Hope everyone enjoyed this one. I hadn't planned to do a third TMNT episode, but I was encouraged to do one by another RU-vidr I know, so here it is. In case you missed it - I'd like to release a Castlevania episode this month. Check out my post in the Community Tab for the channel to see what people have been suggesting so far. If you see something you like, give it a thumbs up. Finally, I received an email that Displaced Gamers has been nominated for Top Tech RU-vidr Award for #Noonies2022. That's really cool! You can vote here (for me or any of the other nominees!) if you'd like - www.noonies.tech/2022/emerging-tech/2022-top-tech-youtuber
love these videos - if a similar subject ever comes up again it would be interesting to use a TAS to force something like the button press conflict indefinitely!
I can't figure out how to put in one of your game genie codes in Retroarch, you go to cheats and then select NES, select the game, and then it's just a huge list of game genie codes (but yours isn't in the list for Retroarch) and there's no way to add a code manually. I messaged Jane's Rolfe and Mike Matei about making a video trying out your fixes, I'm sure they would love it. Fingers crossed that they find out about this stuff. That would be cool to watch.
This is one of my favorite topics that you’ve done. I was a teenager when this game came out and this explains why it was so frustrating to play (in addition to the bad collision detection you covered also)
I feel like you've done a lot already with Castlevania. Some things that might be interesting would likely be a comparison of Castlevania X (NA release) versus Rondo of Blood (JP release). I would personally love to see things about Symphony of the Night because the glitching in that game is insane but thats stepping out of the realm of 8bit/16bit.
I posted this on the last vid but nobody replied. =< Like, is there a way to get it back to 60 fps and make the code optimal enough to run efficiently? Do the enemies HAVE to respawn immediately? (Probably would fix the above issue if they didn't) Do the turtles' velocity HAVE to stop when they hit a ceiling like that? Make the turtles' weapons MAKE SENSE in their damage and other attributes. Such as: (the higher the number, the better. These are just rating numbers, not frames or anything. Range: the larger the num, the longer reach) Attack: higher num, stronger attack Speed: higher num, faster attack Stun: higher num, longer stun Knockback: higher number, more knockback Leo - range 3, atk 3, speed 3, stun 3; knockback 2 (balanced) Don - range 5, atk 2, speed 2, stun 3; knockback 3 (ranged/keep away) Mike - range 2, attack 2, speed 4, stun 4, knockback 2 (stunner) Raph - range 1, attack 1, speed 5, stun 1, knockback 1 (quick damage, basically multi-hit like bombs or the screw attack from Metroid) With this you could switch turtles for the situation you are in, though it would be slightly cumbersome to pause and do so. But you'd get more mileage with each turtle this way, and overall more fun! I have forgotten, does Select do anything? If not, and if possible, the game could pause and you could quickly select the turtle. Ah, I think it allows you to switch sub weapons..so it could still work that way, with the pause function still. Up/down to select the turtle, left/right to select the sub weapon. What do you think?
Imagine being an NES game developer and thinking "ugh, it's good enough, let's ship it. it's not like anyone will know about these bugs anyways," and then like 30 years later this guy just totally picks it apart. (Joking aside, I love these videos, you do a super good job explaining everything)
REALLY liked this one! As a game designer, rather than a programmer, this is the kind of thing that really gets my motor running lol And with that said, I don't really love the "just double the attack speed" solution, both from a balance perspective and from a "feel" perspective. Just being able to mash at infinite speed is fun for a bit, but doesn't generally tend to make games better, most of the time. I understand that a lot of what you did was in service of "quick and dirty" solutions to make them easier to implement for the sake of the video and allowing for applicable GG codes. So this isn't a criticism of the video at all. But assuming that we just had free reign to do whatever we want (while still trying to maintain the spirit of the game as originally presented) let's think about how we can do that! For Leo, I agree with getting rid of the two frame windup. Gone. Between that and fixing the audio bug, I think most of the "input lag" issues would be resolved, and the big problems with the animation revolve around slow feeling response time due to the distribution of active frames. Which is to say it's not input lag or animation length that's the problem, rather that 33% of your attack time is spent in the largely useless overhead position. To preserve overall feel and balance, instead of simply cutting the entire animation in half I would prefer more of a "redistribution" of frames over merely deleting them. Cutting JUST the overhead from 4 frames to 2, and adding them to the lower position will give the swing a bit of weight while increasing responsiveness and extending "useful" active frames. I would also experiment with things like 2-2-6, 2-2-4, and 2-4-4, as well as MAYBE shortening the cooldown. Leo always felt pretty good to use cause of good range and power, so I think he just needs a tweak to help timing attacks and "feeling right". Donnie I think just needs a similar, though simpler, tweak. He's already the most powerful turtle by A LOT, so anything that's just a straight buff would be game breaking. For him, I think cutting his wind-up in half (by four), and his starting "rear" attack animation by two, but adding another 6 or 8 frames of rear attack animation after the main hit would feel really good. The active forward frames would come out six frames faster, but with the same number of total frames (or +2 to to compensate for the faster activation). I might even do something like cut the cooldown by two and add another two to the forward strike, though again you want to be careful of making him too strong, and also holding a position for too long can start to feel really weird. Great video once again! The deep dive code walks are great, but this one was right in my strike zone for fun and interesting content 😂
It's perfectly reasonable to have long animations and cooldowns for balancing. What I'd consider most important is that the animation starts as soon as possible after the input. The hitbox doesn't even need to start immediately, but at least start the animation to give that feedback.
