Dodonpachi Saidaioujou [Cave 2012, ARCADE]

[moozooh]

I believe it's not the sound output tract's fault, but poor mastering, severe downsampling, and generally subpar quality control practices, so THD measurements won't help. Just like if you play a poor recording through a professional-grade equipment, it won't show bad measurements on the equipment's part, yet it will sound awful.

[Muchi Muchi Spork]

I'm trying to get a fix for it right now. My guess is they would rather deny acknowledging that there is a real issue than to risk having to take back tons of boards.

If you used to work in recording and production, do you still have access to measuring equipment? FR graphs and/or THD/noise measurements could make an even more compelling case, especially if you compare the quality over multiple sh3 boards. Admittedly, that's a lot of work with not much chance of reward, but at the very least you can back up your points-- hopefully well enough that somebody will notice.

Even if they won't put out a fix, getting them to acknowledge it would probably make it much less likely for something like this to happen again.

Just my $0.02.

I just have a few bits I kept for writing songs by/for myself. For the recording I just use the top end of programs' safe volume levels that have worked correctly for every other arcade board that I've tried. It doesn't matter how low you turn the volume, the problem is there, so there is nothing to argue about. If you can hear it at all, the static etc. is there. Record it as low as possible and then turn up your computer speaker or normalize, whatever, it's full of static always. Cave wouldn't understand any arguments based on English high tech anyway. We'll be lucky if they translate simple messages correctly.

I believe it's not the sound output tract's fault, but poor mastering, severe downsampling, and generally subpar quality control practices, so THD measurements won't help. Just like if you play a poor recording through a professional-grade equipment, it won't show bad measurements on the equipment's part, yet it will sound awful.

Yep. Most heavily on the severe downsampling. At least that's how it sounds. It sounds like you just downsampled it beyond the limits that a program should offer it, to the point where it is just "destroyed".

[Softdrink 117]

I just have a few bits I kept for writing songs by/for myself. For the recording I just use the top end of programs' safe volume levels that have worked correctly for every other arcade board that I've tried. It doesn't matter how low you turn the volume, the problem is there, so there is nothing to argue about. If you can hear it at all, the static etc. is there. Record it as low as possible and then turn up your computer speaker or normalize, whatever, it's full of static always. Cave wouldn't understand any arguments based on English high tech anyway. We'll be lucky if they translate simple messages correctly.

So basically you just approximated a line level signal for the recording, and based it off of that?
I agree that there's nothing to argue about. :s
Lol.

I believe it's not the sound output tract's fault, but poor mastering, severe downsampling, and generally subpar quality control practices, so THD measurements won't help. Just like if you play a poor recording through a professional-grade equipment, it won't show bad measurements on the equipment's part, yet it will sound awful.

DUH!
I feel silly now, especially since I know all of this firsthand (I'm a pretty serious audiophile). Admittedly I'm running on very little sleep right now (finals week, engineering courses...), but still. Ugh.

For the sake of argument I'm going to point out that it's not impossible that there is some distortion caused by the equipment, but you're quite right. It makes you wonder what they were thinking when they did the mastering, and what they were using for reference. I can't imagine cabinet speakers are poor enough that something like this would go totally unnoticed.

[Muchi Muchi Spork]

I've recorded over 100 pcbs and they all sound fine but this one. It's not rocket science to get a level that doesn't cause static due to the level. Recording programs have continual monitors that tell you when you are safe, in a risky area or definitely too high. I've recorded it at multiple levels including very low levels and the quality problem is always there and it's there in every video I've seen on the web. So you can keep being an audiodouche all you want, I'm done here.

[moozooh]

It makes you wonder what they were thinking when they did the mastering, and what they were using for reference. I can't imagine cabinet speakers are poor enough that something like this would go totally unnoticed.

By this point I'm reasonably sure this is by no means a professional work. They receive proper high quality masters from the musicians, as reflected by the CD releases later on, and what most likely happens next is they give it to some techie who has to fit it in the allotted ROM space and maintain a certain volume level. So what the techie does is reducing the parameters until all tracks fit, without bothering with filling up every last bit, and applying gain until everything sounds loud enough to be heard through cab speakers, without bothering with distortion and dynamics. (Although obviously dynamic range is a concept entirely counterproductive to a general arcade environment.)

Apparently not every aspect of Japanese industry enjoys utmost care (as if the repeated corner-cutting in their nuclear industry hasn't shown that already).

[Softdrink 117]

So you can keep being an audiodouche all you want, I'm done here.

I think there's been a serious misunderstanding here. I'm not trying to in any way invalidate what you've done; I'm asking questions because I'm curious. I've never used any kind of recording software. I have no way to know what steps you took, or how many pcb measurements you've taken. Please understand that I'm ignorant-- I'm only trying to contribute. It was wrong of me to make the assumptions that I did.

I'm deeply invested in music as a hobby, so I'd like to think I know a little bit about it, but that's all. I have great respect for those who work in the field professionally.

My sincerest apologies for any offense.

If you'd like to continue this conversation, I'd be happy to do so by PM.

