RMS' code handling the emitter better than the game does
I sort of disagree with this philosophy. The job of my code is not to make the world's greatest particle render system -- obviously we have many game engines, and blender, for that kind of thing. Instead the job of my code in this instance is to act as a preview for Warcraft III. And so, if it looks different than Warcraft III, it is simply broken.
To that end, I looked at this model for a bit and whatever is happening is rather extra. The orange particles don't show on Ghostwolf's viewer, but when I hack the model and change some particle settings I can get the orange particles to show up on Ghostwolf's viewer however when they do, they start looking like what you saw on Warcraft 3. I'm confused by it. I slowed down the number of particles and if I just emit 1 every few seconds, Ghostwolf's viewer sometimes emits these short fat particles that look different, like what's going on in the GIF of the game itself that you included. But on Retera Model Studio the preview was erroneously showing a smooth set of particles going from large to small on the orange tail.
For what limited time I spent looking at the problem, I don't understand why. Retera Model Studio was meant to just basically copy Ghostwolf, because his particle previewer was so accurate to the game. Apparently I skipped over something that causes mine to show the orange when his doesn't, but if we standardize the model to show orange in both, then the Ghostwolf viewer renders like the game and my copy of him doesn't. I don't know if that means I lost something in translation, or what that is. (Despite copying him, there was a language boundary where I had to try to write equivalent code in a different code language.)
This seems to point to a bug in both Ghostwolf's viewer (not displaying the orange) and also a bug in Retera Model Studio (not displaying short/fat particles that sometimes make the orange tail look broken up or not contiguous). The "bug" in Ghostwolf's viewer seems to be that having nonzero but constant weird TailUVAnim and TailDecayUVAnim, which is essentially a nonsense thing to put on a particle that only has
Rows 1 and
Columns 1, apparently breaks. This is fairly sensible, however, because the texture doesn't have WrapWidth and WrapHeight set on the
Textures\LavaLump2.blp reference. So I would generally say that a model formatted in this way is largely nonsense, and it's weird that the War3 game and Retera Model Studio somehow magically clamp this value to 0 anyway even when it's hardcoded to 1 and to 3 in the model data, whereas Ghostwolf's Viewer and Warsmash can't render it like that and abide the UV anim to load 100% transparent pixel data from the edge of the non-wrapped texture.
But even when I work past that weird model data that doesn't match probably any example in the War3 game and is most likely a user error/typo causing this emitter to not show in Ghostwolf and in my Warsmash copy of him, it is weird to me that
some other thing that I didn't find in the particle data yet causes obvious visual differences between Ghostwolf and Retera Model Studio previews -- in such a way that Ghostwolf matches the game and shows short/fat particles periodically appearing that don't match the apparent size or location of other particles in this emitter, that are absent in Retera Model Studio's smooth "larger to smaller" preview of the emitter.
If somebody figures this out, I want to know about it. Would benefit my Warsmash project, as this is an indication of a potential bug in Warsmash and also in Retera Model Studio, and I have not identified the bug yet.
