New Derivative Model

[Deus Siddis]

I have started work on a new model (currently at 4,000 triangles) for the Derivative interceptor and have attached some WIP screens of it in blender.

I appreciate comments from any folks reading this.

I especially need some input from jackS on this as the (low poly) modeling is mostly finished at this point and I need to know if this can be approved for official VS inclusion before continuing on into its map creation stage.

PS, you have to click on the attached images to see them in a non-nightmarishly lossy fashion.

[jackS]

Low poly model matches intent of 2D concept art, so you should be good to go.
(Might want to make the 2 rocket pods a little bit more prominent, but as long as the texture makes it obvious they're rocket pods, it'll be fine.)

[Deus Siddis]

Low poly model matches intent of 2D concept art, so you should be good to go.

Cool.

(Might want to make the 2 rocket pods a little bit more prominent, but as long as the texture makes it obvious they're rocket pods, it'll be fine.)

There's actually 3 rocket pods, each one integrated into the 'skeletal' structure that holds each engine pod to the hull, as indicated by the concept artwork and, to some extent, the in-game description. If you look closely at these "wing" structures, you can see that there are a good number of (fairly exposed) modeled rockets sitting in there like in the concept. I might need to make the ship a little more 'skeletal' in parts so that these missiles are more obvious.

Speaking of weaponry though, there seems to be a large discrepancy between the weapon mounts for the three unadorned vessels in-game and in the concept art.

So could you set the record straight on what the numbers and types of weapon mounts are for the derivative, determinant and convolution?

[bgaskey]

Looks nicer than the old model and considerably better than I can do Nice work.

[jackS]

Low poly model matches intent of 2D concept art, so you should be good to go.

Cool.

(Might want to make the 2 rocket pods a little bit more prominent, but as long as the texture makes it obvious they're rocket pods, it'll be fine.)

There's actually 3 rocket pods, each one integrated into the 'skeletal' structure that holds each engine pod to the hull, as indicated by the concept artwork and, to some extent, the in-game description. If you look closely at these "wing" structures, you can see that there are a good number of (fairly exposed) modeled rockets sitting in there like in the concept. I might need to make the ship a little more 'skeletal' in parts so that these missiles are more obvious.

Sorry, I should have been more clear, and I also miscounted (I forgot the ones on either side of the cannon)-- I meant the 4 smaller rocket pods you can clearly see here:
http://vegastrike.sourceforge.net/artst ... ektiv2.jpg

Speaking of weaponry though, there seems to be a large discrepancy between the weapon mounts for the three unadorned vessels in-game and in the concept art.

So could you set the record straight on what the numbers and types of weapon mounts are for the derivative, determinant and convolution?

The current in-game models are irrelevant. Please ignore them completely, as they were used because they were available, not because they were ever intended as Unadorned craft. The armament should reflect the concept art only.

[Fendorin]

very nice ship

i hope you will make a texturing soon as possible

it' s true we need to remplace the old derivative/determinat/sickle/Hammer !!!

[Dunkelheit]

why to remplace those ships?, i like the actual derivative, and much more than this one, and the old sickle concept is exelent but it do need a better model, but based in the same concept.
well this is my point of view.

[jackS]

re: why replaced -- Various ships getting replaced to have a cohesive design style for the manufacturing/using factions involved in addition to those being updated for purely aesthetic reasons.

[bgaskey]

I propose that if the derivative gets replaced, the current derivative model gets moved to some other ship that needs art still. No sense in wasting perfectly good artwork just because there is something that fits a bit better. A lot of factions still need their indigenous ships.

The only problem i can see is that it would be a bit disorienting to experienced players, but that will pass in time while the artwork survives and continues to contribute to the game experience for generations to come.

[Deus Siddis]

Thank you Fendorin for your compliments, and all of your own many art contributions to this project we all enjoy.

And yeah, all of the textures are finished now, the diffuse, specular, ambient occlusion, glow and (tangent) normal map. Special thanks to chuck_starchaser on the spec and diff, for bringing me back to the realistic-lighting side of the force. "You were right about me. . .tell you're sister. . .you were riiight. . ."