What I get from this is that the first major step to "fixing the game" would be to convert it back to 60FPS, which would be a whole project in itself because not only is the data processing designed around 30FPS, but the game already struggles processing it like that too.
I think this video is the perfect example to give, if someone goes: "Why can't you just..." - when complaining about things in games. It's not that they are necessarily wrong with their complaints, but - it's probably more complicated. It's always more complicated
Intentional delay could be part of experience, ie hitting with forks shoud be faster than swords; but uncontinuous detecting controller is terrible. Great video.
What an intro! I have never heard a better explanation of lag, especially not in 30 seconds. Knowing something is a far cry from having it put so concisely.
I either didn't watch, or didn't remember, your videos about the NES controller. Fascinating that this bug exists that a controller read might be corrupted by the music and so you have to read the controller multiple times.
I always thought of Donatellos attack lagging more was out of compensation for the distance covered by his attack. Which always came in handy when dogging Rocksteady on the crates. So I can't say if it was by design, and if so, whether it was good design or not that each turtle has its own balance of speed, distance, and direction of attack. Either way this was a really good video and I absolutely loved the music playing through out it.
Yeah I'd give Raph Donatello's damage to make him useful, and Don could take Mikey's damage and Mikey uninjured could use Raph's. Mikey does enough damage below half health to one shot some things, but the killer is that Raph's 9 damage is the lowest when turtles are wounded. This means his unique advantage of killing flying enemies in one hit is negated once everyone is injured enough, since every other turtle's damage is 10 or greater at that point.
I mean, the fact that turtles hurl more damaging scrolls when hurt below 50% but not boomerangs definitely gets me. If Donatello were to throw a boomerang below half health then switch back to his main weapon while the boomerang is still present, he'd do MORE damage somehow. Edit: Thought Don did 17 with his wounded bo staff, it's only 16 which is the same as the boomerang
Rebalancing TMNT could be a lot of fun. For example, Raph is the second strongest turtle at full health but the weakest at half health. I would give him Donatello's damage values to make up for the fact he has such a terrible hitbox, making him a viable character right away. Don could benefit from a faster attack and then get Mikey's old damage values. Mikey having Raph's damage values makes sense to me as well, leaving Leo mostly unchanged.
As a general rule, I think Raphael = Damage, Michelangelo = Speed, Donatello = Range, and Leonardo = Balanced would be a good rule of thumb for a balanced team that fits their personalities and weapons.
@@autobotstarscream765 While I do think these rebalances make sense, the manual writes that Leo is good for energy conservation, Mikey is courageous, Raph is good against flying enemies, and Donatello is powerful. Leo being balanced in my book was leaving him in third place and fourth place for damage since he has the best weapon coverage and second best reach. Actually, wrote up a table really quick: Leo's wounded value nerfed to 8, Raph's healthy and wounded buffed to 15 and 17, mikey's healthy and wounded buffed to 9 and 15, and Don's healthy and wounded nerfed to 7 and 12. Mikey still gains more damage from being wounded than any other turtle and it's actually buffed from 5 to 6, and with this change Leo never makes the new anti-air turtle Mikey worthless, because it takes 9 damage to kill the flying enemies.
