Lmao, nice number plate.
Also, presenting a no-memory-corruption run for the benefit of a mental illness charity is a clever move (if a bit risky; I made a double take to ensure it had that "no" at the beginning).
Drop it, ALAKTORN, I know how it feels like to vote No on a submission with many Yes. I'm certain that there are other people that dislike the extremely broken nature of this category, although this criticism didn't appear in previous submissions.
I'm not ALAKTORN, I'm DJ Incendration. Why did you need to call me ALAKTORN? Also, how do I manipulate the TID to 0x64d0 (in BizHawk 2.3.1)? I tried to do it with a bunch of inputs and frame delays, but I can't seem to get it. Thanks in advance, from DJ Incendration, not ALAKTORN!!
If only it was just animation... Pacing, framing, editing—every directorial aspect is a mess at best, failure most of the time. And the sound design is not much better; even voice acting feels lifeless. I don't blame them.
I mean at first I thought it was only going to be worse than S1—which would be forgivable by itself—but it ended up being just bad. I have seen dumb otakubait comedies done with more love and effort, which is not the kind of competition you want for a franchise like this. Apparently the show is already so deep in production hell that episode 2 listed 5 chief animation directors, 3 animation directors, and 3 assistant animation directors in its credits (normally you have 2–4 people fulfilling these functions). The blame should be shared between both the production committee for poor organization and the studio for poor resource management (and agreeing to this job in the first place, seeing how their schedule was already swamped), and I sincerely hope it will receive a bad enough backlash for them to change their approach to S3 and beyond.
For sure, it's just that AngerFist isn't normally into stuff like that. As far as I know he prefers gritty, visceral action series.
Katanagatari is kind of the opposite: it relies on building up to the fights—while the fights themselves are often short, spaced far apart from each other, and not necessarily violent. But it really becomes great once you get emotionally invested into the characters, because they're all wonderful and distinct personalities.
Btw, I felt like the final episode in particular took some inspiration from Bruce Lee's unfinished version of Game of Death, and possibly from the 2009 film Goemon (at least visually—I know the novels were released before the movie).
Also, holy shit the second episode of OPM S2 is a smoking turd. Not going to watch this travesty any longer.
Hello fellas! Was a long time ago I checked for new anime series. My preference hasn't changed a bit, looking particularly for a show like Shiguri: Death Frenzy, Kurozuka, Ninja Scroll etc. If you know any new series like those, please let me know.
Oops! Missed your post (actually both of them). :v
In terms of style, you might be interested in Devilman: Crybaby. Extremely gritty, violent, lots of nudity and shit. Rushed ending though. If you like it, check out Kemonozume from the same director; an unconventional and very underrated show. Actually, you know what? Scratch that—watch Kemonozume first.
The currently ongoing Dororo is on the tamer side, but is closer to your tastes thematically (feudal era Japan, swordfighting, demons, people dying and whatnot) and is extremely competently done. Also Onihei I suppose, but that's more like a very realistic episodic procedural cop drama set in the 18th century Japan.
Also check out Attack on Titan if you haven't yet. And JoJo. Can't go wrong with either: good gritty action, decent characters, well-written plot and excellent visuals. And if you've liked One-Punch Man... Well, I won't comment on the second season's quality—for now—but do yourself a favor and watch Mob Psycho 100 from the same author. It's every bit as good; arguably better. Similar themes but more interesting characters and more plot development compared to the season 1 of OPM.
On to the stuff I'm not sure about. Katanagatari is a contentious recommendation; it kind of fits the bill thematically (swordfighting, ninja...) but not at all stylistically (there's a fuckton of dialogue and not remotely as much action; the art style takes a while to get used to). But it's an amazing show for what it is, it's clever and has amazing soundtrack and animation, so it feels unfair not to mention it. People seem to really be divided about its ending, though! Golden Kamuy... I would certainly recommend it if it wasn't so inferior to the manga. The latter is a top read, one of the best stories I know—with amazing characters, too. The adaptation is... acceptable I guess? It follows the plot very closely, but you'd be losing out on a lot of expressiveness of the original.
And on to the stuff that hasn't aired yet. If you've liked Baki you're probably going to enjoy Kengan Ashura which will air next season. It's stylistically similar and the manga is one of the most fun martial arts stories I've read. Excellent character cast! Also you absolutely shouldn't miss Vinland Saga when it airs. It's going to be amazing on every level.
