Giga Wing Dreamcast Input Latency

[phase3]

After spending some time playing Giga Wing on authentic CPS-2 hardware, I've found playing it on the Dreamcast a bit frustrating due to some very apparent input latency. The difference is far from negligible. After spending an hour playing GW on the DC, I find myself having to stop compensating for some expected delay in other games.

Why isn't this a properly documented issue? I'm struggling to find any discussion of this anywhere, shmups archives included. Surely I'm not the only person to notice this. I don't think I'm imagining it. It's almost as bad as attempting to play something on a commercial LCD display.

I'm making this thread so there is some proper documentation. If I had known about this latency, I wouldn't have bought the Dreamcast port.

[davyK]

Interesting. I'm playing DC Gigawing at the moment. I have never played the original.

Here's a review of the game that mentions lag. You will see a few comments I make in the video's thread about this. It's the only review I've heard refer to lag.
https://www.youtube.com/watch?v=Mu-E-VwWWq8

I do notice some lag with firing but it varies between ships. There is a bit of lag with starting to shoot but since you are pretty much tapping that button throughout it isn't an issue. There is lag on the default character with shooting for example. Shooting continues for a brief period when you stop pressing fire so I always put it down to the design of each ship's weapon system. I play it via RGB SCART on a CRT as it looks rough on a modern screen even with VGA.

I don't feel any lag with ship movement or with bomb use. But I'm not what you would call highly skilled so maybe wouldn't notice! Is the force shield delay longer with DC?

[Obiwanshinobi]

I recall the arcade one lagging prominently on MAME (3-4 frames, about as bad as Battle Garegga there). Was the emulation accurate in that respect?

[Shepardus]

OP's talking about the Dreamcast port lagging, not the PCB. But yes, there's 3 frames of input lag in MAME.

I do notice some lag with firing but it varies between ships. There is a bit of lag with starting to shoot but since you are pretty much tapping that button throughout it isn't an issue. There is lag on the default character with shooting for example. Shooting continues for a brief period when you stop pressing fire so I always put it down to the design of each ship's weapon system.

In the arcade version you need to tap fire to shoot (holding the button activates the force shield) so each tap lets out a burst of fire. What you're describing is probably the plane finishing a burst.

[davyK]

OP's talking about the Dreamcast port lagging, not the PCB. But yes, there's 3 frames of input lag in MAME.
....
In the arcade version you need to tap fire to shoot (holding the button activates the force shield) so each tap lets out a burst of fire. What you're describing is probably the plane finishing a burst.

Yeah - the DC version is the same. Tap for fire , hold down for reflect. I wonder if the reflect charge time is different on DC? Agree re the plane finishing a burst.

Think I remember reading the DC version has less slowdown too.

[LordHypnos]

Not sure if it's what you're referring to, but the DC version has a significant amount of lag specifically on the reflect barrier (That shield you hold down the a button to activate. Way faster in MAME. Additionally, there are some parts with slower or faster patterns in the DC port (possibly due to slowdown, I don't really know). It's not a great port. Much more difficult than the arcade original IMO, not that I've spent that much time with either. Really quite a shame.

[tomwhite2004]

Mars Matrix also suffers from really bad lag on its Dreamcast port, wonder if it was Takumi or Capcom themselves that handled the conversion?

[davyK]

If the lag difference is noticeable with the reflect force activate time then it's possible it was deliberate.

I do lose lives because of not being able to activate reflect in time - it seems that you have to anticipate and therefore learn the level to be effective but I just assumed that was the way the game was designed. Otherwise why not just have a reflect activate via a 3rd button and have it respond instantly provided it is charged?

[LordHypnos]

Mars Matrix also suffers from really bad lag on its Dreamcast port, wonder if it was Takumi or Capcom themselves that handled the conversion?

I have heard this before, but I have put hundreds of hours into Mars matrix on the DC and am at least somewhat sensitive to lag, but the only lag I've ever noticed was only on activating the mosquito and only when using autofire. This is almost certainly intentional, and easily avoided by not using auto. I even still use auto in a lot of places, and it's pretty easy to adapt to. I guess there was also a small amount of lag when playing with a controller adapter, but that can't be blamed on the port.

Also: i have noticed mame has really bad lag for Mars matrix if you're not using shmupmame

I would not be surprised if the problem with giga wing was actually caused by the longer shot duration, which I think I remember noticing relative to mame. I can't comment strongly on giga wing though, because I haven't played it nearly as much.

[phase3]

Although there may be a more pronounced delay when firing or when using reflect, I find the most damning delay comes from moving the craft. To those who wish to test this issue themselves, please note the following observations:
-There is less apparent delay in response time to initial input, I find the main problem is the game's failure to register when I've stopped moving. It's harder to tell when tap dodging, but when making more sweeping full screen movements, I notice the ship doesn't really want to stop where I want it to, occasionally resulting in a lost life.
-I find that my ship is slightly behind where I perceive it should be, resulting in an effect I'm going to call "coasting" or "ghosting." For example, when weaving into and out of a particularly dense pattern, I notice my ship isn't in sync with my input. It's hard to describe, but it's a noticeable sensation that usually results in loss of confidence that I will be able to accurately dodge something, even if I have the ability.
-When going from forward or backward into a diagonal then back again (↑,↙,↑) it feels as though the ship takes the longest path possible (something like ↑,←,↓,↑) to achieve my desired command. Needless to say, it can feel quite jarring.

I have tested my copy of the game on a couple different DC consoles, video cables, and with multiple different controllers, all of which have the same problem. I am using a CRT with S-video. I have never noticed the issue in Mars Matrix, a game I have spent considerably more time playing.