Always look forward to these. As a half-ass Java and DB2 programmer I often think it would be more rewarding to do assembly coding. Everything now is configuration files and build systems -- not to mention I regard myself as a more linear thinker when it comes to programming as opposed to object-oriented. At any rate, keep up the good work!
6502 was my second programming language (1st was python). NES is very difficult but it can be done. Biggest thing is your code must be fast or it wont work. You'll do things that your CS teachers told you not to do, and you'll just have to be ok with that. But I can say that once you understand the language it all makes sense and you'll know why you're doing what you're doing (something I can't say about most high level languages)
I think it would be interesting to compare this to what Konami did with the NES Castlevania games. They clearly have a delayed attack cycle for Belmont's whip. Does it also have input lag rng, or is it purely by design? If it's the former, is this just something that Konami did with a lot of their games? Is their programming "style," if you will? Fascinating video! Keep up the good work!
While this is not a definitive answer, comparing Castlevania with Contra, I would assume Castlevania's design is intended. When Simon jumps, there is no way to fix the trajectory, just like IRL jump. (Both Contra and TMNT have a jump physics with leeway to aerial horizontal movement) Also in IRL, using a whip requires some wind up time or it just falls flat to the ground. The jump feature and the attack feature in the game suggests its players with methodical approach to beat the game than "just rush-in, dodge and kill everything" approach that Contra suggests. (It is hard to say what the devs intended for TMNT because it falls somewhere in-between and then steps sideways.) If input lag was normal for Konami games, I don't think Gradius or Rush 'n Attack ever became popular.
I am glad you did this episode. While the other ones shocked me more about TMNT in general, this one shocked me more about these old games inputs in general. The fact it could just randomly have different lags to execute is wild.
I'd love to see an analysis of TMNT (as well as Ninja Gaiden) surrounding enemy placement and respawn. I always hated the inconsistency in behavioe of killing and enemy, only to have them reappear because the screen scrolled slightly. Would be nice to have a code where enemy groups were specific to each area and weren't 'random', changing every time you entered a zone.
That would require *way* more than a game genie code. Those enemies respawn because the game doesn't store the fact that the enemy spawned by a particular spawner was killed. It only stores whether that enemy exists or not. If it doesn't exist, and the spawner is reached, it will be spawned. You would need to turn a two-state system into a 3-state system (unspawned, spawned, killed). That's outside of the bounds of a couple of Game Genie codes.
@@mrwizard5012 If I'm not mistaken, in one of those official books Nintendo released containing some behind the scenes stuff and interviews, Miyamoto said that stuff like that usually happened because the machine simply did not have enough RAM to store the state of things like "that enemy has already spawned and now is dead". And when you look at the NES specs and see how much RAM it has, that story sounds 100% true.
I love how you incorporate Game Genie codes, to fix the bugs, as in the minus world video. Using it as a kid for more lives, it's so cool to see now how it adjusts the game by memory.
Does the disassembly you show in these videos exist as complete code anywhere online, or is it just snippets that you annotated explicitly for use in the video?
Just got randomly recommended this by RU-vid and I already know I’m gonna love this video, so I’m gonna like it and then watch it when I’m eating dinner later
Amazing video! This and Legend of Zelda were my first NES games. I loved TMNT, and it's very nice to see how each turtle attack are totally different from a more technical point of view. As a kid, it was a challenge to manage each turtle, where to use them, and to master their attacks. It was a very important part of the game.
This was pretty neat! I'm disappointed I didn't see you replace the RTS with a loop when the 2 inputs don't match. That'd have been interesting to see what happens.
Amazing video. I'd love to see a video covering the major difficulties that would present themselves if one were to attempt to remake TMNT into a 60 fps game.
This is some of the highest quality and entertaining content in my subscriptions, hands down. It’s though provoking, nostalgic, and damn interesting. Thank you!
This has been super interesting, if you want an interesting yet really flawed FDS game to dig into I'd recommend Relics: Ankoku Yousai it's an interesting example of a first-time nes dev getting really ambitious and not writing the best code as a result. Underneath the rough design feels like a really fun game though. I'm planning on taking a whack at fixing it someday and am definitely gonna reference your commentary on design fixes in these.
It'd be cool if we could get some patched rereleases of these games you feature similar to Cheetahmen 2 The Lost Levels. Like, we get modern ports of Disney Afternoon games and 1st party games and such, and that's great and all but the only real benefit those ports have is "easier to access" and typically "cheaper than original carts," whereas games like Turtles could be patched to play like how they were intended and be a definitive version. Them with the release there could be 2 options (rom perfect and debugged) similar to how the NES and SNES classic have different aspect ratio options.
