Solis

OK, here's the fixed version, it should be the right direction now: http://www.geocities.com/canisinvicti/solis_walk-fixed.zip

Should that line be added along with the "anim solis_walk.mb -rotate 90", or should I delete that and only have the one that has a 0 rotation value?

OK, finally stopped being lazy and exported the new version of the animation I made. I ran it with ascottk's script minus the -align ALL part, so hopefully it will still work. Works fine in the external model viewer, but I still need to test it in the game. But in any case, hopefully this should work and I won't have to mess with it anymore (although it doesn't loop quite as well as I hoped, but it's passable) http://www.geocities.com/canisinvicti/solis_walk.zip

OK, I gave it another try, but using the def file you had there it gave a "could not find joint "ALL" to align model to" error. I deleted the "-align ALL" from the def file and it seems to have exported it then, but I'm not sure if that screwed something up or not. In any case, here's what it exported, I guess it's worth another try to see if it works this time (on a side note, how do I insert new animations into the game to test?): http://www.geocities.com/canisinvicti/solis_walk.zip Now I'm really confused as to why the model viewer could load the animation correctly with the first animation if the bones don't actually match though...

OK, I've had some success in getting it working. I've managed to generate a new .md5anim file of a walk cycle I created which is compatible with the original zombie model, you can take a look at a very rough version of it here (ignore any texturing errors, I just loaded the basic mesh in a model viewer for reference): http://www.geocities.com/canisinvicti/zombiewalkv1.avi Unfortunately, it's going to be difficult working on animations as it is. As I mentioned before I'm unable to import the mesh itself, so I'm working with the raw bones themselves without any visual reference as to how the model will actually look in those poses. The hardest part however is that Motionbuilder doesn't associate the bones to the character control panel, so I can't use any of the effector pinning and I have to select the bones directly in the view window. I'm pretty much animation each bone freehand, which is taking a while and it's why the test animation I did looks pretty stiff and jittery (especially where the feet touch the ground). I might have to take some time to set the model up in Motionbuilder so that I can use the extra options, although since I only need to make two animations I'm not sure if that would take more time than it's really worth. Oh, and also, I had to import the animation on top of the original .mb file ascottk made or else Doom 3 gives an error when I tried to do a modelexport of the bones alone. So if anyone else works with this remember that you do need the full mesh on the model in order for Doom 3 to make even a md5anim file for it. In either case I'll work on it over the next few days and see what I can come up with. It moves at about half the speed of the original version so it should even be good enough to use as a placeholder for now. If anyone wants to check it out and make sure it works in the actual game here's the md5anim for it, I'm not entirely sure if I did the origin bone correctly so it might be worth checking now to be sure it moves the way it should: http://www.geocities.com/canisinvicti/boneywalk.zip

OK, that binary seems to work, thanks. When I tried to convert it before, it got through the Doom 3 exporter, but when I tried to use the md5anim on the original model the model would become badly deformed and only random jittering motions would be applied to it. I think it was the origin bone that was screwing things up, either that or I was renaming the bones wrong. What did you have to do to get it working, ascottk? Just remove the prefix and remake the origin as a polycube like the guy on Recall to Hell forum said? I noticed that the animation itself seems to have a lot of problems though, a lot of joints bend incorrectly and jump around, so I'll just make the animation from scratch using the bones instead of trying to modify the existing animations. I did notice that the model itself doesn't work when it is converted into motionbuilder (it had the same deformation issues in the animation that I had when I tried to convert it), but the bones seem to work correctly so I can just ignore the mesh and animate the bones directly, hopefully it won't slow it down too much. I'll see what I can come up with and post it here in the next few days. As long as we have an md5anim file that's compatible, we can apply it to the original model without changing anything else about it, right?

Sorry for the period of silence, I was planning on checking back when I had made some progress but unfortunately I've hit a brick wall. I really haven't found a way modify animations for existing models without having the source materials. Although, I think there's a way to adjust the speed of animations without needing to change the animation itself, which might work as a temporary solution if we need to. But I'm really at a loss here for changing it otherwise, the only useful tutorials I've seen focus more on building models from the ground up. Personally I'd almost rather we just rebuild the zombies using the D3 source material and then make all new animations for them. It would be easier to work on animations if every standard bipedal creature used the same skeletal structure.

I guess I'm a little confused as to how all this works. Are we sharing the same skeleton between all Dark Mod characters, including the zombies? If so, then wouldn't I need to be editing the characters that already have the TDM skeleton applied to them rather than these Doom 3 format models which use a different skeletal structure? And if we use the same skeleton on all the characters and the md5anim files are separate from the mesh itself, shouldn't I simply be able to animate the characters with any generic model that we're using the skeleton on and apply that to the zombie later on? So for example I could take the Thief model and animate it, and then once the MD5anim is saved simply move it to the Zombie directory and the animation would work on the zombie? I just need to know what exactly I would need to animate for the animation to be applied to the Dark Mod zombie that we're using. If we have a .mb file of the basic zombie stance (unless we need to make that), then it should be simple to go from there. Will the zombie need any other of the standard animations (idle, alert, etc.) or do we just need a walk and run for them? If we're customizing them and changing the skeletons on them, the only animations that would work for them are the ones we've done for the other Dark Mod characters, so I'm not sure if we'd need custom ones for these or not.

