It's all in your head

[Udderdude]

Really interested in seeing the full reveal and if it really "ruins the game for everyone".

[Plasmo]

[CStarFlare]

What happens next is game ruin

[Bananamatic]

don't tell me that the game fucks up so hard that the goal is to get as close to 99,999,999 as possible without exceeding it in both loops
even worse if it's done to avoid some weird arcade rule that would cause a leaderboard shutdown

[blossom]

That's an interesting point - if the glitch is as bad as the top players claim (still have my doubts), then a counterstop and a leaderboard lockdown for one of the most popular arcade shmups would probably be detrimental for business.

[Prales]

.

[casualcoder]

I guess MarkMSX pointed it out already that whatever happens next kind of doesn't matter because as soon as SOMEONE discovered it, it ruined the potential for a legitimate/fair leaderboard.

That bird is long since dead. Now I'm just interested in the autopsy.

[blossom]

I don't understand what you're saying - assuming what blackisto found is indeed "the glitch" and not some new glitch, how would players not be on equal footing with a fair leaderboard?

[Bananamatic]

I guess MarkMSX pointed it out already that whatever happens next kind of doesn't matter because as soon as SOMEONE discovered it, it ruined the potential for a legitimate/fair leaderboard.

That bird is long since dead. Now I'm just interested in the autopsy.

The potential was ruined the moment someone decided to use the glitch to get a high score while pretending everything is normal...meanwhile, they're using a glitch that has capabilities to destroy the leaderboard and even the operators seem unaware of it

fix the rules instead of pretending nothing is happening

[heli]

Well, what was the bug then ?, or is it still sercret.

[casualcoder]

Well, what was the bug then ?, or is it still sercret.

I think it's going to be explained at the event that's on right now. I'm not really looking into it too much, and will probably catch it later.

Re: ruined leaderboard. I guess the assumption is that the glitch either produces unreliable results, heavily weights the final score (such that scoring details that used to be important are no longer), leads to a counterstop or some combination of any of those undesirables. Even if it's fair and everyone can exploit it equally, strictly speaking, it leads to result where pretty quickly nobody is going to care about the game. The only way I can see that not happening is if the glitch accounts for only a small part of the overall score or is trivially easy to produce and leads to roughly the same score result every time. Then it just becomes a matter of subtracting that amount from a run to get the "real score."

I don't know much about the glitch, but I'm sensing that's not the case here.

[Prales]

.

[Udderdude]

All hail Glitcho, our new overlord and savior. Counterstop TAS when?

Also, I don't think this glitch actually ruins the game in any meaningful sense.

[Queen Charlene]

the glitch is hard enough to do consistently that i don't see it ruining the scoreboards, yeah. just put it in its own category for those gods who do manage to trigger it consistently and it should be fine.

[heli]

You all talk like you already know it.
Why are you making me extra curious ?
I want to hook up my saturn and go for it.

[mikejmoffitt]

It was linked earlier but here it is again.

The bug is a result of leftover stale data being left in one of the CPU's registers, and Cave forgot to assign a score value to Glitcho. That means you'll almost always get 10 pts for destroying him, but if you get him on the same frame that a "Laser Hit" is registered, you'll be awarded your GP Value instead of 10 pts, all because of a stale memory read.

in the PSX and Saturn versions... I'm not sure if this bug is still there. Those are both very direct ports, probably machine translated to a large degree, so the chances are high that the same mistake is there (e.g. forgetting the hardcoded score value for Glitcho). However, since the CPU architectures are different from the original Motorola 68000, I don't know if the same register is being used, or if the score math could have changed, so it's a complete unknown what the effect on score would be if you were to reproduce the bug there. It's not as if they are already treated the same, but this is now an additional reason to not count scores from those home ports in the same category.

It is unlikely Cave ever knew about this bug nor went out of their way to fix it. As a result, I bet that barebones Xbox 360 port included in that other game still has it, and the behavior should be the same. Campaign Version will certainly have it. Whether or not new versions have it is up to whoever does the port... for that, though, we'd need to know the intended value for Glitcho, since it's not 10 pts.

Also - Enemies only take damage every other frame, when they are flashing the bright blue color. Similarly, the Laser Hit counter only decrements every other frame. This gives us a semi-leniency across two frames, so long as the player is aligned to the correct frame. [/b]

[casualcoder]

http://electricunderground.io/the-secre ... -revealed/

Bit unclear to me in general but it doesn't sound super hard to pull off. Apparently there are a few chances at different points with wild score potentials.

I suppose the really huge score potentials are dependent on the existing counters being pretty high so that's probably the hardest part, but it doesn't seem trivial to "mortal" scoring runs.

I suppose the best way to test it out is to do a save state and devise a way to narrow the frame window with a visual queue. People will get good at it. It's just like 1-frame links in fighting games. Practice it enough and what at first seems impossible ends up a high-percentage play.

[brokenhalo]

Counterstop using Kinect controls when?

I don't think this is as "game-breaking" as people have theorized. Getting the stage chain down is already pretty tough, and then executing 3 or 4 frame perfect tricks during the course of the stage brings the difficulty up quite a bit higher. I'm sure someone will put together a counterstop TAS in the coming days, but we've had a Donpachi counterstop TAS for a few years now and nobody has ever managed to pull that off in a real run yet.

Good luck to the handful of people on this forum who have a chance at pulling this off though. It'll be awesome to see this in a real run.

[Bananamatic]

Even in japan only WTN managed to exploit it to a significant degree and we have 1 active player here capable of chaining stage 6 normally, what makes you think people will suddenly learn the game with several frame perfect tricks on top of that?

I suppose the really huge score potentials are dependent on the existing counters being pretty high so that's probably the hardest part

There is nothing particularly difficult about having a high gp value halfway through the stage, especially compared to hitting 4 frame perfect inputs

It's just like 1-frame links in fighting games. Practice it enough and what at first seems impossible ends up a high-percentage play.

