Nope wasn't the X-mirror

I am so confused. Every time I think I've figured out what is causing the problem, it does something I was not expecting. The X-mirror thing was not accurate at all. The best guess I can come up with now is that its something about the way that the bone weights overlap. But its difficult to figure out exactly what the "rule," is, so to speak. At one point I thought that it was "Only adjacent bones can have common weight." But that's not what causes it alone; I did a test model of a segmented rectangular prism where weights from a grandchild bone overlapped with weights from the grandparent. It imported fine. I really hope that soon everything will become clear.

That blasted X-mirror

Today I worked on the re-weight painting as much as I could. At first it was working. I was periodically checking if it was importing as I went, and it was going ok, until one time after I painted close to the middle of the mesh, and then imported to test it. Then the old error showed up. I'm thinking that it must have to do with the X-mirror option in the weight painting. I might try just painting one side, then using that script that I actually thought was the culprit of the problem at one point to mirror it all at once. Ironic that it might be what turns out to be what saves the day.

The Wrong Solution, and Pure Awesomeness

Today I started classes. This semester is mainly General Education, but I do have one Digital Media class. And that was the first class I took today. And. It. Was. Awesome. The entire class was Pure Awesomeness.

In Axis Mundi news, I have about half of the vertex groups in my armature on a "White List." This means that they will not cause a failure in the importer. Then there are 2 vertex groups that I have on a "Black List." This means that if they are imported they will most certainly cause a failure in the importer. I have not classified the remaining bones. The good news - in fact very good news - is that when I clear the weight painting on a Black Listed vertex group, and then re-weight it, it imports correctly. This means that I've pretty much figured out how to fix everything and make it import properly.

I thought I had also figured out something even more elusive about this whole importing fiasco; the cause of the issue. Sometimes the cause of a computer issue is significantly more difficult to determine than the solution. In this case, I thought I had found the solution AND the cause of the problem, however during the course of writing this blog post I tried to verify my hypothesis and I got results that I did not expect. However, I think it is still worth including what I was thinking up until just a few minutes ago. The text in italics is what I wrote when I thought I knew what caused the problem. It still contains true statements that are relevant to Blender and describes what did in fact happen, it's just that the conclusion of the sequence of events is not actually what caused the importing error.

I'm pretty sure that the cause of the importer's problems with the vertex groups can be described by the conclusion of this sequence of events: At one point during my weight painting I accidentally painted without having X-mirror enabled. (This wasn't my fault, by the way, in Blender 2.48a the X-mirror option does not stay turned on between sessions. That's one of the very rare times I've thought something in the Blender interface is completely nonsensical.) I then proceeded to weight paint one side of the mesh for a while, without realizing that the painting was not automatically being done to the other side. Once I had completed a not inconsiderable amount of work, I realized the mistake that had been made. Of course I didn't want to have to do the same work over again for the other side of the mesh, so... (and here's the punchline).... I used a script that copied the weight from one side of the mesh to the other. More specifically, the script "Mirror Vertex Locations & Weight." I've never used it before, and while it was helpful, that's probably what caused the error in the importer.

So anyways, all this post really amounts to is me talking about a whole bunch of technical Blender stuff because going to my Digital Media class has gotten me more excited than usual about all things 3D, Blender, and Computer-y.

AAAAAAAAAAAHHHHHHHHHHHH DigitalMediaexcitedblendidhfilmovifajsdfhaklehf.

.

.

.

I probably need to calm down before I develop an aneurysm.

Even Closer

Today I successfully imported many of the vertex groups for the model. That means that some parts were deformed by the animation, and some weren't. I'm going to slowly add groups until I find the one (or ones) that are causing the problem. This post is very good news, because it means that there's nothing wrong with a good number of the weight painting that I did. So I can at least keep about half of my weight painting at this point. I'm hoping I'll just narrow it down further and further until there's only a couple of vertex groups left, and those will be the ones that are causing problems and have to be re-done. And then it's on to texturing!!!!

College

So I moved into college today. Besides that I worked a little bit on some music, because that's what I felt like doing after I got back from doing all the college things that I felt like doing. I have a little longer, so I might work some more on figuring out what is wrong with my vertex groups. We'll see how that goes. I will probably post immediately once I figure out what's wrong.