I have a few theories on the perceived issue:
-It could totally be in my head.
-Certain copies of the game are better about lag than others.
-Certain consoles are more capable of handling the game than others. (I'm not sure if that's possible, but one of you might be.)

EDIT: As I spend more and more time playing the DC port, I find my scores nearing what I was able to achieve on the CPS-2. Perhaps spending time with it may acclimate you to the ports idiosyncratic lagginess, but it may also be that the difference in latency between versions is less severe than I had originally perceived.
Something I find interesting is that the ships I am good with on the CPS-2 board I find more difficult to use on the DC, and vice versa.
A few things that may contribute to giving the impression of some input latency (if it indeed it doesn't exist, which I'm still unsure of) are the fact that your ship sprite likes to dance around a lot when moving, making it appear that the ship is moving when it isn't intended; also, the bullet stream ascending from your ship doesn't exactly flow 1:1 to the top of the screen the way you may expect it to, which plays tricks on your eyes. It's all quite odd.
Nonetheless, I find the "lag" isn't entirely game breaking, and Giga Wing is brilliant fun no matter how you play it.

[Jeneki]

I have heard this before, but I have put hundreds of hours into Mars matrix on the DC and am at least somewhat sensitive to lag, but the only lag I've ever noticed was only on activating the mosquito and only when using autofire. This is almost certainly intentional, and easily avoided by not using auto. I even still use auto in a lot of places, and it's pretty easy to adapt to. I guess there was also a small amount of lag when playing with a controller adapter, but that can't be blamed on the port.

Auto-pierce will sometimes delay the shield, yeah I've noticed it often. One of the Shmupmame versions added an auto-pierce button as well, and it does the same thing there (in addition to screwing up the pierce timing on stage 3 boss). I got used to 1 button only, after I got the CPS2 and had to play that way.

[LordHypnos]

snipped

I wouldn't be too surprised, actually. Maybe I'll try to test that tomorrow at some point. Been a while since I played. I just wanted to mention the shield lag because it's something that is reasonably well documented and easily noticeable, and I think might be due to a longer animation of the shot

Auto-pierce will sometimes delay the shield, yeah I've noticed it often. One of the Shmupmame versions added an auto-pierce button as well, and it does the same thing there (in addition to screwing up the pierce timing on stage 3 boss). I got used to 1 button only, after I got the CPS2 and had to play that way.

Ah, well in that case it's probably not intentional, and just a side affect of auto, Maybe something to do with the animation needing to finish. Nonetheless, is hardly a port issue. The one thing that bothered me with the shmupmame autofire was that I couldn't leave the autofire key depressed while activating the shield. Made it basically unusable for me. The DC doesn't have that problem. Sorry to get off on a tangent.

[Kiken]

DC GigaWing definitely has input lag. It isn't totally game-breaking though (will definitely make some top end scoring exceedingly risky). The major issue is the removal of the slowdown, particularly during stages 6 and 7. For the most part, many of the game's patterns aren't too difficult to dodge at full speed, but the really hectic stuff (like the stage 6 blue radial scatter pattern and the stage 7 screen filling radial pattern) now pretty much requires the use of a bomb, a preemptive Reflect Force or a combination of the two to survive (these patterns were definitely designed with slowdown in mind). It's less than ideal in the grand scheme of things, but I do prefer the AST.

If anything, the DC version is good for practice (it isn't missing anything, near as I can tell... it's just laggy and overclocked).

[davyK]

Bit of a a pity that. I enjoy the game though - as I've never played the original I don't know any better


Just had a thought....I have the PAL version (runs at 50Hz or 60Hz via RGB SCART and also at 60Hz on VGA). I'm selecting 60Hz mode when playing (obviously!) on a CRT. Sometimes the PAL versions get little tweaks - especially on Dreamcast....maybe it's different from the NTSC version? I honestly can't detect any lag - ship movement feels pretty responsive*. Anyone else have the PAL version?




* It could just be me and my lower skill level though and I'm talking out of my backside.

[davyK]

Burned a copy of the US version - didn't seem any different but maybe better players would notice a difference. It also ran on my PAL machine without using my DC-X disc so I'm not sure if it is a genuine US ROM I used.

Anyhow - another question. I've been looking at youtube runs on both arcade and DC versions of this and they have all been done using the player 2 side. Is there any reason for this?

Looking at the arcade runs the reflect activation period does seem to be shorter than in the DC port. Maybe it has been adjusted for different bullet speed?

[LordHypnos]

Burned a copy of the US version - didn't seem any different but maybe better players would notice a difference. It also ran on my PAL machine without using my DC-X disc so I'm not sure if it is a genuine US ROM I used.

Anyhow - another question. I've been looking at youtube runs on both arcade and DC versions of this and they have all been done using the player 2 side. Is there any reason for this?

Looking at the arcade runs the reflect activation period does seem to be shorter than in the DC port. Maybe it has been adjusted for different bullet speed?

self-booting CD-ROMs (including burned ones) are region free, on the dreamcast, so it could very well be a US version.

At one point, I hypothesized that the longer wait for shield might have to do with a longer fire animation (this would have been to make the game require less tapping using manual). I'm not sure why I thought this, but it would make sense, assuming that the animation needs to finish before it can go to shield. I suppose this could be tested.

[davyK]

There are regional differences. The JPN version displays the score at the game end screen differently for example. Going by videos I have seen you can skip the cutscenes faster in that version too. So maybe there are other changes and tweaks.