Back on topic:
Moozooh, that's a great explanation. Given the limited memory, and the fact that any quality issues probably wouldn't even be noticed by the vast majority of the audience, it's a sensible compromise (albeit one with unfortunate consequences).

[THE]

According to the documentation of the cv1000b sound chip it supports the following sample rates:

48kHz/24kHz
44.1kHz/22.05kHz   
32kHz/16kHz

I assume that Cave started with 32khz in the first games and gradually did go down to 16khz :-(

[EOJ]

16khz is close to the sound quality of a telephone.

[brentsg]

The only reason that the vast majority of the audience will not notice is that most will be in loud game centers.  The sound is poor enough that it will be obvious to anyone playing in a quiet environment.

[cstarflare]

Has there been any more said about the yin/yang thing? Last I heard was that it might be tied to Autobomb on/off; has this been confirmed and has anyone been able to tell what the differences are?

[MikeNeko]

Well, ym770 handles variable bitrates per block during stream playback... but still, i could make back in mid 80s perfectly acceptable sounding 8khz or 11khz 8bits samples using the existing -and poor- pc-hardware.So it s not a cellphone-style/am radio quality issue from samplingrate.

But... well... ADPCM is a ..err...a... difficult mistress, you must filter high frequencies in a smart way, carefully, on your waveform otherwise the conversion to lower resolution of the samples destroys the efforts made by the sfx artist (NGDevteam must have had some headaches with that xD).

Anyway, i would be interested in hearing if the second batch had been touched up or if they didn t care and just asked for the same roms.

[THE]

Well, ym770 handles variable bitrates per block during stream playback... but still, i could make back in mid 80s perfectly acceptable sounding 8khz or 11khz 8bits samples using the existing -and poor- pc-hardware.So it s not a cellphone-style/am radio quality issue from samplingrate.

Interesting, I didn't know it can do dynamic changes in sampling rate. I just skimmed fast through it to decide if we use it for our hardware, but dismissed it quickly as being not suited for my quality needs. I just designed my own custom SPU it's more fun this way anyways

But... well... ADPCM is a ..err...a... difficult mistress, you must filter high frequencies in a smart way, carefully, on your waveform otherwise the conversion to lower resolution of the samples destroys the efforts made by the sfx artist (NGDevteam must have had some headaches with that xD).

Yeah, ADPCM is horrible. I alway procrastinate it to last minute to put new sounds/music in it.
Still, a professional hired/talented sound engineer shouldn't have much problems with it, but Cave probably just has an intern for this or just doesn't care.

[emphatic]

16khz is close to the sound quality of a telephone.

It's those damn phone ports' fault!

[Van_Artic]

well, looks like some player wanted more of a counterstop

http://www.youtube.com/watch?v=f2LpC0jfHgg
http://www.youtube.com/watch?v=WE3lCR6PIZY

score goes above 999bil, doesn't counterstop

i'm starting to think this is intentional as EOJ stated before!

[Erppo]

I like how they were obviously going to make it damn sure that nobody is going to counterstop the game this time and left enough space for scores waaaay beyond what looks reachable. But no, you counterstop the chain value instead.

i'm starting to think this is intentional as EOJ stated before!

Making the system this stupid purposefully sounds lot worse than it being a bug.

[Naut]

That is retarded. I would've rathered it counterstop so they'd be pressured more to make a new version.

[EOJ]

Props to Cave for anticipating scores in the trillions - they've learned from their mistakes. Bug or intentional, either way it's very much part of the game and I think it's pretty cool after watching the stage 5 video. If you can pull that stuff off, the reward is huge and it must be really awesome to accomplish that in a run. In any case, like in any other Cave game, scores for all of the stages will have be perfected for that coveted world record spot in Arcadia.

[moozooh]

Scores from all other stages combined make up whopping 1-3% of the potential score you can get from a single bugged chain, and there's most likely a place for two of those. EOJ, if you really think this is cool, I don't believe you're qualified to complain about DFK 1.51 or DFK BL Arrange's scoring anymore?they're nowhere near as broken. :P

[EOJ]

This looks fun, like riding a rollercoaster that suddenly takes off for outer space near the end of the ride. Dfk 1.51 is not fun. Dfk bl arrange is pretty fun. That's my opinion.

[moozooh]

Fair enough.

[moozooh]

Yet another superplay-grade demonstration with A-L:

http://www.nicozon.net/watch/sm18182260 (stages 1-2)
Gets 227 million on stage 1 and ends it with eight hypers in stock. I called the possibility, but didn't think it would allow so much room for imperfection. Actually, now I'm wondering if having two hypers at the time of activation and more aggressive aura-ing of the mid-boss would be beneficial in any way. Nine hypers still seems out of the question. Stage 2 finished with 1.44 bil and eight hypers in stock.

http://www.nicozon.net/watch/sm18182630 (stages 3-4 from a different session)
Starts stage 3 with 1.40 bil, ends with 4.81 bil and seven (pretty much eight) hypers in stock. Uses hypers twice in the stage, at levels 10 and 1, both times not attempting even a minimal recharge at the closest opportunities. Thus, most likely improvable. Ends at the end of stage 4 upon taking an unfortunate hit with over 10k chain and rank 20; the final score was 6.3 billion with a clear potential to reach 8.5+ billion after the tally on a no-miss scenario.