3/4 of the way there

So today I managed to get the Mesh, and all 64 bones (that's right, I said 64) of the Armature to import correctly into RCT3. But there is a catch. The importer will only accept the armature if I delete all of the vertex groups that correspond to all of my tedious weight painting. So, the model imports, but it is not properly deformed even though the armature is imported with it. The suspicions that I had a few posts ago have now been pretty much completely confirmed; it's the vertex groups. Now I just need to figure out what the heck is wrong with them.

Halfway there

Good news. I've successfully imported the mesh of the model that I am still not revealing. But only the mesh. I had to take off the armature, and replace it with one that was different. Why that should make any difference at all is very strange to me. But it supports my theory that whatever is wrong has to do with the vertex groups.

Importing problems

Ok, here's the deal. Two days ago, I went over the model that I have been weight painting and found a handful of mistakes and little things that I fixed. That was good. The day after that, I continued doing the same thing, as well as doing a little animation to test out the model in RCT3. That went well, too. The problem came, of course, when I tried to import the model into RCT3. It gave me the same ping that it gave the higher poly mesh. So to make sure that size wasn't the issue, I took a different model that imports successfully, subdivided the mesh a few times, and then imported that. This .modxml file was 12 MB in size, and it imported successfully. The model that won't import successfully is 5 MB. Clearly, size is not the issue here.

Now, at first this kind of made me regret going back to the lower poly mesh, since it wasn't the complexity that was the problem. However, it was so incredibly frustrating to weight paint the hi-res version that I think it is worth going back to the low-res version just to make the weight painting easier.

Back to our timeline: Yesterday all I did was try and figure out what the importer doesn't like about the model. The only thing I concluded was that since it can't be the size, it must have something to do with the vertex groups/weight painting. However, the Warrior of Light model uses vertex groups to define weights for the various vertices, just like this model. And there are no problems with it. I figure I just have to keep looking and I will find some strange thing that I did that causes the error. That's the nice thing about Blender. It's always something you've done. It may be frustrating, but you can always correct it. However, what's really frustrating is when a program malfunctions, and you have done nothing wrong. Then it's almost impossible to get it working right.

I'm looking at you CryEngine 2.

Call me Michelangelo

So I continued weight painting today. I am at the point where I realize there is no more to do technically (Every bone in the armature will deform the mesh pretty much correctly) but for some reason It doesn't feel like it should be done. I guess I will just keep posing it until I am satisfied that no more adjustments need to be made. And I probably should import it into RCT3 for goodness sakes. I keep saying I'm going to do that but for one reason or another it never happens.

A Lesson from the Ant

You lazy fool, look at an ant.

Watch it closely; let it teach you a thing or two.

Nobody has to tell it what to do.

All summer it stores up food;

at harvest it stockpiles provisions.

So how long are you going to laze around doing nothing?

How long before you get out of bed?

A nap here, an nap there, a day off here, a day of there,

sit back, take it easy - do you know what comes next?

Just this: You can look forward to a dirt-poor life,

poverty your permanent house guest!

Proverbs 6: 6-11

I just read this last night after I published my last post. Incredible how relevant it is. Of course, I shouldn't be surprised at this particular book having something relevant to my life... it's always relevant. But I thought it was striking how directly relevant it was to what I had just written, and that's why I am putting it here.

Painting, and a commitment

I hit the weight painting really hard today. Only about halfway done, but I'm doing it completely manually instead of creating the vertex groups from the bone heat of the armature. It's a fun challenge at the current resolution, but any more vertices and I'd be very frustrated. I'm glad I never spent this much time trying to weight paint the subdivided mesh.

I generally try not to make any promises when it comes to my filming schedule. However, I believe that most things in life are a balance. And I don't think that making promises about my filming schedule is an exemption to this rule of thumb. So, although I will not be stating anything having to do with specific dates, I think it would be good for me to announce that I will not be taking any breaks in work for the foreseeable future. I will not spend a day getting distracted by Minecraft. I will not spend a day benchmarking Crysis. I will not spend a day owning people in Modern Warfare 2 Multiplayer. I will not spend weeks taking a break from production to do "mental production." My work is engaging, has clear direction, and there is NO JUSTIFIABLE REASON (that's in caps to help me remember, btw) for me to not be working every chance I get. This is not me stressing myself. This has nothing to do with when I finish The Battle for Axis Mundi. But it's the season for work. And my work is pretty darn fun, so there's no reason I shouldn't be working.

This is not good.

SO... for one reason or another, the Minecraft trial that came with my Humble Indie Bundle has not expired. And it probably won't ever expire, seeing as how I'm not letting it connect to the internet to patch itself. So now I have Minecraft.

Yeah. This is obviously not a good thing.

