NES games with the best animation

I'm curious to find some of the best examples of quality animation on the NES console. Two that come to mind for me are Metal Storm and Moon Crystal. Wondering if there's anything on the same level in terms of smoothness of animation and frame rate.

Maybe Prince of Persia or Aladdin? Aladdin was back ported from the SNES. Why restrict yourself to NES? People were sketching sprites on graph paper and entering them as hexadecimal numbers into text only consoles in those days. Of course their animations could be improved on. We have so much better tools to work with today.

little samson

Mike Tyson's Intergalactic Power Punch had some pretty smooth animation. I've seen people watch the intro and say "This is so fake! NES games can't have smooth animation like that!"

Boulder Dash is another good one. The characters are small, but energetic and lively. The animated background tiles also add a lot of life to the game.

But really, I think the final boss from Mr. Gimmick takes the cake. It's a huge, realistically-proportioned, well-animated person with lots of poses. His animations seem more like a series of keyframes than animations, but I really think the artists and programmers did the best with what they had.

Moon Crystal

@darryl.revok I do not know where your avatar come from, but it looks like much much better animated than the average NES game. If it comes from a released game, then you probably have your answer

By the way it is crazy how much the art of static good looking pixel art and of good animation are two completely separate things : You can have all frames looking amazing and the whole animation look like pure crap, or the other way around.

Just looked at some clips on YouTube... I don't see it. Didn't notice anything particularly smooth.


Mentioned in the first post.


I'm pretty sure that's from the game he's making. You even commented on his art, BTW.

I tried really hard, but couldn't think of any other examples. I guess this kind of animation just wasn't very popular on the NES.

I forgot about this one; it wasn't one that I played much. Interesting that the sprites in this game lack outlines. I started out trying that and it didn't really work for the art style I'm using.


I didn't even know this existed. So apparently there were three versions of the Genesis port. This is the version that had Disney animators. The official EU version is... well... the animations themselves are pretty fluid. The way they move around the game screen is unbelievable choppy. Unless there's something wrong with the timing on this video I found, this looks almost unplayable. (At first glance I thought, this looks like GBC, and it turns out it's a port of the Game Boy version)

The Super Games version looks a little better as far as control, (still choppy) but the animation quality is much lower.

The port of the SNES version looks alright except for a few glitches in the video I saw. I'll say it looks better than most NES animation.


The options for defining shapes become so much greater when you have a bigger palette. There's better animation on later systems, but then again I can watch anime and see much more detailed animation than that. When it comes to defining shapes and motion within the limitations of the console, I'd like to see the best examples that have been done so far. Part of my curiosity comes from wanting to study how they did certain things, and another part of it comes from wanting to see where exactly the bar is for this one particular aspect of NES game design.


Not bad. I can see why this would get suggested. The animation in this game really bugged be though because I feel like they figured out how to draw things spinning and decided to use that for just about everything.


I haven't beat this yet so I had to look it up. It's pretty impressive. He's very large for NES. His body could move more for the sake of animation, but given the overall amazing quality of the game, I can see why they didn't even need to, if they even had space for it. The boss is already impressive as it is in this game. If the boss was in Moon Crystal though, for example, he'd need to move more or he'd look stiff compared to the player.


Why thank you. Yes, it's mine. I need to upload a newer version. I made a couple tweaks.

Definitely. This is my first foray into animation, and I ended up redoing my first batch because of this. When I went back and looked at some games that I remember being visually impressive, like, say, Zen Intergalactic Ninja, I see that the animation itself isn't anything to write home about. Some games like Kirby are A) probably some of the best animation on the system, and B) right on the verge of being impressive for sake of animation but not quite. Like, I feel the visuals of that game are super effective, he has a ton of powers which have accompanying animations, and the game itself is so smooth that the entire thing feels smooth. But, if you just look at the animations, you wouldn't think "that's an impressive animation with a lot of frames". I feel the same way about Mr. Gimmick. Being that those two are still near the top in animation for the system, I feel like that only leaves a handful of games that really stand out for this one particular reason. To me, so far, I've seen it in Moon Crystal, Prince of Persia, and Metal Storm. I feel like there must be more Japanese games though that I don't know about.

Does an unfinished homebrew count? I have one in the works, that has animation rivalling Contra/Super Metroid. The best part? I'm not the one programming it!

Dragon's Lair and the transformation in Guardian Legend

Does it count? Hmm... I dunno, can I see some animations?

Are you talking about Cat's Quest? It looks good but I thought you were going NROM with that one, which wouldn't leave room for a lot of animations. My apologies for not reading the entire thread if you changed your mind along the way and already posted it.


