Although most of the enemy cast can be dealt with by shooting from a crouching position, the stage design does a good enough job of using the stage terrain in a way to force you out of it, particularly Stage 1, 3, and 5, on account of actually presenting different planes for you to vault around, and using doors to spawn enemies right next to you.
Stage 3 is mostly a retread of Stage 1’s ideas, but with the addition of Mutants to spice things up. Initially you will have to get used to their ability to leap over your shots when they get close, but over time you can mitigate the threat they pose by memorizing most of their positions. One interesting thing of note is that Mutants cannot enter the screen via doors (nor can any other non-Masker enemies, although all other non-Masker enemies beside the Mutant suck and are better off not appearing at all) and only appear from the sides of the screen, which all things considered makes sense, if they spawned within leaping range they’d come at you so fast that you can only survive by memorizing these spawns. However, this lack of response time could be mitigated if Mutants could only spawn through doors on the plane that you’re currently not on, so you would still have adequate time to react.
Stage 4 opens with another staircase where you have to wait for the enemies on it to drop down into your line of fire, and then giving you Owlbats to deal with. After that you have to drop down into a pit, but if you take the most straightforward path of simply dropping down, a Ninja Masker will suddenly spawn in your face, which will make you collide with another very inconveniently placed Ninja Masker. Instead, you have to descend platform by platform to get down safely. This isn’t particularly hard to execute or get around, it’s just some bullshit designed to get first-time players considering the Ninja Masker isn’t remotely telegraphed. And after that you get a platforming section, which again involves more trial ‘n error. The first part is the running men-on-fire that keep bouncing around the screen, and split into four ascending spirits when you shoot them. You have to memorize what platforms to shoot them from, else their erratic bouncing will make it infeasible to get them in your line of fire (especially if you’re on a higher platform and they’re on a lower one), and shooting them while they’re close means their released spirits can get you by surprise. Again, it’s easy and consistent once memorized, but their erratic nature is primarily designed to catch first-time players by surprise. But the second part involves the actual platforming itself, which takes some time getting used to, as you’ll probably often end up falling off platforms by accident while trying to jump. This is mainly because the player’s hitbox and neutral stance is pretty thin. Most platformers allow the player character to eke over the edge in order to make jumping off the edge off a platform more lenient; you can see how the character’s sprite is standing with one foot on the platform and the other on thin air. But as this isn’t the case in Rolling Thunder, you have to be very precise with your jumps, which IMO is a bad fit for a game that primarily isn’t even about precision platforming at all.
Stage 5 is where all enemies and bullets start moving faster. Even though the actual enemy compositions aren’t that hard by themselves when compared to previous stages, it’s certainly more engaging because of their numbers and all the different directions/doors they are coming in from, on top of the increased enemy/bullet movement speed. In this stage the foreground/background plane switching comes (somewhat) into greater play, although bizarrely going into the background is never required to progress, so you can just ignore this altogether. There is an interesting dynamic to combining background/foreground switching on top of the existing plane switching; being in the background protects you from enemies on the foreground and plane above you because enemies on the top plane cannot jump down into the background (but enemies can jump from the background towards the top plane), but you still have to contend with enemies spawning from the background doors, and you eventually have to exit into the foreground through a chokepoint doorway. Being in the foreground means you no longer have to deal with the background doors, but you’re still vulnerable to enemies on the top plane, and on the top plane you’re vulnerable to background doors AND the bottom planes on the foreground AND background. This kind of asymmetric set-up in terms of potential risk is something the terrain could have played into--where there’s trade-offs in taking one plane over the other, with hard counters forcing you away from one particular plane. Sadly the terrain in Stage 5 never plays with this, so the foreground/background switching ends up being an underutilized gimmick.
Stage 5 isn’t the end; you have the second loop as well. It’s not just the same thing as the first loop but slightly faster; you get new enemy types, revamped enemy placements, some stages feature redesigned terrains, and the Stage 4 in the second loop is a completely new stage that doesn’t resemble Stage 4 in the first loop at all? Given how short the first loop is, (10 minutes) this is basically the second half of the game, and if you intend on playing this game you shouldn’t skip out on it.