Though be forwarned I have not added a crap load of texture detail in this first version at least, since I don't really know what any of this is actually going to look like rendered inside VS yet. There might be some of the overdraw issues we've all been talking about on this forum recently to. Between the poly count and adhering to this ship's particular concept artwork (and maybe some of my own bad habits ans laziness ) this was difficult to avoid. Maybe I'll redo the model as a more optimization focused version later on if users experience noticeable slowdowns when this ship is in view.

I also made a low lod with a third the complexity of the high lod, so hopefully the lod code is working alright.

So now the question is, where do I go from here?

What sort of gpl compliant thing am I supposed to do with the 60 megabytes of source file I have, send them all in?

Do I have to run the converter gauntlet to get a game-useable version of the model and textures or does someone else do that? Does one triangulate the mesh before conversion?

Who has access and authority to commit it to the SVN?

Which is the best to use, general public license, public domain, creative commons?

[safemode]

you might want to talk to pyramid or chuck to see what of your source files or what not is necessary (well, requested would be better), to be uploaded along with the actual data that will be used in game.

[pyramid]

Great to hear another model is completed.

My first suggestion would be to actually make it workable in VS before sending anything in. So actually yes, you should run the conversion combo straight through until you can see the ship in your VS as you have intended it to be seen. I hope chuck can give you more details on the triangulation, and other modeling/conversion related questions. Not sure how up-to-date is the tutorial on the wiki.

The size is no problem, you can send the all the files (data for inclusion, masters, and sources) to me or jacks (or alternatively post it somewhere for download) and someone will twiggle it into the existing data set. I'll probably do the commits and align with jacks on the ship stats and spawning parameters.

License type is GPL by default, or any other of the mentioned open licenses of your choosing as long as you avoid the NC clause in CC.

Hope that helps

[Deus Siddis]

My first suggestion would be to actually make it workable in VS before sending anything in. So actually yes, you should run the conversion combo straight through until you can see the ship in your VS as you have intended it to be seen. I hope chuck can give you more details on the triangulation, and other modeling/conversion related questions. Not sure how up-to-date is the tutorial on the wiki.

Unfortunately after spending some time going through the available information on forum stickies and wiki howtos, I have been unable to make any real progress towards getting this content into a VS readable format. The converter exe's don't even run for me, they just demand extra libraries or whatnot that was not part of the download. There doesn't seem to be complete documentation on adding LOD, Shield and Collision meshes and then all of the shader maps- Diffuse, Specular, Shiness, Glow, Ambient Occlusion and Normal.

The size is no problem, you can send the all the files (data for inclusion, masters, and sources) to me or jacks (or alternatively post it somewhere for download) and someone will twiggle it into the existing data set. I'll probably do the commits and align with jacks on the ship stats and spawning parameters.

Okay, sources sent. Just let me know what else you need me to do in Blender and GIMP to get you or anyone who knows the current conversion process the files up to the right specifications to be converted into the game useable format(s). There just isn't the documentation and/or tools that I can find for me to get this further while including all of the current generation models and maps that I've made for it to meet our new standards.

Hope that helps

It's a start.

[Deus Siddis]

On the topic of the old derivative model, it is now more or less crap by todays standards. It was a perfectly fine model a decade ago, but well. . .that was a decade ago. I wouldn't re-use it.

As far as its overall design goes, the modern day Russion Su-47 is a more badass example of this carnard-with-forward-swept-wing design than even maybe the Nicander. But I totally agree this overall design is superbly badass.

So if jackS has another concept for a VS ship that uses the Canard and FSW layout? I might be up for modeling that from scratch with all models and shader maps and the works instead of the old derivative model taking its place.

[charlieg]

Deus: post your .blend file so somebody else can do it?

[Deus Siddis]

Deus: post your .blend file so somebody else can do it?

There's alot more to a model than just the model itself. There's a little more than 50 MBs of texture sources that are part of the package. Zipped, it all turns into 20 MBs. That's too much information to post on this forum, it won't take that much. But if anyone wants a copy, I'm fine sending it via email. Already sent one copy to pyramid to add to the VS source content repository.

[loki1950]

The Linux version of mesher runs ok but there is no windows version of the latest with techniques that works that i know of ATM klauss did build one but i don/t know if chuck has gotten that far yet in his shader work.