Yeah the NES version does have pretty good animation even if the game is abysmal. It's weird that good animation in an older game usually goes hand in hand with dreadfully sluggish controls. There's no reason that you can't have a standard platformer with more frames of animation.


Yeah this is pretty cool. Neat game too. Compile made some fantastic games.

If a game has good animation but sluggish control it's either because:

a) Control is too dependant on animation.
b) Animation is fluid but has bad timing.

In some games, especially on platforms without hardware sprites, I wonder if it has something to do with needing to restrict movement to tile multiples. Or it might have something to do with failure to transition from running frames to jumping frames.

It might be due to the nature of fluid/realistic animation, similar to how Street Fighter III players complained that the characters were telegraphing their moves.

Jumping may be an instructive example. In a Mario game, when you push the button, Mario leaves the ground. Right away, at full speed. In real life, you can't do that; you have to bend your legs first, and for anything more than a short hop you'd be lucky to get off the ground in 10 frames. The same applies to other types of motion.

If you're animating something in a cartoon, you can time the transition period so that what we tend to think of as the motion itself happens on cue. But if you're animating a video game, you can't start the animation early; you have to go by button presses. And that introduces a lag between when the player expects the important part of the motion to happen and when it actually does. This requires intelligent balancing of fluidity and promptitude.

A similar phenomenon occurs in music. Lots of instruments have attack periods in the tens of milliseconds, sometimes longer. (In real life, a musician will simply take this into account and start early, but MIDI or tracker arranges often fail to account for this and end up with lag.) When preparing an instrument for a MIDI keyboard, it can be necessary to chop the attack off the samples if you want the instrument to feel right when played live - for instance, a lot of the attack on a piano, especially one in a poor state of repair, consists of the noise made by striking the key, which happens before the key hits the keybed, meaning a typical MIDI keyboard can't tell you've even hit the note until that part of it is supposed to have already played...

...

Or it could just be that the developers only had so much time and effort to spend, and spent it unwisely...

In my personal opinion, having more than 1 or possibly 2 frames of lag before jumping in an action game is a huge design flaw. This shouldn't ever be done, not even for better animation. In a game that is not action oriented this is no problem, however I don't know a game that is not action oriented and where you jump. If such a game were to exist, then a nice jumping animation is acceptable.

Now what would be clever is to switch between good animations in moves that aren't too time sensitive (such as landing after a jump, or getting up after a crouch) and moves that are time sensitive (such as crouching or jumping).

Do you think that games with laggy controls are that way because animations for new actions have to be loaded? If so, I wonder how it would look if the game allowed the actions to take place right away, keeping the old graphics until the new ones are loaded. I kinda plan to implement something like that... even if the tiles can't be loaded in time, new actions start anyway, and only the animation lags.

Not Cat Quest, that project is *my* baby, nobody else gets to work on it!

It's my stupidly named "Not Metroid" project from last year, that has transformed into a collaboration between myself, and Collectorvision Games. I don't *think* I've posted this .gif here until now?


It was even officially announced! So I'm allowed to talk about it in more detail! ...sort of.
http://www.retrogamer.ca/moderne/neo-re ... e-et-wiiu/

EDIT: Oh yeah, the article miss-spells my name as "Justin", rather than "Dustin". XD
...Honestly, I'm kind of used to that, from French Teachers in elementary school.

^^^not all of it, but you at least posted the running animation once in your topic.


Not most times. I've seen one or two speedruns that allege this but no substantiation. However, a bigger factor for smooth animation is that, obviously, you don't want to be jerky--which means that you either need a transition between any two states (so, different from running->throwing than from standing->throwing) but one way to reduce the magnitude of transitions you need is to have common intermediate states- so the former turns into running-stopping-standing-throwing. Obviously, this takes more time for the character to display, and the computer can't know you want to throw until you hit the button (to start with stopping), so there is longer between button and throw. This can be avoided by accepting jerkiness or using more animation frames (or, with models, some kind of tweening...but that has its own problems.)

But the more important one is from that caveat that the computer cannot start the animation before you send it the directive...so if your guy has a wind-up before his throw, then the wind-up must be completed before the ball leaves his hand and after you press the button.

Consider the Prince of Persia standing jump. He crouches, then leaps up. To remove or shorten the crouch would reduce the realism of the jumping animation, but would be necessary to make the jump happen closer to the button press. Super Mario in SMB3, on the other hand, has "running" frames and a single in-air frame (well, two, but only one for going up) and absolutely nothing in between. So, you press the button, and Mario is immediately in the air in his in-air pose, having never animated a jump motion of his limbs.

