Spring RTS Engine

I made an AK, would like to know what you think of it. I also have lower level of detail versions. For LOD, which is implemented, at least i think.
I will post the lower level detail versions later, when i figure out how to automatically make (cropped)side-on and diagonal pictures. Have not even found out how to see basic statistics . I am using blender. The way i put the pics here currently is retarded too, please click on the picture to see them full size.

It is just about as far from the OTA AK as the spherebot is from the peewee, though. Looking to CA, i don't think its design matches general looks of the other units of CORE. What is the direction of the core look anyway? Not sure though, the other (non OTA)units to compare with are more heavyweight, and the outlaw might be close enough, depending on how you interpret its polies.(It seems a little low on them.)

There was another AK+1 made, but many people, including me thought it looked too heavy, more like a Warrior.) It might be the cheek, or the shoulders.(Although the OTA AK has squared shoulders too.) Or the backpack. Smoth seemed to take the criticism a little too hard though, the modders just want their game to look consistent.

I am willing to have another go at the AK. At least i can promote LOD a little, it did not seem to be very hard to lower or raise a units' LOD. I don't think it will be that much harder when textures are added either. I am not sure of the memory cost, but it does remind me when i made a multi-level quad-tree (With many nodes stuffed in an array.) To do this to a depth n, space cost is (math, not code), an array with every resolution of power two below and with some power is only a third larger then one with only the top resolution. I can't tell if this will be the practice in our modeling, but i know it will when models get very high-poly. And so what, if it is in our case is two thirds more or twice the current memory usage, higher end computers have enough space(Which is well worth cpu and gpu advantage.), and lower end computers may be able to opt to only get the lower LOD versions.

Needs to be more muscular for a start, maybe have it stand a little more upright, not as much as the other ak+1 but more than it is now.

For CA, the core look is dictated mostly by the models already in game- primary Mr.d's.

Several elements should be taken from mr.d's models. Things like the mechanical, low-tech look of most of them, the red stripe-eyes etc.

Some people say we shouldnt be using mr.d's models because they're based on the OTA ones, but most of them arent close enough that it really matters. After scrubbing the core logos, they should be fine.

Here are some concepts showing designs that might marginally match mr.d's models.
http://s131.photobucket.com/albums/p315 ... /?start=40

Specifically, to your model: Think about how this thing has to move. It has hip joints but no knees, the torso probably doesnt pivot well either.

It all looks like you've used a bit too much smoothing, rather than taken the time to think carefully about your polygon placement. Try to start with a general idea of your model beforehand, dont let wherever the polygons fall dictate the geometry of your object.

It doesn't look quite right yet. Needs to stand up straighter, have knees and be a bit less rounded imo.

Core units are typically very angular looking, they use alot of triangular shapes with sharp corners but also use thick solid construction, CORE also makes use of 1/3/5 odd numbered parts throughout CORE design.

I think the general design you have there, especially the head area, resembles the AK well enough.

The unit looks overly skinny through the arms, legs and gun pods, those are the areas which need to be worked on IMO.

I've always found it best to work almost directly from the original model's dimensions and basic outline when redoing the OTA models, adding in extra details and smoothing out the rough low poly work to bring it a bit closer to modern RTS quality.

3d party units are the playground to explore new things, but the original units I always like to keep as close to original as possible, the TA cinema movie and cutscene renders are particularly useful in this regard.

I always say just do what you want, but expect alot of bitching from everyone when it comes to redoing the original units if you change the style too much.

Thanks for the replies :). Many good points i didn't see before. To be honest i think starting over is better. The model looks more 'biomechanoid' then CORE.(Or maybe too mechwarrior like too.) I am not sure how much further i will take this particular model.

The unit was partially in purpose so skinny. How to make it slightly bulky and still make it look vulnerable and fast is a problem. It also seems to have a window, and i guess the chin is to make it look more like a face.

So i started working on something more like smoth was making, but i don't like how the body looks currently. Maybe i should try take aspects from both.

Mr.D. wrote:I've always found it best to work almost directly from the original model's dimensions and basic outline when redoing the OTA models, adding in extra details and smoothing out the rough low poly work to bring it a bit closer to modern RTS quality.

Thats no fun
More seriously, the gameplay is significantly different then TA, so i think it doesn't hurt if the units are different. Am all for taking units that look good and are consistent with the rest, regardless of whether they are based directly on TA. On the other hand, i am having trouble convincing myself that what i am making now is worthy of being called a AK.

Can't be animated and it looks strange... as if you applied a catmull modifier.

speaking of loses, why hasnt any one started working on that? rrr different variations of the same model with lower and hi detail for the los levels?

Because this is way too much work.

I decided not to continue the old model and make a new one. I think this one looks pretty cool. It is based on that other one i tried to make it look like a much lighter kbot and faster. I think i succeeded.



I am not sure whether lower LOD is too much work, but if people to it acceptably, it should be used imo. If people like this model i will try lower-lod it after texturing, so i can get a feel on how much work it is myself. Meanwhile, i have not textured anything yet anyway, so i may just try texturing it regardless what you say.(Texturing might be hard though.)

PS still looking for that function to automatically make some views. Not spend that much time looking for it though. PS2, it is not whining when i use non-convex quads :).

Looks too mechy, regret would probably offer cookies for this.

It's better. Still looks a bit hunched over to me, but it suits the style more. Would have to see it running and decide whether it looks too much like a scurrying Quasimodo to be the unit of swarming mechanised doom.

Neat body design.

The guns are a little weak- I think it's mainly just the welding flaws are making it hard to see the mesh edges. But they feel... non-descript, and more a like a projectile weapon than a futuristic laser cannon.

The legs need to be completely redone.

