[DATA+TAGS] H2MT - The Essential Collection

[ShadowSpartan]

Yeah, I hear that man, its just that I believe the H2EK could rival and even surpass Custom Edition in many ways and I hate not seeing it come to full fruition. Even if we do not have all of the tags many could be improvised in one way or another. Many shaders can be created by hand so I'm curious if Shadow's extracted bipeds have proper UV coordinates.

They did have the texture coordinate data, but the shader tags they referenced were empty, hence why they were all white.

There is a lot more to my campaign such as more locations and characters, but without those tags it would suffer because we really need access to the original tags such as animations, bipeds, and campaign-only vehicles, so I decided it might be better as a machinima...still I long to finish that campaign someday, so if you ever get the free time know that I do have a plan for those tags, "I don't just want them, I need them, and that is an adult decision."

There are other limitations stopping campaigns from happening though, such as everything campaign related being stripped from Sapien....so unless someone created a new Sapien, everything would have to be placed by hand in Guerilla. Also, if I remember correctly, Korn said at one point that the AI pathfinding data creation had also been stripped from the tools, which makes AI pretty much useless.

[Resinball]

So if the UVs are good, couldn't the shaders (the ones not dependent on post processing) be recreated by hand by using Entity to study the originals? More work, but possible right?


Yes, with the campaign work I have done I've had to manually place much (but not all) of the items such as characters/weapons spawn locations. Its not very hard, I just place a temporary equivalent in Sapien then copy the XYZ and rotation coordinates into h2guerillaNew. There is a setting in Sapien that allows you to change the scenario type, before this would not compile but now we have the unlocked H2 Tool which will compile any scenario type. I almost have a new custom Main Menu working however some built in check is preventing it from loading, though it does load from SP and MP areas. So making a scenario into a campaign scenario is no longer an issue, much thanks to adolif2 who made these additions to H2 Tool. Then there is the bin directory; I recreated the files that go in there by studying and comparing the headers from existing map files.


The path-finding is certainly an issue, thats why I was thinking that extracted original scenario files might be the answer as the path-finding from them might be reusable. If the path finding remains from an extracted scenario that is then compiled it could be recycled. To test this theory I would need one extracted BSP/scenario/lightmap (if those three tags can be extracted that is). At least then we would know if its possible.

[Ki11a_FTW]

You're not going get any pathfinding data that is gonna remain in the scenario file that you can edit. IIRC pathfinding data is generated when a BSP is built in tool, and thats built upon by the AI navigation tools, which we don't have in sapien.

[Kornman00]

IIRC, enough of the ai pathfinding code is stripped that it doesn't even clear any of the pathfinding blocks before the point at which the actual PF creation code would run. If not, that's a simple enough fix.

Pathfinding data is located in both the structure bsp (which is the autogenerated data) and then in the scenario (which should be the designer specified data). Lightmaps have nothing to do with it. However, IIRC, I finished all the bsp/lightmap recreation code in BlamLib.

The code/definitions in Sapien for the campaign-related data, like Shadow said, were stripped in release builds. There's nothing there to work with.

Tool still has all the needed of the physics/collision model import code, and it's theroetically possible to recreate the render_model import process using or RE'ing the code used in structure importing. Or one could even hack up a process which exports a model as a structure bsp and then processes the imported tag and strips/postprocesses the underlying render model data into an actual render model.

Animation import code still exists to some degree. I was working on recreating this code via RE'ing Halo1's import process (the two don't differ that much). It appears that when Hired Gun was working to get the EK localization-able, they inadvertally changed all the import-based code to read input files as unicode, which makes no real sense when .ass and .jms files contain no or accept non-ASCII characters. H2tool contains code for processing 'old versions' of jma files (version number should be on the first line). CV and I tried playing with this, using a Halo1 jma saved-as in unicode, however I don't think tool actually spat out a usable animation graph (can't recall the reason why).

I've thought about doing a kickstarter for some OS stuff (both H1 and H2), one of the project bullets I had in mind was 'finish tag extraction user interface' (for both h1 and h2). Though, there is also anothe major idea (for H1 mp) that would probably be considered 'worth' more development time than things like that.

[Ki11a_FTW]

If the H1 netcode was fixed(leading, syncing, etc) i could honestly say I may make another MP map.

[Kornman00]

I've privately discussed possibly taking the time to increase the maximum players to 24, 32, or maybe even higher, but most of who I've talked to think that there wouldn't be that much support for a Kickstarter for such a project ("except for maybe mexicans"). Retroactively adding hit-scan support was another idea for the would-be KS. And I'm pretty sure I already have all the needed engine research to add-in some more adaptive object (projectiles, players, and vehicles) hit detection with little effort (the engine already tracks most of the needed data).

[Ki11a_FTW]

I'm a user of OS, and I see it's potential. From my eyes, if everyone in the CE community saw such a feature, (people who don't know about halomaps, or modacity), it would completely boost the community. The problem is, how can they notice it? The only thing I could think of is making it some kind of update, or message in the lobby. I mean maybe an agreement with dennis on halomaps could be made to include all of these features with the game download, thats the best I can think of.

E: hell, a while ago I made a few test maps with OS, using halo's default BSPs. One of them was blood gulch. After days of making custom detail maps and high-poly scenery, and using my HD halo tagset (which is essentially halo 3 models on H1 rigs), it made the game look so much better. But it still has that sloppy feel to it. If these features you are saying will one day be developed, I'd gladly chip in my HD H1 maps as a support to get more players to get OS.

[Resinball]

Thanks for the detailed info guys. Its sounds like there are still a few major limitations but most of them can/could be circumvented.

[arbiter901]

How much you think is needed to fund something of that sort?

[Kornman00]

I estimated between one and two months to: reverse engineer all the addresses (haloce and haloceded) that need to be patched up, then actually coding those patches, then testing. I already have pretty much all the structures documented which are limited to k_multiplayer_maximum_players so it's mostly just finding all the references to the player fields in those structures and making them point to another part/address. It would probably not take up to two months, but I figured it's a decent amount of time to account for any unforeseen circumstances. The KS would have had to cover, at the very least, the time I spent off of full time work (I still have bills to pay). So for one month it would need to be almost 2k. But I wasn't just going to work on more players for the KS project because I think more can be done in those ~two months that others would like as well. Eg, syncing device objects, halo3\reach equipment (for campaign and possibly MP), h1/2 tag extraction, etc.