A comparison of CV-1000 (SH3) PCBs with their respective ports

[EOJ]

In this thread I attempt a more detailed and accurate assessment of each port of a CV-1000 (SH3) CAVE game. I only attempt assessments for games that I own (or have owned) both the PCB and port of. Furthermore, I need to have spent a good amount of time comparing the two.

First of all, a disclaimer should be made: NONE of the ports "feel" the same as the PCBs. This is because the slowdown on the ports, while for the most part is well-emulated, does not slow down exactly the same in regard to speed and syncing with inputs. The PCB slowdown feels smoother and more organic in transitions in/out and overall speed, and this is partly because the inputs are tightly in-sync with the slowdown, compared to the ports where they are not. This creates a distinct difference in the "feel" of the respective versions of each game. For games that rely on nearly continuous instances of split second timing for scoring (such as Dodonpachi Daifukkatsu and Mushihimesama Futari God Mode), the different feel of the PCBs becomes more apparent. Whether one prefers the slowdown "feel" on the ports or PCBs is completely subjective. However, after one becomes accustomed to the "feel" of the ports similar high scores can be achieved in most (but not all) cases. Another important thing to keep in mind: nearly every port has slowdown missing in some spots and extra slowdown in others. Unfortunately, some ports are horrendous in this regard. This makes some ports easier to score on, others harder, and others about the same (but still fundamentally different from the PCB).

All CV-1000 (SH3) PCBs have 2 frames of input lag. The best the X360 can output without screen tearing is 3 frames, but some ports have worse input lag than that. I sourced all of the input lag results from this page:
http://electricunderground.io/shmup-input-lag-database/

Although the main purpose here is to compare the PCBs with the ports, I also discuss and grade the extra features provided in each port.

[EOJ]

Akai Katana
SLOWDOWN ACCURACY: F (Overall about 40-50% accuracy)
Here it is, the most inaccurate SH3 port on the X360 in regard to slowdown emulation. It is so bad that the same high scoring run with the same patterns lasts about 39 minutes on the PCB, but only 29 minutes on the port. That's a whopping 10 minutes of missing slowdown!! The first three stages are not too bad, but, inexplicably there is practically no slowdown in the last two stages in the port (which is particularly odd because these stages have the most slowdown on the PCB). Overall, where it does slowdown, most of the time, it does not slowdown as much or for as long of a duration as on the PCB. On the bosses some patterns have slowdown and others don't, seemingly at random. One of the worst parts is the Stage 5 boss and the final boss. They have very little slowdown on the port, but tons on the PCB, and considering you really need to no miss and no bomb the game for an optimal score, it is MUCH more difficult to achieve that on the port. The superplay DVD by phan demonstrates this can be done, but his best score on the port with Type 3 (525mil) is less than the WR with Type 3 on the PCB by AKF (540mil), and AKF stated he could theoretically do 550+mil on a perfect run -- I highly doubt that could be achieved on the port. The best scores for Types 1 and 2 recorded on the port are even further away from the WRs on the PCB.
PORT ACCURACY: C
The port has 4-5 frames of input lag in comparison to the PCB's 2 frames. I am not aware of any inaccuracies other than the lack of an option to choose the original low res graphics. The entire game (not just the sprites) was redrawn in high res, creating a sharp and pleasant image, but noticeably different to the PCB. We are not allowed to enter our initials in the high score board.
VISUAL OPTIONS: D.
We are not given any visual options except an option for smoothing on/off.
CONTROL OPTIONS: A.
Single inputs for each shot type and bomb are provided. Nothing else is needed for this game, however.
EXTRAS: C.
A decent selection of wallpapers is provided, and that's about it.

Deathsmiles Mega Black Label (Japanese)
SLOWDOWN ACCURACY: B (Overall about 80-85% accuracy overall for both modes)
One of the best Cave ports out there in this regard, but Arcade and 360 modes differ in their slowdown accuracy. Arcade mode is nearly perfect in the first six stages, including all the important slowdown on the LV999 bosses, though some of the slowdown does feel a little slower on the port compared to the PCB. The canyon stage is also very accurate. However, the Ice Palace is missing a noticeable amount of slowdown, perhaps 30% or so. With some characters this is worse than with others due to the laser-induced slowdown (e.g. Windia's laser creates more slowdown than Sakura's). In the castle stage, there is actually more slowdown in the first third of the stage, making this easier than on the PCB. And thankfully, the important slowdown is there on the Castle boss Jitterbug. When we turn to 360 mode, the slowdown is very accurate outside of the LV999 stages. In the LV999 stages only the first third of the Lake Front stage is missing some noticeable bits outside of Power Up mode. When we turn to the Lv999 bosses, there are two that are missing important slowdown. First, the Swamp boss Sakura: in her first phase there is no slowdown on the large blue discs she shoots out, making it very, very difficult to avoid these compared to the PCB. For Windia, this is a big deal. If you get a half-life bar hit here, you cannot use the half lifebar hit on the discs in the last phase without dying, forcing you to use an extra bomb here. This leads into the Forest boss at Lv999, which is missing all of the slowdown in the first phase (it is also missing all the slowdown on the red apples in Lvs1-3, though there is any easy way to avoid this bullet pattern), which, again, makes it very difficult to get through with just one bomb. And if you die here, you get a score drop that makes it impossible to properly execute the optimal scoring run for Windia. Other than this, 360 mode is very accurate, and the Ice Palace is so close to the PCB I can barely distinguish the two in regard to slowdown. The only other part of X360 that is missing noticeable slowdown is Jitterbug (though it is there if you play with Follett, perhaps due to her laser). The first phase is missing all of the slowdown, and his last phase runs too fast, making it just a little harder to beat him without bombing (this last point is not a big deal, though).
PORT ACCURACY: A (Arcade Mode) / A- (X360 mode).
The port has 3 frames of input lag in comparison to the PCB's 2 frames. I am not aware of any inaccuracies in Arcade mode other than the lack of true 240p graphic output, which is common to all X360 ports. When we turn to X360 mode, only two inaccuracies pop up, neither of which is significant. First, the Ice Palace is not shown on the stage map. And second, there is no limit on how many times you can choose a LV1 or 2 stage. Thankfully both modes allow us to input our initials in the high score board, true to the PCB.
VISUAL OPTIONS: D.
We are not given any visual options, not even an option for smoothing on/off (which would help in Arcade mode). No option for scanlines, etc. Screen sizing is handled well, however, and it is easy to set up the screen to display without any vsync tearing in the background, providing a clean, crisp image.
CONTROL OPTIONS: A.
Single inputs for each shot type and bomb are provided, along with extra buttons for Left and Right rapid fire, as well as a separate button for Power Up mode, should you choose. Excellent!
EXTRAS: C.
A decent selection of wallpapers is provided, and that's about it.