Similarly, consider Megaman's Buster. He has one shooting frame [with variations for walking and jumping and climbing ladders, so that no further transitions are needed!]. You press the button, he immediately is in that frame. There are no in-between animation frames of his hand turning into the Buster first, or him extending his arm from whatever position it was in before. It just goes from one to the other as you push it.
Contrast with Link from A Link to the Past's sword (or rod, or cane) swing. The sword, moving smoothly through its arc, illustrates it better- but the early part of the swing where it is still to the side, it can't be hitting in front of him. (This isn't much of a problem in this game--perhaps a better choice would be Soma swinging the Reaper's Scythe in Aria of Sorrow.)

And if you had the effect happen while the animation was still winding up, that would be bad, because the batter shouldn't be winding up when the ball goes through the strikezone, and yet get a hit.

In any case--my point is, in a lot of cases, there is a necessary trade-off between immediacy of action(generally considered 'responsive controls') and smooth animation for that action. [Turn-around animations, or swing follow-throughs, are good examples of this on the back end of an action "reducing responsiveness" and "cancels" are players' way of circumventing it when it doesn't just let you start a new action by default. If you've ever seen a Megaman X series (as Zero) speedrun, you can see how ridiculous this gets.]

Probably not. All but a handful of MMC3-era games use CHR ROM anyway, so loading new animations takes 10 microseconds.

As for CHR RAM games, I'm doing Battletoads-on-steroids in my current project, and going from run to the first frame of jump animation takes one 8-tile load transaction that completes in one frame. Sprite tiles are double buffered, and each frame has a most likely following frame that the engine automatically loads into the sprite slot's other buffer. For example, the next frame after stand is walk 1, after walk 1 is walk 2, after walk 6 is walk 1, after jump is crouch-land, and after crouch-land is stand.


In fact, that's the failure mode I've implemented for prediction misses, such as the first frame of a jump or punch combo, and the player usually doesn't notice. It does cause larger bosses (which take up two 16-tile sprite slots) to occasionally break up for one or two frames when one slot handles its miss before the other does.


One compromise is to make A hop without delay and Down+A perform a long jump with all the crouch frames. Super Mario 64 does something close to this.

I am not sure what exactly you have in mind, however, it sounds like a bad idea to me. If the player press "jump" and that the player is logically jumping within the game but still appears to be standing for a few frames, this will be incredibly weird. The player will not know that it's sprite is logically jumping, so the controls will feel just as sluggish as if they were really sluggish. In addition to that, it could lead to that "weird hitbox" symbol where your character on screen touches something but isn't hit or the other way arround, very typical of bad games. So in my opinion it's a very bad idea.


Very nicely animated indeed. Especially the hair.

I assumed the intent was to still update the position of the sprite, even if the animation frame is wrong...

I honestly don't think that a 1, 2 or even 3 frame delay in animations that run at 15fps or so will make such a big difference for the player.

I think instant response and long wind-ups are both acceptable ways to do gameplay, but in both cases the game has to be designed around it.

I think Prince of Persia is actually a great example, all the motions are very slow and deliberate, but also very precise to control. This is largely due to how actions are confined to a grid, leaving large windows of time to press the keys you need for the actions you want.

The delay from a wind-up can be a good source of conflict and challenge, too. When you can't start and stop on a dime, you have to take care when committing to an action.


The reason it is so often bad is that a lot of games are poorly designed, and in a lot of cases the animation is just straight up in conflict with the game's design. If the animation department can change the timings of things in the game for a purely animation-aesthetic goal without collaborating with design, the result will be poor gameplay, and this does happen a lot more than it should. Instant actions, on the other hand, mean that there's no animation involved, so at least it cuts out that potential conflict.

I feel like this is the approach they took for Metal Storm, and similar to what I'm doing. If there's not need for an animation to convey an instant action, for example, walking, there's no reason not to make it fluid as possible. If you jump or shoot though, it's immediate. You can still make nice, fluid animations without delay, you just have to make some minor sacrifices on animation so as not to let gameplay suffer.

I changed some of the more "realistic" gameplay ideas I had for my game because I could very quickly tell they weren't going to feel good to play. For example, I wanted you to have to hold the attack button to draw your sword, and have it sheathed before you could walk at full speed. I just decided to instead make her attack animation draw and attack in one move. There's a little bit of a delay in the attack, but for an attack, I think this is good. I expect the combat to be a bit more like a fighting game, so I'm going to have leading frames and recovery frames in an attack animation. You draw, swipe, and then sheath. If you hit attack at the right timing, you'll attack again quicker before sheathing. That's the current control scheme with which I plan to experiment. Some characters will draw and strike faster and they'll probably have a longer recover, or vice versa.