OK, no problem, I just figured I'd offer just in case. I do have an updated version of my low-poly-full-fingered hand I can upload if the person that wants to take up the task wishes to use them (I haven't heard if the other person that was making that got back to us or not).

Unfortunately modeling isn't my forte, however I can probably do #2 and #5. I don't know how to set it up for UV mapping or rigging however, so once I modify the basic models someone else would need to finish those steps. Are we going to have some kind of system so we can see what models have had what done to them? Perhaps we should do this in steps so that two people aren't modifying the same base model and we end up with two different versions that have each had a single change made to them.

Oh, sorry about that, didn't notice the thread had a new reply. Right now it's not no, I haven't compiled it into a Doom 3 model yet. I can upload the motionbuilder file if you want, if someone wants to look at it and maybe try to figure out why it does that slow slide throughout the animation it would help.

Hmm, rather than modeling a head with "metal" on it, would it be possible to simply make a normal head, make the helmet an attachment, and then maybe allow the helmet to be removed in the game for whatever reason (for example if they get knocked over from a strong force, it would look cool to see their helmet get hit away). It could be an interesting game idea, although obviously just there as an option in case we come up with a situation later where it would be appropriate for a guard to lose his helmet. Or, maybe guards on break could sit down somewhere and put their helmet to the side, making them easier targets (I can't imagine they'd want to sit around for 12 hours with a slab of metal on their head constantly).

Well it certainly won't win any beauty contests close up, but it's light on polygons. It was actually designed to be subdivided into a higher polygon hand, so it's at the bare minimum like that. It was eventually turned into this: But at 752 triangles that's probably way too high to use on our current models as it is since it's most likely pretty far above our polygon budget (did anyone come up with a set limit for that?). I'm not sure how high our current hand models are (presumably pretty high if we can shave 350 polygons from each hand if we use mitten hands), but whether we want to use oDDities mitten hands or BT's reformed hands or an optimised version of one of these, we should probably see what they each will look like and how polygon efficient they are and then decide on which works best. On that note, does anyone have a standard, model only character (before the normal maps, textures, etc.), preferably in Lightwave .lwo format if possible, that I can take a look at and see how the hands look attached to them? I'd like to take a look at what the basic model geometry is like so that I can get a clearer picture of what the hands we're going to use will look like on them (generally if something looks out of place on the flat shaded model, it will still look out of place even with a nice normal map on it).

Leg and arm armor, possibly. It could also be used to identify notable characters, for example a person with a sash to signify someone of importance, or something. Or even bracelets and necklaces, accessory items like that.

Hmm, actually nevermind, it looks like when I first downloaded the file from the email, the download was cut off and the file was incomplete. I tried again using Firefox instead of IE, and it worked. Thanks, I'll take a look at it a bit later.

That should work, what do our current models weigh in at polygon-wise? A basic model for the shadow should be possible with around 600-1000 polygons, although we should be careful that it lines up well with the high polygon model so that the shadows aren't too ridged or look off from how the model does (I always find it weird in games where the characters have blocks for their hands in the shadow). Although, this might be something we can put off until later (even after the first release) since it shouldn't be much of a problem to add it later and it doesn't serve any purpose except to increase performance (granted, it's always good to be as optimised as possible, but it isn't something that's absolutely crucial to getting the mod out). In either case, it would require someone to model or remodel a lower poly version of each character, which would probably be the hardest part.

Well, if we want to save time but make sure we don't screw over ourselves later, we could rig it for now and then do the actual animation later on once we get all the more important tasks out of the way. Since the heads are on their own "channel" in the Doom 3 engine, we should be able to add facial animation later and not have to touch or re-export the actual animations for the characters as long as we use the same rig, correct? If we're going to do that, it would be best if we create a rig for the head that covers every potential uses we'd have for the facial animation: eyes, cheeks, eyebrows, several sections along the mouth, jaw, etc. The face is one of the more difficult things to rig, but once it's out of the way it should make doing basic facial animations fairly easy. At least the meshes were made for that, props to whoever modeled the heads and took that into consideration.

From what I saw, the noblewoman uses the same rig as the standard characters, just with the leg bones assigned to the robes. It would be nice if the legs could actually be on a separate set of bones from the robes so that if the robes lift at times and you could see their feet as they move (for example as they're running away from you, their legs would probably kick up the back of the robes and you could see their legs), but that would probably take a while to do and most likely isn't worth the tradeoff. So the easiest way would just be to stick the legs to the bottom of the robe and attach them to the leg bones that are already in the rig, and then they should move along with the robes and hopefully will move right when they're in ragdoll mode. Although I'm not sure how the robes look when they turn into ragdolls, do they actually collapse inward like cloth material would? If the robes are assigned to the same bones as the legs, I'm not sure how that would affect the ragdolls. Well, considering how often you've said that, I think it's obvious at this point that it will have to be done for one reason or another I think we'll eventually have to reexport the animations that are already done anyway since most of the solutions to our current problems seem to involve that, so should we maybe concentrate on how to perform all the tasks together and then only need to reexport the animations once? It would probably be best if we set up the rigs and models how we want them, import as many animations to the new rigs as we can and reexport them, and then work on finished up the rest of the animations. I wouldn't want to hold back from making neccissary changes just to salvage older animations, since it would probably hurt more in the long run and is something that would be harder to do the longer we wait.