So Stage 6 (or ST2-1) features way more enemies, tougher enemy types appear more often, and previously flat hallways now feature enemies poised to drop down from the windows above. The second half features laser gates that you have to time your approach through (while being mindful of their
wack hitboxes), but their inclusion feels kinda wasted. You can usually dispatch all enemies first and then safely pass through the gates while the lasers are down, although this way the laser gates always pose the exact same challenge regardless of how the enemies are placed. Laser gates would be more interesting if you were forced to pass them while having to deal with enemies that you can’t permanently remove. Doors to both of your sides that spawn Maskers at a higher rate would be a great fit). Laser gates aren’t used again until Stage 9 (and in the same lacklustre capacity to boot), so there’s some wasted potential.
The most bizarre change between loops is that the final stretch of Stage 6 features enemies that will
jump up unannounced out of the sandbag mountain you are standing on. You cannot see this coming because the sandbags are obscuring their sprites (save for teeny bits sticking out), and whether they will jump up to begin with is mostly random. You can kind of route around this, but the only way possible against a threat that you
literally can not see coming is through trial ‘n error. This part is a headscratcher and I cannot even guess what the rationale behind it is.
Stage 7 plays mostly the same as Stage 2 with slightly tougher enemy types, the only noticeable changes are the addition of Owlbats in the crate sections and Mutants in the staircase section. For Owlbats you want to memorize their positions so you can pre-fire them while they’re still on the ground, so you don’t have to wait for them to stop flying. The Mutants don’t really add anything to the staircase section, all you do is stand on top of the staircase and shoot while standing as all the Mutants leap into your line of fire. It’s a static threat that doesn’t overlap with anything else, and is repeated several times in a short timespan in the exact same way. Just a waste of time.
Stage 8 doesn’t look all that different from Stage 3, but then it keeps going, and it
just keeps going, until you’ve realized that they slapped a remixed version of Stage 4 at the end of this stage. The No Miss clears on YouTube can only manage to finish this stage with less than 20 seconds left on the timer because of its sheer length, and only with an airtight route and a compliant AI at that. The role of time limits should be to discourage excessive dilly-dallying and encourage more aggressive and engaging playstyles. All making them this strict/making stages this long is make only the most optimized routes feasible, while making any elements of RNG a massive pain in the ass.
One such RNG element is the new Panther cage. Whereas you could just pass by it via the foreground in the first loop, now the foreground path is blocked off, and you have to get around it by going into the Panther cage. And the only way you can do so is via two narrow doorways, while there’s a massive group of Panthers on the other side unpredictably walking around and making it unsafe to enter it. All you can do here is wait until an opening presents itself--costing valuable time, or try to force one by jumping and making all Panthers jump along with you as well. But all of this is subject to a lot of RNG, because sometimes the Panthers will decide not to jump (because they randomly decided to growl), sometimes they will decide to leave their cage and get the jump on you, and how they move around is unpredictable.
Particularly cruel here is that if you move a slight bit to the left before entering the cage,
a Panther can spawn on the left side of the screen and immediately maul you before you can react. There’s nothing inherently wrong with enemies spawning in from the left, but the player character’s position then ought to be locked to the center to give you enough time and space to react to this, as opposed to Rolling Thunder 1 where your position is locked a one-third screen width from the left. This makes RT1 more suitable for threats that come in from the right, but certainly not from the left edge of the screen.