Enjoy the Choice

[pyramid]

@Deus
I've got the obj file you've mailed, however when converting I get the following error:

Code: Select all

>mesher derivative_high.obj Derivative.xmesh oxc
some OBJ functions not yet supported: - aborting

and

Code: Select all

mesher derivative_high.obj Derivative.bfxm obc
No such file or directory
(Check the .obj and make sure it references a .mtl file)

The archive you've sent does not contain mtl files and by the output I suppose that mesher requires them.

Next, I did import the obj file into Wings3D and exported also as an obj file. This created me a mtl file, and while the conversion to xmesh has failed with the same error, the conversion to bfxm was successful:

Code: Select all

mesher Derivative.obj Derivative.bfxm obc
0_0: 4648 faces, 3878 vertices, 0 lines, 4648 tris, 0 quads
Total faces: 4648
Min [-29.329 -17.982 -28.630] Max [29.329 33.000 22.400]

It would be nice to have pontiac's xmesh export plugin, but it seems that the link top his site is broken.

Next, I've converted to xmesh, to edit the texture statements. They were missing, so I've added them:

Code: Select all

texture="derivative_diffuse.texture"  texture1="derivative_specular.texture"  texture2="derivative_displace.texture"  texture3="derivative_glow.texture"

Please note, this is preliminary until the LaGrandeNoodle has been updated to combine with the newest shader technology.

I'll do LODs and shield meshes later, so now converted back to bxfm plus converted the textures to dds (nice resolution 2048x) using nvcompress DXT1 with mipmaps.
Right now, the mesh shows up on my screen but not the textures, so I'll have to check what's going on. Maybe it's my old version of the VS.exe, maybe it's my graphics card (i'm testing on a low-end system with ATI mobile), maybe it's the shaders.

[Deus Siddis]

@Deus
I've got the obj file you've mailed, however when converting I get the following error:

Code: Select all

>mesher derivative_high.obj Derivative.xmesh oxc
some OBJ functions not yet supported: - aborting

and

Code: Select all

mesher derivative_high.obj Derivative.bfxm obc
No such file or directory
(Check the .obj and make sure it references a .mtl file)

The archive you've sent does not contain mtl files and by the output I suppose that mesher requires them.

Damn, I looked up the specifications for .mtl files and saw very little information in them relevant to VS in its current state. No normal, glow, specular, AO, etc. Mostly because this is all as very old and outdated a format as it is standard and compatible unfortunately. So I didn't think the .mtl would be needed since so much extra material data is needed to link to the majority of the shader maps that is not in a .mtl file.

Next, I did import the obj file into Wings3D and exported also as an obj file. This created me a mtl file, and while the conversion to xmesh has failed with the same error, the conversion to bfxm was successful:

Code: Select all

mesher Derivative.obj Derivative.bfxm obc
0_0: 4648 faces, 3878 vertices, 0 lines, 4648 tris, 0 quads
Total faces: 4648
Min [-29.329 -17.982 -28.630] Max [29.329 33.000 22.400]

It would be nice to have pontiac's xmesh export plugin, but it seems that the link top his site is broken.

Not sure why it would be complaining about unsupported features. The .obj files I sent were fairly trimmed down (too much so even in the case of no .mtl files).

Next, I've converted to xmesh, to edit the texture statements. They were missing, so I've added them:

Code: Select all

texture="derivative_diffuse.texture"  texture1="derivative_specular.texture"  texture2="derivative_displace.texture"  texture3="derivative_glow.texture"

Please note, this is preliminary until the LaGrandeNoodle has been updated to combine with the newest shader technology.

Sorry, I forgot to mention this before, but displace is just a source file, it is not meant to be used in game. It is used on the super high res mesh which in turn is used on the game use res mesh to create the normal map that I sent you. In fact only the derivative displace xcf is truely needed for the source directory, it can create a new derivative displace png to be used by blender in about 5 seconds.

I'll do LODs and shield meshes later, so now converted back to bxfm plus converted the textures to dds (nice resolution 2048x) using nvcompress DXT1 with mipmaps.