700 polygons seems like a lot to go from full fingered to mitten hands...it SHOULDN'T add more than a hundred polygons or so to insert 2 slots for fingers in each hand. Actually, if we want to go the super-low-poly route, here's a really basic full fingered hand I made a long time ago for another project: It weighs in at 188 triangles, and isn't very optimised since it was converted straight from quads to tris. A mitten hand version would be around 140 polygons. Would something like this work on our current characters?

Thanks, I can get on the FTP now. Wohoo, now I'm official. That file doesn't seem complete though, both on the FTP and in the Email you sent me. I can get some of the files by running the archive recovery, but the rest of them it gives an error when I try to extract them. How big is the rar supposed to be? Oh, and it might not be giving you permission to upload if the FTP is set to not allow overwriting of files. You can just name it something different and upload it again, and then someone with full access can delete the original.

While I do agree that facial expressions wouldn't be seen very often, I think they should at least take effect when guards are in alert and attacking modes. Seeing a guard taking swipes at you and running at you with his sword drawn baring a totally blank expression would look pretty odd, and would hurt the whole illusion of the AI being an actual character. Also having a facial expression for when a character spots the player would be nice as well, if a guard or civilian steps right up to the player and then sees them, they should at least have a look of shock or surprise on their face. I always found it looked kinda cheesy in the Thief games when a character would make this elaborate jump backwards and raise their arms, yet their facial expression stays as a blank stare. Their face should reflect their actions in extreme situations, at the very least. If this wouldn't be too difficult to impliment (a day's worth of work or less), it would be nice to at least get basic facial expressions for a few of the major actions.

Personally I prefer not using motion capturing for our models for a few reasons. For one, it's harder to really tweak and modify, so if it isn't exactly how we want it as it is we'll probably be spending quite a bit of time trying to do that anyway. Also, sometimes motion captured animations don't "flow" into premade ones as easily, at least in Thief the animations were more or less consistantly stiff, having a combination of totally lifelike movements and ones that are less so can look a bit odd. This is especially a problem when some parts of the model don't have motioncapturing, for example in the walking animation the hands are totally stiff, so those would end up needing to be animated anyway. And I'm not sure how easy it would be to apply motion capturing to the rigs we're using anyway without messing something up. I don't think it would be too big of a problem animating parts that we would be able to use motion capturing for. There really isn't that much that would be motion captured that would fit what we need, other than basic walk/run cycles.

Thanks, and I've replied to that questions thread. And no, I do not have FTP access.

Well, I'm not entirely sure of most of the technical/programming aspects when dealing with Doom 3 engine animations, but I'll try to chime in about what I can. 1. I'd prefer full fingered hands, although I think it should be dependant on the model. It would be nice being able to choose between depending on the character, for example in Half-life 2 the main characters of importance (Alyx, etc) are full fingered, while most of the enemies just use mitten ones. Are we going to be using the exact same rig for all the characters though? If so we'll have to decide on one or the other, personally I prefer full fingered hands becase it's beneficial for some characters (like the archer as mentioned in that thread, or for specific actions and cutscenes, if we decide to have some). 2. Not entirely sure, since I don't know how objects have attachments applied to them in Doom 3. Although if it's like I think it is, we can simply keep the sword on the model we animate with and the rig animations could be imported to the Doom 3 model that has the sword attached in-game. In that case we could remove them right away. 3. Hmm, well from what I understand we'd need to edit the models at the neck and set the points as attachment areas, but how a head is reconnected to that I don't know. 4. With any luck this should only be a small tweak, since there are already bones for the legs in the model and all we'd have to do is attach the feet to them. Although I'm not sure if extra bones would need to be added to the robes themselves to get them to flow realistically. I was kind of hoping we could simply have robes be a physics material in the game and it would animate them in realtime, although that option does have it's problems (especially given how Doom 3's physics engine works). 5. I'd like to use facial animations, but it might be difficult to add, especially since we'd need a working mouth (unless we just want to do simple things like frowning and such, then it shouldn't be hard). 6. Err...sorry, no idea at all on this one.

Yeah, right now I'm just using a skeleton def file which has the bare minimum that's needed for the model to be compiled in Doom 3, so the standard def file would help. If it's not too much trouble it would be nice if I could get the files needed to fully compile the Elite Guard so that it's easier to see how it will look in the game. Right now all I have for reference are the Motionbuilder animations and a very weird ghost-like silouette in the game. Also, which draw animation were we talking about? I downloaded the "Elite 7 side on.fbx" file, and it seems that some of the sword-through-sheath issues are due to the animation rather than the model. I could touch it up if we're going to be using that one, if there's not much else for me to do right now (was a decision ever made as far as rerigging the models?)