?

 

I have no idea and I am not on my PC until like wednesday so I can't test.

[rebb]

@Tels :

The material with the missing fragmentMap assignment is actually the wattle material .

"textures/darkmod/decals/cracks/wattle_corner" - fragmentMap 1 is empty there, at least in my version of the material.

[Tels]

@Tels :

The material with the missing fragmentMap assignment is actually the wattle material .

"textures/darkmod/decals/cracks/wattle_corner" - fragmentMap 1 is empty there, at least in my version of the material.

 

Is that after applying my script? Maybe that material has some problem? sorry, can't run D3 on laptop

[rebb]

That's after applying the first diff-file. But that material has the diffuse-map in a "map diffusemap" stage, maybe the script didn't account for that yet, so it could well be that it's already fixed.

[Tels]

So again what do we do now? Convert more TGAs to DDS?

[rebb]

If thats on the place anyway, i'd say so.

The question is just, which textures are going to "ship" and can be converted already.

[Springheel]

Most of the textures are already converted. The only ones left are the character images and any .tga files that were added after the textures were converted to .dds.

 

The character images cannot be converted yet because most of them are more than 5 folders down, and D3 won't read them. I have to move them up a few layers, and that's going to take some time. To be honest, it's pretty far down my todo list.

 

I'm a bit unsure about the issue, however. Are you saying the new frob won't work with .tgas?

[Tels]

Most of the textures are already converted. The only ones left are the character images and any .tga files that were added after the textures were converted to .dds.

 

The character images cannot be converted yet because most of them are more than 5 folders down, and D3 won't read them. I have to move them up a few layers, and that's going to take some time. To be honest, it's pretty far down my todo list.

 

I'm a bit unsure about the issue, however. Are you saying the new frob won't work with .tgas?

 

That is what I gathered, the TGAs get a green tint for some reason

[Springheel]

That's a bit of an issue, isn't it? Are we going to force mappers to download a .dds converter for their maps?

[rebb]

It only seems to happen on some materials, but that might be due to most of them already being dds, as you say.

 

Apparently Doom3 does some strange stuff in the background when there is a conditional .vfp stage and a .tga is used as the diffusemap ( there's no tint when the vfp stage is not conditional ).

 

Makes me gag .

[Springheel]

It only seems to happen on some materials, but that might be due to most of them already being dds, as you say.

 

Probably best to nail down whether that is the issue. If the new frob can't be used with .tga diffusemaps, that's potentially serious. Mappers making their own maps may not have access to a .dds converter, or even realize that they need to convert images to avoid having that happen.

[greebo]

@rebb: can you run a few tests whether a working material with DDS might stop working when using TGA? Use a "good" material, save the DDS diffusemap to a TGA file and remove the DDS temporarily (to make sure the engine doesn't use the DDS behind your back). Does it still work?

[rebb]

Ok, i just converted a wall-texture from DDS to TGA and it showed the green tint, backing up my initial findings.

 

Again, figured this could be an issue and got hesitant about the whole thing, and my question from #208 still stands .

[Springheel]

What's the general consensus on this, do we want to pull this through or stay with a purely material-based frob-highlight ?

 

I'm a bit hesitant. The new frob is an improvement, but is it enough of an improvement to deal with the hassles of not being able to use .tga? And what other issues might come up with different cards/drivers (thinking back to the ambient light issues)?

 

I think I would recommend NOT trying to do this for 1.0. The current frob system is well tested and works decently. I don't know that we should try altering it this close to release when the new system has potential problems. After 1.0 we can go back and look at it again.

[greebo]

Ok, i just converted a wall-texture from DDS to TGA and it showed the green tint, backing up my initial findings.

Ugh, that's bad - I wonder what the hell might be causing this. Would there be a way around having the stage conditional? I assume having the stage "always-on" is not an option, as the fragment shader needs to be run over every fragment in the scene then - pretty expensive, isn't it?

[Tels]

I am also curious why it would show up as green with TGA and normal with DDS. Are they different somehow in the alpha canal or byte-swapped?

[rebb]

@Springheel :

 

You're right, with these issues it's probably not a 1.0 candidate.

 

@greebo :

 

It could be possible by making the effect a "blend add" and then multiply the output with parm11, but as you mention it would be calculated even if it ends up being invisible.

I'm not sure how much of a performance hit this could have, but it probably has more overhead than a simple non-vfp stage would have.

 

@Tels :

Who knows, maybe id left some sort of debug functionality in the game, which highlights such cases with a green tint, thinking that nobody would ever try to do something like this anyway.