Should've done that earlier

So I finally did what I said that I would do a few posts ago. I tried to import the model into RCT3. Unfortunately, I ran into a snag. It looks like the RCT3 Importer doesn't like how detailed my model is. It doesn't give me any error messages, it just appears to begin loading the .modxml, deliberates for a good while, then hits me with a classic "ping!" with no accompanying message. I am guessing that it is just too big. So I'm thinking I'll just have to go with the base mesh that I modeled, instead of the subdivided one. Like I said, I think RCT3 would be able to handle the subdivided mesh, but I didn't think about the fact that it might not be able to get through the importer. That's OK, though, because I was starting to get a little overwhelmed with the weight painting of the hi-poly model. Hopefully I'll have an easier time with the lower poly version.

Do it right! Take 2!

Today I reviewed the edits that I described in the last post. The bad news is that I didn't like the way it turned out at all. The good news is that today I filmed and edited in replacement footage (Again. If you're counting, that's the second re-shoot/re-edit for this particular [very short] scene), and I actually liked the way it turned out. In fact, I was very excited to get this part polished up, because I never considered it properly finished in the first place. I just kinda pushed it to the side and told myself I would finish it up later. I guess I've done that now.

I continue to amaze myself with the amount of crap that I can write about making this film, even though it continues to be unreleased. But I guess I explained that in the last post, didn't I? I'm not just trying to finish the film, I'm trying to make the film something I will be proud of. Nevertheless, the fact that it is taking so long does weigh heavily on my mind.

Motivate THIS!

So it has been a few days since I should have posted. This is the post that I meant to write a few days ago. As such, it is slightly out of date. Meaning I haven't worked in the past few days since I forgot to write this on the 5th.

I worked on The Battle for Axis Mundi some; it was editing. I always feel like there is room for improvement with this film, especially with the editing. I want to avoid any shot feeling out of place. I want each shot to flow into the next, and for no shot to have unnecessary, gratuitous screen-time that says "Hey, I'm just too lazy to think of something else to put here that makes sense, so I'm gonna keep showing you this shot." When that happens, a shot that could have been awesome turns into suck. And I want to avoid that and other editing/storytelling mistakes. So I filmed some extra footage that wasn't too hard to do, but I hoped would make some stuff make more sense and flow better. Who knows if I succeeded? I sure don't, (to be fair I haven't finished editing it in, rendering it, and reviewing it) but I can only hope that if I tweak it enough, maybe... just maybe... I can create something that I can be proud of.

Using that thought as inspiration, I came up with this:

That's pretty much what I look like whenever I am working, and most definitely when I am editing. So yeah... that's actually one of the most relevant pictures I have ever posted. Cause' whenever I have an update, that is basically what I've been doing.

Rigging

Today I got into the meat and potatoes of rigging, i.e. weight painting. It's a lot more time consuming painting this model than with the Warrior of Light, because it has so many more vertices. But I really like how it is looking so far. I can't wait to get it fully rigged and textured; I think that there's a lot of creative stuff that can be done with the animation of this model.

Tweaked Out

Well, I may have alluded to being close to finished before, but now it is official. I am finished modelling the model I was modelling. Now I am going to import it into RCT3 to make sure it will work. It has a lot more vertices than the Warrior of Light model. That model has 306 vertices. The model that I just finished has 8,333 vertices before subdivision, (that means 8,333 is the number of vertices that I worked with by hand) and after one subdivision (to make it look a little smoother) it has 33, 679 vertices. Using those numbers, the new model has 110 times more detail than the Warrior of Light model. That shouldn't be a problem for the RCT3 engine to handle, because I have done crazy experiments before where I have hundreds of the same Warrior of Light animation going at the same time. It does lag a little bit, but nothing that would crash the game.

Here's an interesting tidbit: at one point I was considering doing two subdivisions on the model instead of just one. That would give the model 134,341 vertices, which is 439 times more detail than the Warrior of Light model. I don't think I've ever successfully had that many WOL models in one park before (although I haven't tried), so I decided to step it back a notch for performance sake. And honestly, when it comes to sub-surfacing, there are rapidly decreasing returns for each successive division, so even if it would run in RCT3, it's a little silly to have so many vertices define such a simple shape.

So yeah. Done with the thing I started modelling ages ago. November 2010... was it? ...Nope, just checked and it was..... 0_o....... October 2010!

That's right, the model that I just finished is the model that I first mentioned in this post.

Man am I ever dragging this film out.