It's way harder since links are dependent on your own sequence of inputs and are isolated button presses that come down to muscle memory compared to trying to hit something frame perfect according to automatic screen scrolling, not to mention it also depends on your vertical and horizontal position being nearly perfect on top of that

shmups really are 3d games with the 3rd dimension being time that's out of your control, which is absent from most 2d speedruns that ironically enough revolve entirely around time

[Sumez]

the glitch is hard enough to do consistently that i don't see it ruining the

A frame perfect glitch like this definitely shouldn't be allowed on any scoreboard unless you do want the game ruined.

[heli]

The low level programming language caused the bug.
Glad C or C++ exist, imagine writing such a game with ASM.

[Sumez]

You wouldn't have a bug like this if the game were written in C, but that doesn't mean you wouldn't have bugs. On the contrary it's easier to cause memory overflow etc. in a C program.
Honestly I find assembly a lot nicer to work in than C/C++.
If I want to do high level stuff, I prefer complete abstraction with strong types, forced OOP, and managed code. C is a funky middle ground that exists mostly to make stuff more portable.

[blossom]

the glitch is hard enough to do consistently that i don't see it ruining the

A frame perfect glitch like this definitely shouldn't be allowed on any scoreboard unless you do want the game ruined.

This is an anally strict glitch in a game which has an anally strict chaining system. I don't imagine the type of person who is already invested into playing Dodonpachi for score will be terribly bothered with the execution of said glitch. If he were, honestly I'd say the reaction would be strangely hypocritical.

[Bananamatic]

Glitcho acts a little weird in terms of his Base value points but normally gives you a mere 10 points, which is the lowest existing Base value in the game. There are only two other sources for a 10 points Base value: the rockets in the beginning of stage 6 and laser hits.

So what exactly happens when you pull it off on a st6 beginning rocket? Do they actually have the 10 point value coded correctly or is it missing as well?

[Plasmo]

Glitcho acts a little weird in terms of his Base value points but normally gives you a mere 10 points, which is the lowest existing Base value in the game. There are only two other sources for a 10 points Base value: the rockets in the beginning of stage 6 and laser hits.

So what exactly happens when you pull it off on a st6 beginning rocket? Do they actually have the 10 point value coded correctly or is it missing as well?

Saying that Glitcho's base value is 10 is an oversimplification just to make things easy. In reality he has no (hard-coded) base value (made clear later in the article). The rockets correctly have a 10 point base value and moreover wouldn't be an object you could get a laser hit from anyway.

Reading it again, it might indeed appear a bit confusing.

[mikejmoffitt]

Glitcho has no value programmed at all, so him giving 10 points normally is just a matter of happenstance. If the section wasn't clear about that, then that's my miss. We could double check the rockets but something that small and plentiful is more likely to intentionally be 10 pts.

Unless there is a second hasty mistake like Glitcho's code, the chances of us applying this principle to other enemies is basically zero.

As for whether this could happen in something other than assembly... yes, absolutely. It probably wouldn't be a stale register value, but the principle of stale data in variables transcends the architecture and languages in question.

[Bananamatic]

Isn't it basically the same as the missingno glitch in pokemon RBY where the game saves bogus stuff where the encounter table memory is for the sake of a tutorial then due to an oversight allows you to get into an encounter while the game never loads another proper encounter table?

Also, what if this wasn't bad emulation but another legit glitch? Is "bad emulation" to the extent of just the max bonus screwing up without anything else breaking even likely?

[casualcoder]

It's just like 1-frame links in fighting games. Practice it enough and what at first seems impossible ends up a high-percentage play.

It's way harder since links are dependent on your own sequence of inputs and are isolated button presses that come down to muscle memory compared to trying to hit something frame perfect according to automatic screen scrolling...

I see what you're saying and you might be right in principle that it's more difficult than a 1-frame link in fighting games. But I'm not sure it's that far off at all.

When hitting 1-frame links I don't only use the relative timing of my button presses as queue. I also use visuals and sound. I find that my mind "freeze-frames" a particular frame and that acts as my main queue to get the link down. The internal timing is more the window, while the visual frame queue the specific moment I close the window, so to speak.

So that would translate pretty well to a consistent scrolling game. Someone good at this could find the right moment and freeze-frame it in their mind and you could even probably use other visual queues to ensure the ship is in perfect setup. Add that together with a 4-frame window and it doesn't seem too difficult in principle... once practiced to death!

[Bananamatic]

So that would translate pretty well to a consistent scrolling game. Someone good at this could find the right moment and freeze-frame it in their mind and you could even probably use other visual queues to ensure the ship is in perfect setup. Add that together with a 4-frame window and it doesn't seem too difficult in principle... once practiced to death!

In reality the 2-5 full chain (not 1-5) gives you way more than 4 frames of leniency yet I've seen basically no one pull it off

[Sumez]

This is an anally strict glitch in a game which has an anally strict chaining system. I don't imagine the type of person who is already invested into playing Dodonpachi for score will be terribly bothered with the execution of said glitch. If he were, honestly I'd say the reaction would be strangely hypocritical.

We're talking about a timing element that's so narrow that whether you get it is almost random.
And getting it the maximum of four times in one run will rocket your score so far out that any scores that don't are pretty much invalid.

I see what you're saying and you might be right in principle that it's more difficult than a 1-frame link in fighting games. But I'm not sure it's that far off at all.

Outside of one example being rhythm-based timing, and the other being reactions to the game, I think an important difference here that missing one combo link in a fighting game (how common are 1 frame links anyway?) doesn't ruin your entire game. If this glitch is allowed in "official" DDP scores, you're pretty much required to get them.