EDIT: There's a lot to say and show about JoJo, but if I were to pick a single non-spoilery scene to try selling you on the series, it'd probably be this one (it's uncensored on the blurays, obviously). Just the perfect mix of deadpan seriousness and whimsical absurdity. It was this exact scene that convinced me that the new JoJo adaptation was going to be quite the ride.
Well, this is a thing now.
Link to video
Most games still aren't playable, but this is pretty good, eh?
I've downloaded the current build and poked around on their Discord. The emulator is pretty bare-bones right now, missing almost all of the essential features such as controller configuration (which made me not want to continue any experiments for now). Some are coming in the UI overhaul, but there is no ETA on that. Savestates are "apparently possible but are not a priority atm". The devs have been busy lately so progress has slowed somewhat. So TASing is definitely out of the question for 2019, but 2020 is probably within the realm of possibility—at least as long as savestates and sync-stable input replay are possible.
Apparently this is still (once again?) actively developed. The current stable version is 1.8.0, and the latest git commit was made today. Versions released since the last post in this thread have some beefy changelogs, too.
I ran a coupledemos on it using default settings, and only saw what appeared to be minor emulation artifacts in either (the first had some issues with effect timing in a couple spots, and the other some issues with a glow effect). I'd consider it pretty good.
Judging by the multitude of YouTube videos, commercial games run fine...-ish? Anyway, that's it for now.
What it usually means is that the game doesn't calculate a trajectory for a ball (that would require processing quite a bit of floating point math which NES is extremely bad at); instead it has a table that contains data like "for this angle and speed, move the ball X pixels to the right and Y pixels up next frame until a collision happens". That way it bypasses the need to do complex calculations by having the number of variations boiled down to a reasonable minimum and having immediate access to the result of those calculations.
It's the math equivalent of prerendered 3D sprites: instead of processing the entire geometry of a 3D model, store a 2D image of it at 4/8/16 different angles and replace them as needed.
I don't have an opinion on the TAS itself, nor its quality, but let me chime in real quick with regards to authorship attribution, because there is some visible discontent brewing between here and Discord, and it needs to be addressed.
Our attribution policy for modified input is notably lax to avoid situations where every new author would be required to include all past and present collaborators, so the entire body of work on a given game published on our site can be considered a collaborative effort where the content is published already with intent of further reusal—especially since it becomes inevitable as TASes become progressively more optimized and thus converging to a single optimal set of inputs. The contribution of every author is recorded in the site's history to comply with the CC BY 2.0 site license, which also allows building on others' work without asking for permission—this is something every movie author and editor agree upon when submitting any content to the site.
This moves the attribution of a heavily modified work strictly into the domain of community ethics. The author is encouraged to list every source of input that was directly copied into their work, and is discouraged from taking significant portions of input from other players without properly crediting them—within reasonable bounds, of course. But unless an overwhelming percentage of input is directly copied from elsewhere without any attribution, this is not something that the site can—nor should!—enforce one way or another, and it certainly shouldn't influence the judgment process or user rights.
And this also isn't something to admonish beginners for, considering they cannot be expected to be fully acquainted with intricacies of our etiquette in their first few months. It's not possible to tell at a glance that what's been done here can be frowned upon, since the borrowed input constitutes only a tiny portion of the TAS's overall runtime.
In my personal opinion, it would have been reasonable and of proper courtesy to mention in the submission message that inputs in 1-1 and 1-2 were taken from HappyLee [for any sensible reasons]. And that's really it.
Carry on!
As someone who once attempted a no-reboot true pacifist Extreme difficulty playthrough, I can confirm Extreme takes over twice as long to complete because 1) you actually need to explore every stage and kill most of the enemies in it for Nano pickups; 2) you still cannot kill directly, so in many cases you need to set enemies up to kill themselves by splash damage off of your body—but since you can take at most one heavy hit without dying, you have to constantly backtrack to heal yourself. Running this category on Normal is a wise decision in my opinion; Hard would just be slower and not necessarily more exciting because the no-damage Ultimortal run already exists.
Feos, thanks for the comparison encode! Very interesting seeing the two side by side. One trend I noticed was that Baxter's TAS often had to spend a lot of time clearing the last few silver blocks, whereas the bot-generated run tended to get these as it went.