Minor point, but I feel like sense-react-confirm is usually more of a cycle than a timeline. It's rare that a player action isn't followed by something else they need to react to, usually influenced by their last action. (And obviously, technology is responsible for the time between confirm and sense.)
Goes to show how powerful hitbox detection is when even with all that lag on Leo and Donny's attacks, they are still the better of the 4 characters (although part of the reason Donny's is so much better is because he outputs way more damage than his brothers).
I remember when I got a copy of the Megaman Anniversary edition on the PS2 the game was practically unplayable because of the input lag. The game felt very sluggish, and definitely made me an unhappy camper.
I noticed that also with MMA and how it felt slightly different than the NES version, but would hardly rate it unplayable. To be honest I was overjoyed beyond words when I found out about a MegaMan 1-8 compilation for the PS2 and was more focused on the excitement of it that I hardly noticed any input lag. I'm a lover of retro games and having grew up on the Genesis console I thought it would be worthwhile to buy the Sonic Mega Collection, also for the PS2. And man that one suffered quite badly from input lag that even the sound had issues.
It's interesting that despite Don's excessive lag he is still the most popular and considered the strongest by most players due to the higher damage per hit mixed with the longer range. Certainly any fix would need to make some attempt not to just expand this imbalance even further.
Watching these just makes me think about how much of a gift it would've been to the original devs to have the emulators and debugging tools we have now. It's easy to criticize the flaws in games like this, but man, they were really working with stone tablets and chisels back then. Haha
Great stuff as always. Really curious though, what causes using the DMC to corrupt controller input, and in what way is it corrupted? That's a really interesting issue.
For anyone interested in a more comprehensive solution, running RetroArch with 4 frames of preemptive frames/runhead (in the Latency config) will compensate for the input lag on ALL actions (but only set it to 4 on TMNT, you generally want only 1 frame for other games.) Also turn on GPU Hard Sync with 0 hard sync frames (if using the gl video driver) or set Max Swapchain Images to 2 (if using the vulkan driver) to force the graphics card to output new frames immediately like a console would. Finally, set Frame Delay to 13 and turn on Automatic Frame Delay to compensate for USB controller lag.
DMC fix seems great. Not sure about frame adjustment on start, if only because Leo/Don are already the more useful turtles. I can't recall progressing anywhere in the game without Don usually being the main turtle, especially on bosses. That said, I'd be interested in a build with tweaks across your various videos about this game.
14:37 I see that controller corruption happens fairly often, but how often is it that it actually affects and delays your button presses? (To phrase it differently, what percentage of your button presses will get delayed?)
Good video, i have to mention thou that emulator lag is a solved issue, there was a period of time while ago where gpu drivers were way too optimized for thrughput , input lag be dammed and no way to deal with that from software side. Thank god we past that and now any emulator or program can have input lag down to raster lines if they care to.
Now I want a GG code for giving all turtles the attack power of Donatello and combine it with faster times for Leo and Don, that playthrough will be awesome!
Why would you ever use Donatello in that case? Leo has almost as much range (especially up and down), and his sweeping attack hits a lot of stuff. The only think that keeps Leo from being OP is that he only does half-damage. Rebalancing this game would be a good idea, but I don't think you can get good gameplay just from timings and damage changes. You'd really need to redesign the attacks themselves.
@@GeneralBolas I do agree, but at least I would switch between turtles more often instead of just using Don and falling back on Leo when afraid of losing him and use Mike and Ralf as sacrifices. Also, obviously just doing this for fun and mess around.
I typically use Raphael or Michaelangelo in the BOMB DEFUSAL LEVEL (sacrificial lambs) but I also tend to use them - especially Raphael in the later level with the flying enemies. Raphael and Michaelangelo kill most fliers in one hit while Leo takes more and Donatello's attack is too awkward to use while jumping. Raphael is better at attacking enemies diving towards him from above than Michaelangelo. Donatello and Leo's strength are being able to attack through ceilings and floors. Each stage lends itself better to a specific turtle and like MEGA MAN games, you just have to figure out who to use and when to use them. I will usually equip Michaelangelo with BOOMERANGS and always switch to him when I find them so I can add to his ammo. I will usually give Leo and Raphael throwing stars - especially the triple-star ammo. Donatello gets the KIAI SCROLLS since he's so powerful in the Technodrome level's final hallways. I can kill shredder from a distance with whatever weapons are leftover.