I used 2048 for an interceptor because I think it was the wiki that said this was a good idea for future versions and I wanted the derivative to be a totally compliant with the VS project's current standards and preferences and almost future proof example of how new contributions should look (with the exception of mesh optimization, as I could not find peace between technical overdraw concerns of my own and the official concept art.)

But point is I can shrink the textures down to 1024 for optimization as was originally planned, it was built to look good at that resolution. But you're essentially the project's technical art director, so do whatever you think makes sense, doesn't really matter to me either way.

Right now, the mesh shows up on my screen but not the textures, so I'll have to check what's going on. Maybe it's my old version of the VS.exe, maybe it's my graphics card (i'm testing on a low-end system with ATI mobile), maybe it's the shaders.

I really appreciate the work you are putting into this for me. You've already beat your way through alot of bugs in this process that I couldn't even get to.

Hopefully this content pipeline is only temporarily as broken as it is right now, because I see now that it is without a doubt the biggest bottleneck to the VS graphics modernization effort.

[pyramid]

Indeed, the mtl files are not required if you go the route obj>xmesh>bfxm as described in the wiki.

The 2048x texture resolution is a great idea. I might end up going with 1024x for the data set.

The problem I stuck with is that the textures are not shown on the screen. I can see the totally black model but whatever I try changing in th exmesh file, the texture will not show, even when I only declare the diffuse texture skipping all the other textures. This also happens with nvidia 8600GT which points to a problem unrelated to the graphics card. Btw, I was running the latest svn build on linux and the latest exe on win.

I've compared the Hawking xmesh file with the Derivative xmesh file I have produced. While the Hawking model shows up fine, my Derivative model does not and I just can't find any obvious reason for this to happen.

I'll be experimenting a bit more with this and later commit the whole bunch of data to svn so that other experienced contributors can help.

[Deus Siddis]

The 2048x texture resolution is a great idea. I might end up going with 1024x for the data set.

Hopefully more artists will jump on board and turn in higher resolution source textures like this. For a game with an indefinite/infinite lifespan, future proofing is well worth it.

The problem I stuck with is that the textures are not shown on the screen. I can see the totally black model but whatever I try changing in th exmesh file, the texture will not show, even when I only declare the diffuse texture skipping all the other textures. This also happens with nvidia 8600GT which points to a problem unrelated to the graphics card. Btw, I was running the latest svn build on linux and the latest exe on win.

I've compared the Hawking xmesh file with the Derivative xmesh file I have produced. While the Hawking model shows up fine, my Derivative model does not and I just can't find any obvious reason for this to happen.

Getting textures (and animations too) to come through has always been the most difficult part of the model conversion process. In fact I would say it is the most anoying part of producing realtime content, period.

I'll be experimenting a bit more with this and later commit the whole bunch of data to svn so that other experienced contributors can help.

Again, I really appreciate your help. You've already taken this much farther than I could have.

[pyramid]

I've committed the data for the new Derivative as far as I could get.

I would like to appeal to experienced modelers to have a look on the model and post back tips on how the textures can be integrated correctly.

[pyramid]

The newly committed Derivative has now diffuse, specular, and glow textures. I wasn't yet able to integrate the normal and ao textures and LOD but will do it as soon as chuck's Cinemut pipeline is ready to go.

Including the normal map is possible, but gives me some triangulation artifacts on the final result. Due to no experience in modeling whatsoever, I can't tell if the source of the artifacts is the mesh or the texture. Here's a pic

What remains to be done is the shield mesh and review of the ship stats.

@Deus Siddis
The model is still quite low poly. Could you add some greebles to it to improve the perception of scale (doors, handles, antennas, ...). You can also use the texture to improve detail of the model. Right now it looks pretty dull.

[Deus Siddis]

Including the normal map is possible, but gives me some triangulation artifacts on the final result. Due to no experience in modeling whatsoever, I can't tell if the source of the artifacts is the mesh or the texture. Here's a pic

It might also be a bug somewhere in VS's very recently supported handling of tangent space normal maps, because I don't see what could be in the model or map to create that obvious of an issue. It renders fine with its normal map in blender at least.

@Deus Siddis
The model is still quite low poly.

