wungasaurus

> Total virtual memory: 2147352576 That’s oddly close to 2^31.

But you said you even found the new chunk and the wiki part for it? That means you know where to find it and with the wiki how to interpret it?

?????? There is ZERO guessing in file data ids. There is NO need to know any filename. The "everything is in the same folder" assumption is weird and absurd. Just fucking map file data ids to files, extract them, call them whatever you want (although I suggest just calling them something with the id in the filename) and add it. Just take TXID, look at the ids in it, use cascexplorer or https://bnet.marlam.in/files.php#search=type:blp&page=15680 or whatever to map it to a file, download/extract the file, call it whatever the fuck you want, add that string to the m2, change the offset to the newly added string and you're done. No guessing. No estimating. No WMV and looking at geosets. No "all in the folder". Full support for models you don't even know the fucking name for.

Why don’t you just copy all textures and name them fdid_$id.blp, slap them into an mpq and just don’t care about the filename? There is literally no reason to care about filenames for these textures.

Given that information, TXID should be 0, something, something, 0, where the two somethings are the file ids for slot one and two.

The index in the old M2Texture array is equivalent to the index in the TXID array. Special textures are still defined by the M2Texture array. Just for hardcoded textures, the M2Texture name offset will be irrelevant and TXID points to the relevant file data id.

Have you actually read the error message?

Again, it depends on the game version. In 3.3.5 it is just two textures (xtextures/splash/splash.blp and xtextures/splash/wake.blp) that are drawn with a tiny animation. I don't know what you mean by "physics based".

Depending on game version ripples is just a texture animation. You can probably make a model with just that. Are you taking the circular ones or the ones when moving? Either way, a few seconds animation in blender should look pretty much the same.

This is a feature of the engine, not some database. It is technically possible, but not feasible.

Ocean shall always be at height 0. Stuff will break otherwise.

the easiest way to find out file data ids is probably using https://bnet.marlam.in/files.php#search=aggramar_original&page=1&build=6bffbae0d39ffae8147e43e7ba3a52e5 -- the *source* model https://bnet.marlam.in/files.php#search=creature/stonekeeper&page=1&build=6bffbae0d39ffae8147e43e7ba3a52e5 To read the AFID chunk try looking at it again with my list of entries and map that to the bytes you see in the screenshot while comparing with the wiki page at the same time.

Again, the structure of that chunk's content is give at https://wowdev.wiki/M2#AFID_.28Legion.2B.29. It describes exactly how to interpret those bytes. Again, what you have to do depends on what you replace by what, which I do not know and thought was the other way around in your previous post. The *resulting* file needs to have *all* the entries that the *source* model has. It is likely a good idea to reuse the file data ids from the *replaced* model for all pairs of anim+sub_anim the *source* and *replaced* model share. For any additional animation in the *source* model you will need additional file data ids, as said for example from unused files. The latter would mean that the *resulting* file will use some of the *replaced* model's anim files, and the additional anim files named as if they were something completely different. Note that the name does *not* matter other than for file data id determination purposes. The game *only* uses the id in the AFID chunk to find them.

I literally just looked at your screenshot and applied the description linked on the wiki before. By selecting the value in the editor and setting it to a new one? Just replacing the block will end up with two animations still being broken. That video looks like the .anim file itself is not converted correctly.

as an example, stonekeeper.m2 in your screenshot has the following AFID data: {anim_id = 0x004a, sub_anim_id = 0x0000, file_id = 0x000752c5} {anim_id = 0x0054, sub_anim_id = 0x0000, file_id = 0x000752c7} {anim_id = 0x0061, sub_anim_id = 0x0000, file_id = 0x000752c8} {anim_id = 0x0062, sub_anim_id = 0x0000, file_id = 0x000752c3} {anim_id = 0x0060, sub_anim_id = 0x0000, file_id = 0x000752c9} aggramar_original.m2 has the following entries (i.e. the same anim_id, sub_anim_id pair) as well (note: all sub_anim_ids are 0 so omitted) 0x54 0x4a 0x61 0x60 0x62 it additionally has 0x4b 0x3c I don't know which way around you want to "convert", but well.. if you replace aggramar_original.m2 with stonekeeper.m2, you just have to change stonekeeper.m2's file_id values to those of the respective aggramar_original.m2 AFID entries and just replace the files. The other way around, you will have to find two additional files you can safely override (e.g. the unused old stormwind wmo files world/wmo/azeroth/buildings/stormwind/stormwind_221.wmo etc) and use the filedataid of those.

Everything I said relates to the AFID chunk only. You do not need to look at the rest of the file.

Just replacing the contents will not work if the chunk size is not identical or if when interpreting the contents in pairs of 4 byte values, the first value of each pair in both files does not match. Or when expressed using https://wowdev.wiki/M2#AFID_.28Legion.2B.29: anim_id and sub_anim_id need to match when just replacing. Otherwise, only file_id shall be replaced with ids linked to filenames that are overwritten. It doesn’t matter what file is overwritten. if the count of entries in the new file is bigger than the old, more files have to be overwritten, again it is irrelevant which, as long as the file id you give in the chunk and the id for your replaced file name match up.

I have no idea what you mean. If the client of that version reads those chunks, then it is going to load the files by id, not name. Are the correct .anim/.skin/.whatever files visible with the given file data id?

Well, yes.

Deleting 7 bytes makes NO sense at all. The thing you’re deleting is four bytes chunk name and four bytes chunk size. It always has been. For this “technique“ to work, the second four bytes need to be file size - 8, so you can easily check if it even could work. They need to be renamed. Do all animations not work or only those with anim files? Does your target version already have AFID chunks? (Then renaming them or moving them has no effect unless you replace the right ones.)

Even if it was an easily delectable header, it would be 8 bytes, not 7. Any .skel based model won’t be „easily deletable“ though.

I'm not talking about the models but the lua/xml stuff. At least it looks like that. I did not see anything model or MPQ related.

The instruction at "0x00713E26" referenced memory at "0xB41CD381". The memory could not be "read". I can't pin-point it directly, but it appears to be UI related. The code is highly nearby glue/frame xml stuff. This is not a model issue.

Please make sure you know which file crashes and send exactly that with the exact crash log. For this kind of errors it is important for information to be as precise as possible.

The file you sent does include doodads. Is this really the one crashing noggit?