The Leflair Sprite Sweatshop, Voxels need not apply

[Leflair]

The conversion job so far:




Played around with the upper torsos to see how best fit the individual arms later, as you can see I left some of the shoulders "in" on a few of them due to the minimal differences between that part on the various arms - may have to reconsider that when I flesh out the "unarmed" arm frames.

Due to how the unched over body covers most of the arm "away from the camera" I might not actually need to modell the complete arm for those frames, just what juts out of the elbow.

I wasn't satisfied with how the diagonal front legs turned out in the first draft so I redid them to better work with the games way of animating it. Might still require some changes, although this is at 300% zoom so it might not be noticable if a few pixels are a bit "off" here and there once ingame.

There's still some height differences between the Muton and Veronian, due to how the legs are structured on the Veronian (Humanoid vs "bird/bug legs"), most noticable on the full model rotation and the diagonal running comparisons. May require some modification - or it might turn out that the legs work better overall with one of the other alien models. The diagonal arms may "bob around" less than the Mutons due to the more compact motion set.

[luke83]

looking good mate, cant wait to see them in game.

[Volutar]

I see no hand waving in front animation, though some sort of waving in look from behind and from angle. I guess hand waving is something that haven't done yet.

[Leflair]

You mean arm "motion" while moving? Yeah, most of the arms are frozen (copies) currently (with a few exceptions), will take on properly animating that after I'm done with the rest of the frames.

The largest bulk of frames are actually arms, due to having both "unarmed" and "armed" states.

To quote moriarty:
Take a look at my list here: https://openxcommods.weebly.com/uploads/1/1/8/4/11849982/xcom_0_sprite_usage.txt. the sprites are for the most part in groups of 8, for the 8 different facings possible. the order is always the same, and can be seen best when looking at the legs or torso: it starts facing "away to the right" (which is "north" in-game, as seen in the mini-map - the "camera" sits southeast, looking towards the northwest), and then turns clockwise.

[Leflair]

Here's an example of how the Y-level will make things... trickier



The blue line shows the height difference between the current complete models depending on their movement frame.

Depending on which set of "four" movement frames for the legs, you got to track either the inner or outer leg that stays "solid" while the other leg moves. (see pink line)
This becomes a bit tricky with the "bug legs" of the Veronian and the pseudo-depth perspective on the inner leg, which quickly gets "compressed" as hell but I might get away with having the "solid" leg be one-pixel off from the center-line on occasion while the other one moves in the foreground.

The red line shows the height difference between the different frames of the Muton legs (and torso), a total of 3 different levels, with 2 of the leg frames sharing the same one (the left most ones).

As you can see from my current (remake) of the leg animation, the only place to compensate for that height difference is the "free" moving leg, while keeping the 3 midriff levels in mind to get it all to "glue together" correctly later (for the Muton, the other animation sets differ with the Y-level stuff).

Not only that, but due to the "hunch-forward" look of the Veronian, the midriff will at different times be covered by the arm closest to the "camera", which is why I've added some dark purple in the top layer of the mid-riff on the legs - which should mostly be hidden by the arm (Which should impose itself above all other pieces depth) while in animation and should come across as "shadows" when not. I hope that works as I think it does...

This is how it looks "fixed" (I think)


Adjusted all the running animations accordingly, with some new frames too where the legs didn't work right:


Still need to check so that the different "torso's" line up correctly with their lower bodies - and do the arms.

[Leflair]

I've readjusted all the legs so that they are "one pixel" "down" compared to the Muton, and the Veronian torso the same to match-up. Might adjust more downwards.

And this to get some visual ques to figure out the arms 


The top arms should be the models "left arm" (from its perspective) but I'm not sure what's going on with the bottom row, the dark areas would indicate that they are arms on the "other side" of the body, facing away from the camera and they appear to be unarmed like the top row? But they don't seem to quite line up... 

[moriarty]

the lower ones look like some kind of movement animation to me. are you sure you got the right sprites?

[Leflair]

Those are just a small section of the 260+ frames, but yes - directly exported from the game.

Here's the full thing:


Look up in the top right corner, might be why the bottom row in the one I posted look wonky! 

(Not sure what the hell is going on with the 8 frames of a single pixel in the top row)

[moriarty]

I'd say the one-pixel-frames are there because that's where the "kneeling legs" frames are in the xcom_0/1/2 pck files obviously the muton.pck was made in direct analogy. except for the "female upper body and head" sprites at the end.

it's all quite logical. and yes, I guess you've got a wrong offset there, two frames forward. that explains a lot...

[Leflair]

I'm having one more headache though (scroll to the right end):


I'm figuring Paint.net just aint that exact - so even when it displays "32 pixels width" it might be 32.25 because I keep on redoing each grid but once I add up the totals it's off by ".5" or some other miniscule amount that do add up once I start gluing my grid back together.

Any ideas how to pull it off to the necessary exactness?

