MPQ editing or Bypass map size limit?

[deafir]

Hello not sure where I should post this, sorry if I got the wrong forum.

I am using many custom models and sound files in my map. The map is for a private multiplayer LAN party to test my maps concept. Using a custom WAR3x.MPQ file is not a problem. It also wouldn't be a problem to use a Map size limit bypass.

My question is which choice would be better for avoiding bugs and lag. So far I have been using an MPQ editor and everything seems to be working fine for models. I would assume using a custom MPQ file would be more efficient, since the map doesn't contain the models.

What differences would occur between using a Custom MPQ file vs a Map size limit bypass?

Lastly has anyone ran into any problems with using custom MPQ files or Map size limit bypass?

 

[GhostWolf]

You mean I will need to download a 374MB file to run your map?
If anything, I would stick it in War3xLocal.mpq, assuming that doesn't change with patches.
But that too, is 44MB to download for your map.

Unless you mean injecting the models with some tool, in which case, too many steps for the common joe.

In fact, I'll argue that just using a custom MPQ is too much for most people, regardless of its size (plus very not friendly, what if multiple maps decide to go this way, do you now need to keep swapping MPQs??).

Unless there's something I don't know about using MPQs, I doubt anyone will want to go with this.

I also don't see any reason for a map to be more "efficient" in any way, because the models are in the MPQ instead.
The same data gets loaded, I doubt it matters much if it's from the map or MPQ (but if it does, then it will most likely be faster to load from the map either way).

No clue how the limit bypasser works, so can't comment on that.

 

[Chaosy]

The mpqs are technically not hacks, like the custom dll.
From a moral standpoint I would prefer.

Otherwise, the dll file is probably a better choice. I've customized my wc3 with the mpq method myself, no issues for me.

 

[Dr Super Good]

I would avoid looking into >8 MB maps until your map has a decent popularity. Any sort of work around will likely have usability implications which are likely to be highly negative at starting a fan base. Let us not forget that for many people large data is still a problem so downloading a 300 MB extension might take hours or days.

Apparently the limit will be officially raised next patch. However like all patches for old Blizzard games it is likely to come SoonTM.

 

[deafir]

Thank you for your comments and input. But i feel this got off topic and I should have been more clear about the purpose of this map. Its not to be shared online, its for a private LAN party. If its successful I intend to release a public version but obviously I would remove the custom sound files and replace the models with base WC3 models that's a given. I agree that using either of these methods is difficult and not user friendly for online maps.

I was hoping to discuss the differences in using these different methods from a technical view and workflow ease. I was hoping to hear back about others experiences using these methods. I did state in the original post that using a custom MPQ file or a map bypass was not an issue for this particular map. I will edit my original post to be more clear about this.

When I was using the editor's import manager to import custom models, the editor started to run slow when I reached 7MB. When I used the MPQ method the map, while smaller in size, took slightly longer to load but the editor appeared to run fine. This may have been for multiple other reason or my imagination. I tried different map editors and reinstalling WC3 but the results didn't change.

 

[deafir]

You mean I will need to download a 374MB file to run your map?
If anything, I would stick it in War3xLocal.mpq, assuming that doesn't change with patches.
But that too, is 44MB to download for your map.

Unless you mean injecting the models with some tool, in which case, too many steps for the common joe.

In fact, I'll argue that just using a custom MPQ is too much for most people, regardless of its size (plus very not friendly, what if multiple maps decide to go this way, do you now need to keep swapping MPQs??).

Unless there's something I don't know about using MPQs, I doubt anyone will want to go with this.

I also don't see any reason for a map to be more "efficient" in any way, because the models are in the MPQ instead.
The same data gets loaded, I doubt it matters much if it's from the map or MPQ (but if it does, then it will most likely be faster to load from the map either way).

No clue how the limit bypasser works, so can't comment on that.

Do you have any experience or reason to believe that loading materials from the map directly is faster than the an MPQ file?

 

[DracoL1ch]

Do you have any experience or reason to believe that loading materials from the map directly is faster than the an MPQ file?

nothing a human could worry about

 

[GhostWolf]

Do you have any experience or reason to believe that loading materials from the map directly is faster than the an MPQ file?

It depends on how the game works with MPQs, which I have no knowledge on.

At best, the whole MPQ is already in RAM, and you load the map so it's also in RAM.
As far as I remember, the MPQ hashing has a constant complexity, so the amount of data in each MPQ doesn't matter.
In this case, the speed should be the same.

If the whole MPQ isn't in RAM, then the game needs to fetch the data from it and get it into RAM, which is of course a magnitude slower than getting it from the map.

Now after we're done with that - this doesn't actually matter.
You will never notice the speed difference in this context.
My comment was just to say that you shouldn't think this gives you any speed benefits.

 

[Dr Super Good]

If the whole MPQ isn't in RAM, then the game needs to fetch the data from it and get it into RAM, which is of course a magnitude slower than getting it from the map.

Maps are also MPQs so are also subject to random file access.

The difference in speed finding a resource to load between map and MPQ can be considered trivial compared with processing the resource itself. That said if one was to store resources in a map uncompressed (not recommended) and the map was inside file cache then there would be a non trivial difference with processing the resources as decompression takes some time.

Now after we're done with that - this doesn't actually matter.
You will never notice the speed difference in this context.

And that is the truth.

I ran a test with my Java MPQ library to test performance. Touching every byte from every file within 1 of the main game MPQs takes less than 10 seconds when in the file cache.

To put that in perspective, touching every mipmap from every texture in WC3 takes ~20 seconds. If the main MPQ archives and map archive are in file cache there is a good chance that processing the BLP image files need more time than loading all required file data.

This is probably why JPEG content BLP files were dropped for World of Warcraft.

 

[Barade]

I use MPQDraft for my mod to create a custom .exe which has the MPQ archive in it and loads it. The problem is that the .exe files dont work on Windows 10 and 8 and freeze the game. Wouldn't be the "Allow Local Files" registry entry be another option?