It also spent a lot more time actually using the paddle instead of the wall and ceiling bounces, leading to e.g. less shallow angles—even more consistently so than I'd thought. You can clearly tell how Baxter was almost always going for a more human-intuitive strategy which nevertheless led to accumulation of time loss closer to the end of level: even when the balls spent more time closer to the blocks, they almost always spent less time actually hitting them.
For instance, in level 11 which is comprised entirely of silver blocks with no multiball power-up, the strategies used in both runs were largely similar, but Baxter opted to eliminate more blocks as soon as the opportunity arose instead of spreading them out like the bot did. His approach ended up concentrating unbroken blocks at the edges which required a lot of idle time to get in the end—whereas the bot solved the last dozen of blocks by cleverly ricocheting the ball between 3–4 at a time because they were conveniently spread out. It's almost magical.
Level 15 is an even better illustration of Baxter underusing of the paddle in favor of sending the balls to the ceiling gap. IMO this comparison video should be linked in the description to demonstrate the differences in optimized human and nonhuman approaches.
Also could you explain further by what you mean by expect more of this game?
I mean that for how fun it is to play—especially with a friend—it sure looks much more monotonous in a TAS (that tends to happen, yeah). I suppose bumping is to blame for that? But if you say there will be more variety down the line, that's good to know. I'll watch it in any case.
Thanks for the detailed response.
Breathtaking. Some solutions even seemed a little counterintuitive, such as in level 30 where I would expect significantly more time for the balls to be spent in the upper half of the screen. But very entertaining throughout.
Going into this, I knew it was (semi-)bruteforced, but it wasn't until I finished watching that I read the text entirely and realized what it was exactly that you did with the game to achieve it. Recreating the logic perfectly AND spending a year of CPU time running the simulation? Jesus almighty, that is some insane dedication. Certainly a TAS of the year contender in my opinion. Submissions like this are a real gift, and I'm continuously impressed by the quality of works that pop up here on April 1st, of all days.
But you know what? One of the best takeaways from this, for me personally, was that Baxter wasn't far behind the optimal solutions to a set of incredibly hard problems. What a lad!
Huh, I expected more from this game. Alright, I have two questions on the basics.
1. Is there a particular reason you keep adjusting your angle on the way rather than immediately?
2. Is there a way to manipulate the enemies to converge on your location sooner so you can, say, kill two at a time (even if it takes slightly longer to engage them)?
We did this in the form of a world record progression video for warm up mixing cheated, tool assisted and live runs.
The video was a bit on the longer side, but it was interesting to see all four major style decisions—stumping vs. stretching at the apple, front vs. rear wheel at the flower—constantly changing between different the records without a conclusive statement as to which one was faster. Even the very last 13:88 clip used pretty much the same style as the 14:33 from 2000, and not the one from the 12:99 TAS.
I think "manipulated" is just a misnomer in this case. It looked like the drops were punched-in based on the monsters' loot table. So if a monster could drop something, igloosPat would simply choose what exactly the monster would drop, and disable all else. Or at least that's the impression.
Note that it probably won't affect the legitimacy of this particular run because its legitimacy claims extend over the possibility of an item being able to drop with particular affixes from a particular enemy (which it adheres to as far as I can tell), but not that it would actually drop in a reasonable amount of time if you tried replicating it. There's probably no way to replicate all of these drops in an actual TAS in a way that would make it faster. Unique/set items and runes would be by far the easiest ones to procure.
Well, it's not something you'd be able to submit to TASVideos, because 1) there are no means of reliably reproducing the run on another person's hardware by means of a player input file; 2) you have tampered with the game's RNG by means other than providing it input as part of the run. These two disqualify it right away.
But it is something we at TASVideos are thematically interested in and want to see more of.
Ohhhh, this is some good shit. It wasn't that long ago (less than two years I believe) that I was thinking what would a theoretical TAS look like. I did not envision javelins, however! That was truly refreshing to see.
Note that the turrets in stages 4 and 6 are different; the ones in stage 4 spawn once and leave a ground star on destruction while the ones in stage 6 respawn continuously until scrolled to the bottom part of the screen. This may or may not be relevant. Both spawn within the scrollable area, however.