But it is 5,000 triangles which to my understanding is the desired polycount limit for interceptor/fighter craft according to Klauss. I didn't know if planar quads are supported by the engine without a host of rendering bugs or artifacts, but there are a few of those in the model that if not triangulated might bring down the count and leave more room for further geometric 'greebles' or welds (mesh optimization).

Could you add some greebles to it to improve the perception of scale

I am not really sure what the scale is, the concept art's only clue is the cockpit. I don't know if it says someplace how big each ship is supposed to be, but I couldn't find anything on the wiki.

(doors, handles, antennas, ...).

Yeah, I could add access hatches and panels, it would make sense to see those on a smallish craft like this.

You can also use the texture to improve detail of the model. Right now it looks pretty dull.

Unfortunately alot of that sort of detail is in the normal map.

But I did plan on adding more detail beyond that later on though, after I had a better idea for how VS's shaders affected the look of things. Because I had been flying completely blind until just now when I saw that derivative screenshot you just posted.



There is one thing though that might make these issues irrelevant. You have etheral_walker's derivative model now, right? Is it a more accurate and/or more aesthetically pleasing, especially if taken as a member of his match set of unadorned craft?

Because if it is then I don't mind my version getting the boot. I learned somethings and made some mistakes in creating it and there are plenty of other ships that I can create or recreate using what I've learned with this one. I just wanted to the older fugly unardorned ship models to get replaced by something better, they've been eye sores for too long now.

So you can consult jackS about this if you want and let me know what you want to do so that I can allocate my efforts effectively.

[pyramid]

It might also be a bug somewhere in VS's very recently supported handling of tangent space normal maps, because I don't see what could be in the model or map to create that obvious of an issue. It renders fine with its normal map in blender at least.

Might as well be that. Until that's figured out, the integrated ship will remain without normal texture.

But it is 5,000 triangles which to my understanding is the desired polycount limit for interceptor/fighter craft according to Klauss. I didn't know if planar quads are supported by the engine without a host of rendering bugs or artifacts, but there are a few of those in the model that if not triangulated might bring down the count and leave more room for further geometric 'greebles' or welds (mesh optimization).

The triangle count seems reasonable though jackS' priority always has been "visual appeal before poly count".

I am not really sure what the scale is, the concept art's only clue is the cockpit. I don't know if it says someplace how big each ship is supposed to be, but I couldn't find anything on the wiki.

True, this would be something worth bugging jackS about. I was wondering if the intended size of the new design would be similar to the previous model. Currently, the ship is about the size of the mining station radar dish. It seems very big as compared to other craft but the scale can be easily adjusted via the scale parameter in units.csv file. So there is no need to change the mesh itself right now. What might be required however is the change of orientation. If I understand the design correctly the forward direction is where the cannon and missile launchers point. With the current mesh, the craft forward direction is actually opposite. You might consider this for future developments of the mesh.

Yeah, I could add access hatches and panels, it would make sense to see those on a smallish craft like this.
Unfortunately alot of that sort of detail is in the normal map.
But I did plan on adding more detail beyond that later on though, after I had a better idea for how VS's shaders affected the look of things. Because I had been flying completely blind until just now when I saw that derivative screenshot you just posted.

Great This might as well wait until the CineMut family of shaders is done and the workflow defined, unless you'd like to continue the learning process.

There is one thing though that might make these issues irrelevant. You have etheral_walker's derivative model now, right? Is it a more accurate and/or more aesthetically pleasing, especially if taken as a member of his match set of unadorned craft?
Because if it is then I don't mind my version getting the boot. I learned somethings and made some mistakes in creating it and there are plenty of other ships that I can create or recreate using what I've learned with this one. I just wanted to the older fugly unardorned ship models to get replaced by something better, they've been eye sores for too long now.

This decision has already been taken. Yours is the craft that fits into the idea of VS universe. What I said was meant to be an incentive for you to continue the beautification of this craft to make it look visually appealing and realistic.

I have just committed the shield mesh, but due to the ongoing development of the new shaders it will show with a whitish glow when hit. This should be fixed with the CineMut FireAndGlass shader and an appropriate technique that will be assigned to the shield mesh.

Also, haven't figured that out yet myself, but if you are eager to investigate, you might have a look on how to integrate engine exhaust to show when the ship is being propelled.