Moriarty: Does that mean I could make my Voronian have a full set of kneeling animations? What would be the Y-modifiers for the upper body so it matches with the legs? The AI might not use it unless "fixed" though... but it could be fun (there are two different sets of height values for standing and kneeling on all units)

[moriarty]

maybe you are doing "32 pixels wide" and then your "division line" is on the 33rd pixel? although then you should end up being 7 pixels off. weird.

of course you can make kneeling animations... the "AI" doesn't know how to kneel yet, but that shouldn't stop us, I guess. maybe you could also make kneeling animations for the muton, while you're at it? you are definitely a way better pixel artist than I am.

as for the kneeling Y offset, Volutar wrote

for sitting xcom - 4 pixels lower

so that should be it

[Leflair]

Thought it might be the division lines which is why I tried removing them, but the remaining difference is tiny and only pops up when measuring 3+ "sets" of 8 frames. I'll just have to try my best to superimpose my new arms over the old ones and see how it ends up fitting.

I've already done a camera facing sitting frame for the Veronian that's 4 pixel height difference, imagine that. If OpenXcom could include a smarter AI (At least for higher difficulties) than the original that would be pretty grand 

Let's see if I can list and classify all the different arm sets in order from looking at them now (you said they should match your list/the human animations so here we go):
First: Standing perfectly still facing all the different directions, unarmed, unit's left arm.
Second: The same, except right arm.
Third (the one which were two frames "off"): Looks to me like the unit's Left arm, unarmed, movement animation. As it's a movement animation it should be for a direction - Might be the "northern" one going with the clock-wise fit for the torso order.
Fourth: Which should make this the same except for the right arm.
Fifth: Left arm, NE side-view movement animation (fitting with the clockwise order)
Sixth: Right arm, NE side-view movement animation
Seventh and Eight looks like the Eastern movement animation.
Ninth and Tenth: SE movement animation "front".
Followed by the same for the remaining directions, S: 11, 12, SW: 13, 14, W: 15, 16 and NW: 17 & 18 sets.
Which should make the last arm sets the weapon holding arms. The first set, 19 looks to be holding a one-handed weapon and set 20 is for rifles and other two-handed weapons.

Looks to fit your list perfectly.

Is that correct?

Hmm, when I think about it the only reason the game keeps the sprites chopped up like this is to save room (for what was then very premium space).

There shouldn't be a reason why we couldn't just mash the bodyparts together like in my "work sheets" for OpenXcom and have the game just replace the whole model inbetween movement animations (in effect achieving the same result as the original). Would save a lot of time when making new sprites in the future, dunno about you guys but when making completely new stuff, being able to see the "whole" and compare it to other models at the same time is a lot easier than staring at each individual part in GIMP and keeping track of 260 different layers or have a whole bunch of different files

I mean, that way fine for just modifying stuff that's already in the game. Perhaps an option (a txt file flag in the graphics file pack or something) to have the game switch between reading a set of graphics in the two ways to give more leeway. Just a thought.

[moriarty]

looks right to me. don't forget the very last set of arms: the right-arm-firing-a-weapon


 it even makes sense, from a programming-point-of-view... if all unit pck files use the same order, it's much easier to load them, because you only have to define the things once...

btw, what program do you use for extracting the sprite images from the pck file? whatever it is, the whole thing is a lot more eye-friendly if you adjust the window so that a multiple of 8 fits into one row

[Leflair]

PCKVIEW from dashivas mapview, with Save  to get the whole sheet instead of chopped up per frame. Not sure why it unpacks it "out of order" like that.

Once we can insert wholy new species graphics packs, being able to modify the number of frames and the order they load would open up a whole new world. The game already does this when you think about it, because Snakemen and Floaters don't have the same number of animation frames as the Muton/X-Com soldiers. Could even include stuff like transforming enemies when they do some action - which could potentially be linked to another sets of stats or abilities.

Maybe even remove the 8-frame limit if you want to make "smoother" movement animations, and as stated in my previous post edit: Not need to chop the original up in the different limbs.

[moriarty]

Maybe even remove the 8-frame limit if you want to make "smoother" movement animations, and as stated in my previous post edit: Not need to chop the original up in the different limbs.

I do see the reasoning behind having arms separate from torso sprite. If you look at a soldier carrying a rifle from the side, the weapon is behind the arm but in front of the torso. you couldn't do this if it was just one sprite. and there are different arms: power suit arms are thicker, sectoid arms are thinner, so you can't just make a weapon graphic that looks like it's behind the arm.

and for the legs, well, when the unit is walking, the legs are changing, but the rest is not. why load a lot of redundant data?

actually, I'd vote for going in the opposite direction and detaching the head sprite from the torso sprite. not only could you easily implement soldier's heads that actually look like the inventory image, but zombified units could still look like a regular soldier (just a different head), and if you're into gore, you could implement nasty headshot animations as well...