Balance of the design's center of mass is very poor- it would fall on its face in a heartbeat, IRL. The AK was really low-poly, and OTA's designers didn't have to worry about this issue nearly as much, because of the fixed POV and 2.5D engine. Don't use it as the model, in terms of the legs, because their models don't work when viewed in 3D either.

Spend more time looking at your mecha designs from the side- if it looks really imbalanced and it's a two-leg design, it won't appear to be stable to players, either, and no amount of animation prowess can completely fix that issue.

Lastly, if you're going to have feet that should be articulated, they need to be separate meshes, otherwise the foot area will look extremely weird while walking.

JAZCASH wrote:Looks too mechy, regret would probably offer cookies for this.

Hmm don't know what to do about this one. Tried giving it 'eyes' but it was ugly, think any facial-like other then what exists will make it ugly.

crayfish wrote:Still looks a bit hunched over to me

Found a good way to fix that, pulled the head all the way to the back, making the back vertical.(Maybe less mechy too.)

Argh wrote:The legs need to be completely redone.

Couldn't agree more, to be honest, the legs are a bit botched on.. Found them hard to do.(No ideas to nick from smoth there. )
As for the canon, need to think of some ideas. CA had some guidelines that weapons should be recognizable, so i want something recognisable giving the red pulse laser and fitting in this model.

Argh wrote:Balance of the design's center of mass is very poor

Look at the sideward pic, it would fall backward more likely then forward, the tip of the nose wouldnt weigh much considering volume and the feet are actually in the center of the back and the chin. Hmm, while actually i wanted it to slightly lean forward to insinuate speed/nibleness; i guess that should be when moving, though.(I am going to make it worse )

Argh wrote:Lastly, if you're going to have feet that should be articulated, they need to be separate meshes, otherwise the foot area will look extremely weird while walking.

Only the feet are not separate meshes, they probably should be. (Asking for the connected points tells me so.) Have not figured out how to let blender make it separate objects as well though.

In motion... it'd fall forward, the instant the mass was in motion and the center of balance moved. Gotta look at the whole picture of how the leg is going to change the center of balance as the fulcrum moves and the legs extend.

It's really not that hard to design a balanced-looking walker. Simply resist the urge to stick too much forward or backwards of the hips, without corresponding mass elsewhere, and keep in mind that the farther the mass from the fulcrum, the greater the leverage.

I know that all of that engineering crap may seem irrelevant- we're artists, right, so just make it look cool... but it's not. People have been around working mechanical designs all their lives, and can see it and feel it, when things aren't balanced very well.

The 45-degree angle between thighs and lower legs is just fine, so long as the body's apparent center of mass is well forward of the reversed knees. What people tend to do that hoses things, is to put it well past that center, and they neglect a counter-balance.

Here's an example of a chicken-leg design, originally conceived by Guessmyname, with final editing by myself.

A lot of the apparent mass is forward of the rear knees, but it's counter-balanced with the rear. The final result looks and feels very stable and believable when animated. Note how all of the top mass is balanced, almost precisely, on the hip joint- I took great pains with that, getting it to feel balanced on both sides. Basically, if it's balanced on the hips, and doesn't throw too much mass forwards or backwards, it'll probably work.

The problem with a lot of people's attempts to do these sorts of walkers is that they just aren't thinking through the leg motions, and the resulting positions the mass will be in during that.

So, they tend to build stuff that looks great standing stock upright, but terrible moving, or stuff that looks nice from above, but looks really bad if you can see the sides at all... or stuff that's so massively top-heavy that it's obvious it would just tip over IRL. If you're going to go top-heavy, like I did with that walker... pay a lot more attention to balance.

I think that classic "flamingo" style walkers need to keep as little mass above the hips as they can get away with, though. That way, during the animation, the apparent energy of the masses moving around the fulcrum is concentrated in the legs- and, in a good animation, they're balanced against one another.

Argh wrote:I know that all of that engineering crap may seem irrelevant- we're artists

I am a physicist . I guess you are right, it will get torque along the x-axis, and fall, or at least wobble back and forth with each step. I guess the attachment point of the legs needs to be closer to the center. It may only lean forward as much as air friction can provide counter torque. That text could nearly be part of a walker-making tutorial, by the way .

If you're a physicist, you're way ahead of me, on the math. I just eyeball it a lot, think about material weights, which is 90% b.s., since it's sci-fi, who knows what anything masses... and when it doesn't feel right, I go do something else for a few minutes and come back to it. Usually works all right, but I have a lot of experience doing animation.

I think that the key is to actually animate the thing- at the very least, use your modeling software, do some test rotations, really make sure that it makes sense in the extended positions of motion, not just rest positions. A lot of newbie modelers make stuff that obviously cannot even complete its nominal walk-cycle, but looks awesome... not moving...

I guess i did not think about its walk cycle. As for designing scifi stuff, just stick with things that feel right. Use physics only to the extent to make it feel right, and keep it from utter ridiculousness. It always annoys me when people whine about sci-fi is unrealistic physically, because it always is, and you have to live with it. Actually usually in science series (especially in sci-fi) i have to live with all the extreme unlikelyness of the stories regardless of physics. (For instance, Stargate really has crossed the boundaries; nuked a couple of dozen fridges.)

Right, if you're going to make a reverse jointed walker, I (as a biologist) strongly suggest that you look at some pictures/videos of large flightless birds running.

Something like the Emu.

When a chicken walker runs if you take a still photo it does look like it's going to fall over as it usually leans forward. However when they're standing still there legs are directly underneath them.

I'd definitely move the legs forward, however on animation the AK would more or less crouch then push forward.

http://www.fotosearch.com/CRT753/000414cf/

If you could get the ak to move like that it would be tres awesome...

I don't want a unit that looks like a godly machine with superpower guns that dies in a few seconds. Retract the nose a bit and make the legs less mechy.