I feel like, for some of the bigger games, like Kirby or Gimmick, they may not have done more animations because they may not have had the CHR space. I haven't disassembled these or anything, but wasn't Kirby like the biggest NES cartridge and filled to the brim? For most games, I feel like the decision to not have more animation was due to the time it takes to make the animations and the perceived value to the player. For something like Kirby, would more animations have made the game feel better? Maybe just slightly. For something like Moon Crystal, Metal Storm, Prince of Persia, I feel like the designers made a conscious decision to prioritize animation and make that a selling point of the game. Would I have been more impressed the first day I brought SMB3 home if it had more animations? Hard to imagine my mind being blown any more than it already was. But I remember getting Metal Storm and thinking, "Wow! The main character animations are very fluid! This looks awesome! Especially when he explodes!"


That's really cool! So you're doing all the graphics? How far along are you? You're a pretty talented pixel artist. Everything that you've done looks great. Cat's quest looks like LTTP on NES!


Well said.


Which one, X4? That game has great animation and I remember the control being as responsive as the SNES titles. What's the ridiculous part?

Thanks!
Yeah, I'm drawing every tile in the game, on my own, so far.

The progress? The Player sprites, several NPCs, a few Enemy sprites, and tiles for a "planetary select" screen. The game's file menu layout, takes inspiration from Metroid Prime, with the file handling, being animated, something more like Zelda 64.

--It's funny that you mentioned NROM, because right before this game was picked up, I was in the middle of converting the game to those restrictions. Here's the player sprite:

79 tiles! That's 7 less than Mario from SMB 1!
The entire run cycle is there, only the diagonals were removed.

Yeah, Metal Storm is a good illustration of where you can mix it up.

Not the animation being slow (though a full slash does take a little time)--but slash dash cancels.

Okay I just thought of a problem that is about to arise in my game. I drew an animation of Alisha ducking.



Because of DMA limitations, Alisha doesn't always start an animation as soon as she performs an action (like what Tokumaru described). I'm using center-based coordinates, and the hit box is a rectangle around the center point. When ducking, I would need to move the center point down. If it starts the animation late, there will be a glitch where Alisha's sprite pops down for a couple frames, before the ducking animation starts.

The feet sliding together looks a bit unnatural.

How do you handle trying to DMA more information to vram than you can? I would have thought you'd just go past vblank and have part of the screen cut off for a second. I've been thinking of an idea (that probably isn't going to happen...) to double buffer everything where half of the tiles for what would be the next frame are loaded into vram. If the animation changes to something that isn't expected, (like jumping in the middle of a walk cycle) the double buffered stuff would be erased and the first frame of the jumping animation would have to be completed in one frame.

What's really killing me right now is to have it know the location of the sprite tiles for each object in vram... Also, it would have to know the location of the double buffered tiles in vram...

Wait, you know what I just realized? I was thinking of having it to where only half the tiles of every sprite in an object would be transferred to vram for double buffering, but then it would take up twice as much space as it needs to... I guess I might have it draw all the tiles for the even sprites to double buffer, and then all the tiles for the odd sprites to finish it. Either that, or I'll just hardcode it for everything...


I also kind of feel like the hair should sway or something at the end, even if this would take up more frames. It's not like rom's the issue though.

Yes, this is something I realized too, but didn't consider this to be a problem since I usually make the bottom center the reference point for the sprites (but I can change it on a per sprite basis).

@ Espozo

Every object has two copies of the frame number. One "input" frame number and one "output" frame number. The character's AI routine determines the input frame, then the animation engine decides weather or not it can update the output frame, and then the OAM building routine draws the output frame.

So it's like one has the previous frame, and one has what is supposed to be the current frame?

Pretty much.

I think I have an easy way around the collision box problem. For the player-enemy collision, I cheated a bit and used point-box collision, to make collision more forgiving. I could just have a collision point variable, so I can just move the collision point.

R-Type 1-3 actually use a single point for the ship's hitbox. Despite how hard these games can be, they're actually pretty forgiving in a couple of aspects like this. I don't know why I felt like sharing that.

Anyway, I'm just curious, but is the player have its own designated spot in the object table so it's easy to look for? For collisions, I just thought of having the position of the player being one point, and have the hitbox have a separate position relative to the player coordinates. That way, if you crouch, the coordinates for the player don't have to move, only the hitbox. This is important because if you where crouching and then decided to jump, the player would start out partially through the floor, and could lead to you clipping through it.

Ideally the hitbox should be separate from the objects coordinates. I just did it an easy way that seemed good enough at the time.