It seems likely the wind up delay on Leonardo and Donatello is intentional. Their weapons both have larger hitboxes. And Don's is the longest, and he gets the most delay. It's likely an attempt to balance the weapons.
I agree. It's clunky but it does look like it was done by design. The problem is that combined with the multi-frame input lag (especially if you got unlucky with the conflicts) it just feels too sluggish. At least they could have shown some kind of wind up graphic so you at least had some immediate response to pressing the button, then it would feel better.
3rd time a tmnt video, why do i feel you like this game so much? Also are you contributing you findings to an unofficial wiki (if there even exists one like for smw for example)
I am not a huge fan of the game, but I don’t strongly dislike it either. I’ve definitely spent more time dissecting it the last few months than I have playing it the last 30 years. I don’t directly contribute to wikis, but some people have cited some of my videos in various places.
So. TMNT runs at 30 FPS and still can't maintain framerate. Wow, when you mentioned it ran at 30FPS before, I'd forgotten that it had slowdown. What is going on in this game? Is it really *that* terribly coded? I mean sure, there can be a lot of enemies on-screen, but it's not like they're colliding with each other. Just terrain. It should not have been hard at all to load balance collision checks: on frame A, do half the collision checks, and on frame B, do the other half. My guess is that they didn't do that. They hard-coded certain code to be in frame A and other code to be frame B, but one of those two frames can get overloaded.
Great informative video. One other thing that always annoyed me was how some of the turtles strength of attacks varies. Sometimes the same enemy will require one or two hits. This didn't seem to affect Donatello or Raphael. Leo and Mike's attacks were sometimes weaker. I think it was just another way to balance each of the turtles.
The turtles have two separate damage output values. One when they are above half health, and one when they are below. Don is still the power hitter, Leo gets a slight increase, Mike gets a big increase, and poor Raph has the same value for both states. I wish I knew about this as a kid. Such a deep game and I believe most things the programmers did was intentional to an extent.
I've watched this and other videos you've made on the same game... How about a diff patch to patch these issues out? It would be nice to be able to play an idealized version of the game.
4:22 wait so the game checks collision against an invisible sword for no reason? doesn't that basically prove that the code was cut-and-pasted from Getsu Fuuma Den?
8:18 Could you remove the first $00 byte from the sequence to get rid of the wind-up? I guess it would take more game genie codes than you have to move everything one byte over, unless you can change the starting address, though.
My focus is exclusively videos right now. If people would like to talk information and run with it, that’s cool. Maybe I get mentioned in a readme file or something. Ha
Wait a minute... we all know running the retail version of the game , i.e., no changes. Donatello's bow is the best weapon. Now, isn't that because it has such a large wind-up and attack, causing it to stay on screen longer? Here's a question...is every part of the Ninja Turtles weapon's fatal to the enemies in game? If so... Wouldn't it make more sense slowing down the very last part of the attack, thereby giving said weapon more screen time. Like if it hangs for a second after an attack, it would inadvertently take out another enemy or two...that could be neat.
I wonder how much work it would be to move the controller polling so that it happened /before/ the attack gets processed (instead of happening way down the frame?).
Boy howdy! How could Konami be so inefficient with coding? This was back in the day when they were cool! Oh wait it was also back in the day of command lines and in house dev tools. Combine that with tight deadlines... because *there is always a tight deadline!* and you get results like this.
I’ve always wondered what in the six hells takes so long in the logic with this game that it takes two render frames to get it all done. It’s not doing anything interesting or complicated. That would be a fascinating exploration. A decent amount of jank, like the input processing, would just go away if it did logic at 60 like every other game.
Got to say, with all this talk of frame rates, I'd be interested to see how the PAL versions compensate (or don't) For those unfamiliar, PAL TVs have (had? I legitimately don't know if it's still a thing...) a higher quality picture (that is, more scan lines) but a lower refresh rate (50Hz)
Hmm...I don't agree with just getting rid of wind up time altogether. That ruins what little balance the game has (or at least the attempt at balance) and, for a lack of a better term, "weight" of the weapon.
You could always use a table to hold a separate wind-up for each turtle and apply it for those frames first before using the attack frame lookup if you wanted to adjust Don. It can’t be fine-tuned too much since every number used is doubled for output.