The ensuing lava section has to be one of the worst cases of trial ‘n error in the game. First you get an Owlbat that for some reason decides to fly around the bottom of the screen instead of the top, and in order to hit it you must jump on a platform amidst lava. The only issue here is that this platform is placed just close enough that if you jump off the very edge of the prior platform (as the game taught you several times before), you will overshoot and land on hot lava instead. Air control is very limited in this game and your jump arc is mostly fixed, so you can’t adjust your trajectory mid-air by much. The only way to avoid this is to unintuitively jump a few steps before the edge of the platform. Correct me if I’m wrong, but I don’t recall even Castlevania ever doing this. Again, there’s nothing inherently with this kind of setup, but Rolling Thunder isn’t primarily about platforming and never drilled fundamentals like these into you, so being expected to suddenly know about not having to jump from the edge of a platform (while the time limit is almost running out) is a bit obscene.
This is also why the last section of the stage where you have to platform across incredibly narrow pillars also feels rather out of place, but the most offensive thing about it is the men-of-fire that get spawned mid-jump--in a game where you can’t attack while jumping. Even if you land safely, they will inevitably collide with you and kill you. The only way to avoid this is to pre-fire a bullet before you jump, which is some mighty kuso. I did say before that I liked queueing shots for its speedrun potential, but only because it was optional. Forcing the player to pre-fire obstacles is always going to be a problem if you don’t telegraph what’s ahead.
Stage 9 is on the mild side, all things considered. Not a lot of doors or multi-directional spawns to worry about. Instead you get a lot of boring sections where you have to drop down something, but have to wait for a Masker at the bottom to move out of the way so you don’t collide with them. There are also more Stage 3-style Ninja Masker pits, where you gotta drop down several platforms, but standing on particular platforms causes Ninja Maskers to spawn in your face, so you just have to trial ‘n error your way through until you find a consistent route. Not particularly interesting on replays. The only interesting obstacle of note is the two laser gates stapled to each other that activate at asynchronous intervals, since you need to use your brain a little to find a gap between the laser gates, although their RNG nature makes it crappy when playing for speed. One thing I’d suggest for laser gates in general is to have them move back and forth horizontally, allowing for more interplay with enemy spawns by making laser gates more of an overlapping and persistent threat, as opposed to the current implementation where you can just chill near one and take out all enemies first, effectively isolating the element of enemy placement from the stage hazards, instead of combining them.
Stage 10 is more of the good ol' stuff. More enemy swarms, less gimmicks. However, the foreground/background switching is still mostly boring, and there is one crappy part that involves enemies pre-firing at a place you have to drop down towards, which again involves more jumping to get the AI to GTFO. At first I was convinced that the extension of the final stage in the second loop was some bullshit that you couldn’t reliably pass through, namely having to ascend a staircase populated by Maskers that refuse to budge unless you bait them into dropping, so I
used an invincibility glitch to waltz past the whole shebang. Turns out my Google-fu just wasn’t up to snuff and I somehow missed the existence of a No Miss clear before the time of recording. Designing games around gaming the enemy AI like this is tricky, because behavior is not something that you can really telegraph until it’s too late, like the Red Arremers in the Makaimura games or Donovans in Streets of Rage hard countering your air attacks. It’s more tolerable in beat ‘em ups because you have a wider margin of error thanks to the beeg health bar the genre usually affords you, but less so in precision run ‘n guns like Rolling Thunder. At the very least you want to teach the peculiarities of the enemy AI in a controlled environment early on in the game, not the absolute last stretch.
Finally there’s the final boss fight against Piccolo himself, where you must face his most powerful technique of Running Into You. This is actually an interesting fight where you must constantly gauge how many shots you can afford to let loose before having to jump over him again, because the distance after which he decides to turn around after you jump over him is AFAIK random. And often that distance is so short that you have no choice but to immediately jump again. Cool fight, but the randomness makes it bad for speedrunning, which is also especially bad in a survival context, because if you no missed everything up until this part you will likely have about 20 seconds remaining on the timer, and RNGesus making you pay more of it really makes you want to scream. However, all this is assuming you let the fight play out as intended. Instead, if you have at least 25 bullets, you can simply spray him to death before he even gets close. You can do this by conserving SMG ammo throughout the stage, but if you enable autofire your pistol basically functions no different from the SMG anyways.