That paints the following picture for the reference A-L scoring with perfect survival, assuming current knowledge:
≈1.45 bil (stages 1 & 2) +
≈3.5 bil (optimal stage 3) +
≈3.7 bil (optimal and complete stage 4) +
≈11 bil (stage portion of stage 5 + Hachi, as per the earlier vids, including the no-miss and full chain bonuses) +
1 bil for killing Hibachi's first form +
5 bil for killing Hibachi's second form +
6 bil for remaining lives +
0.3 bil for remaining bombs
= 32 billion.

Now I don't expect anybody to actually max out the score, but we may well see Arcadia scores roll over 30 billion not before long?in fact, I strongly suspect the current top scores to be around 25 billion already, at the very least. If these replays, at least stages 3 through 5 are inoptimal, then it's likely there's some 2-3 billion more to take advantage of.

At the same time I don't really see how anybody would no-miss/no-bomb this game under high-scoring conditions, not with this TLB and the rank system anyway. The best stage 5 replay seen so far starts it at rank 5, these vids suggest it would start at rank 20+ under optimal conditions, and the TLB would be reached at around 40.

[TimingTripod40]

I think it will be like Futari God where people will figure out near-optimal strats relatively quickly but the TLB and clear bonus will stop them from actually maxing the game out for a while.

So I was pretty much spot on here?

Though BL Maniac is probably a better comparison since the first God mode scores were only in the neighborhood of around 4.5 billion while this looks like it's getting pushed pretty close to its limit already, aside from the whole TLB thing. Not that I'm complaining as it still looks like a pretty fun game.

[moozooh]

To be fair, I don't think any of the Pachi games so far have been maxed out per se, except maybe some DFK 1.5 ship/style combinations. You may remember the scope of some of the recent improvements. I'd say at this stage it might be similar to the 640 million DDP score NAI made early on, which was reasonably close to the top?yet the current record still has a lot of room nobody is willing (or able) to master.

[Kaneda]

Now I don't expect anybody to actually max out the score, but we may well see Arcadia scores roll over 30 billion not before long?in fact, I strongly suspect the current top scores to be around 25 billion already, at the very least.

I wish I could tell you the top score at Game in Ebisen, but I was sworn to secrecy. It's pretty goddamn high (and this was a few weeks ago).

New Arcadia in a few days, I'm curious to see if the Expert uber-scores are in there.

[moozooh]

Man, this suspense is killing me. :D

[Erppo]

To be fair, I don't think any of the Pachi games so far have been maxed out per se, except maybe some DFK 1.5 ship/style combinations.

On the other hand, all the other Pachis are 2-loop games which probably makes it considerably more difficult to do everything perfectly. I'm suspecting SDOJ scores to get really close to the max sooner for that reason.

[moozooh]

Right. There's also the fact that players have messageboards, wikis and video hosting sites they didn't quite have in pre-SH3 era. But then again, this time the players have micro-optimization of gauge filling/recharging to master, and then there's the devious combination of very expensive life stock at the endgame and one of the fiercest TLBs to date. I wonder if that'll even things out a bit?

[EOJ]

The new arcadia isn't out until tomorrow, but a guy on 2Ch claims to have it and posted approximations of the scores (he didn't want to give the full scores as it's not out yet):

A-S 27.8bil (Hibachi ALL)
A-L No scores submitted
A-E 15.9 bil ALL

B-S 18.3bil (Hibachi ALL)
B-L 24.1bil (Hibachi ALL)
B-E 273.8bil ALL

C-S 22.8bil (Hibachi ALL)
C-L 21.7bil (Hibachi ALL)
C-E 6.8bil (Stage 5-Hibachi (NO CLEAR))

These are the best scores as of about 5/20/2012 (the cutoff date for Arcadia submissions for the current issue).

[EOJ]

The above scores are accurate, I just checked the August issue of Arcadia. I didn't buy it, however, and I didn't bother to write down the remaining digits of those scores as they'll be beaten in a month's time. The usual suspects got the scores, i.e. Clover-TAC got one, and the low C-E score was by SWY-Yusemi. I guess C-E is the hardest to score with?

Also interesting is Arcadia talks about the 273bil B-E score and they say "this score used a special technique to get a really high GP bonus". They don't call it a glitch or bug, for what it's worth. They also never called the PS infinite lives thing a glitch/bug.

There's also an interview with IKD done in June 2012 about SDOJ. I skimmed through it, but not much of interest popped out. He seems happy with the way the game turned out and I got the impression there won't be any further Arcade revisions. From what I recall, he said that he thinks it will take many years for people to master the game.

[SuperPang]

The new batch have the same 4/20 version so they obviously decided against fixing the bug to avoid confusion. It's now officially a "special technique"