Dodonpachi Daifukkatsu Black Label
SLOWDOWN ACCURACY: A- for Power/Bomb style (Overall about 90% accuracy), C for Strong style (Overall about 70-75% accuracy).
This port is unusual in that Power/Bomb style are very accurate, on the same level as the DFK 1.5 port. However, Strong style is quite inaccurate in several important spots. First, for Power/Bomb, although as always there are small bits missing here and there and some of the slowdown speed is too slow or too fast in spots (it is notably much slower in parts of ST1 and on the last midboss in ST5), it is hard to find any major inaccuracies outside of one of the patterns in the second phase of the ST5 boss in which the slowdown at max rank ends about one second earlier (and the transition-out is faster) than on the PCB, making it trickier to milk the boss without using a 'safety' hyper here. For Strong style, there is slowdown missing on the last phase of the ST1 boss (so, the extended milking pattern done by top players on the PCB is much harder to do), extra slowdown during the first big enemy at the start of ST2 (making it a lot easier on the port), extra slowdown in phase one of the ST2 boss (making it easier on the port), slowdown running much faster on the last phase of the ST2 boss (making milking this pattern a bit harder), and slowdown missing on phase one of the ST3 boss (making milking this pattern harder). The slowdown in hyper mode is overall faster in Strong style as well.
PORT ACCURACY: C.
The port has 3 frames of input lag in comparison to the PCB's 2 frames. The major inaccuracy here is the complete removal of the sprite overload glitch that causes the on screen display, and one's own ship, to disappear when too many stars and bullets are on the screen. This is a minor issue in Bomb/Power, where it rarely occurs on the PCB, but in Strong style it occurs very often in hyper mode, often lasting for 4-5 seconds at a time. Removing this makes Strong style significantly easier on the X360 compared to the PCB, to a level where it is possible to score much higher on the port than the PCB. I am not aware of any other inaccuracies except there is no option to play with the original graphics, rather we can only play with hi res sprites over low res backgrounds. Thankfully both modes allow us to input our initials in the high score board, true to the PCB.
VISUAL OPTIONS: C.
We are not given any visual options except smoothing on/off. No option for scanlines, etc. Screen sizing and orientation are handled well, however, and it is easy to set up the screen to display without any vsync tearing in the background, providing a clean, crisp image in horizontal or vertical orientations.
CONTROL OPTIONS: A-.
We are able to map the standard four buttons however we like. No other options are present, but an option to map A+C (laser + shot) to one button would have been nice.
EXTRAS: B.
An excellent selection of wallpapers is provided, but nothing more.

Dodonpachi Daifukkatsu Ver 1.5
SLOWDOWN ACCURACY: A- (Overall about 90% accuracy).
This, along with DFK BL Power style, is the most accurate X360 port in regard to slowdown. Yes, there are minor bits of slowdown missing here and there, but the major slowdown is nearly all there. However, some important slowdown is missing in ST2-5, making a few tricky spots a bit harder on the port. On the flipside, the "tateana" (vertical scrolling section after the first midboss) in ST2-5 has more slowdown on the port, making this part much easier than the PCB. The slowdown speed (and in/out transitions) is also very close in feel to the PCB, which is something lacking in many other ports. 
PORT ACCURACY: A.
The port has 3 frames of input lag in comparison to the PCB's 2 frames. I am not aware of any inaccuracies except there is no option to play with the original graphics, rather we can only play with hi res sprites over low res backgrounds. Thankfully both modes allow us to input our initials in the high score board, true to the PCB.
VISUAL OPTIONS: C.
We are not given any visual options except smoothing on/off. No option for scanlines, etc. Screen sizing and orientation are handled well, however, and it is easy to set up the screen to display without any vsync tearing in the background, providing a clean, crisp image in horizontal or vertical orientations.
CONTROL OPTIONS: A.
We are able to map the standard four buttons however we like. No other options are present, but none are needed for this game.
EXTRAS: B.
An excellent selection of wallpapers is provided, but nothing more.

Dodonpachi Saidaioujou
SLOWDOWN ACCURACY: D in 720 ~1080p (60-65% accuracy), B in 480p (80-85% accuracy)
This port is very accurate on the bosses. It is difficult to find any differences here except the last bullet pattern in the second phase of the Stage 5 boss, which has a bit more (and slower) slowdown on the port. However, when played in 720-1080p resolution, the stages are a huge problem. Unlike many other ports the problem isn't a lack of slowdown, but rather additional slowdown. Huge, noticeable swarms of slowdown. These become very apparent beginning in Stage 3, and they increase in duration and frequency throughout Stage 4 and also throughout Stage 5. The end result is scoring, and doing the overflow, is much easier on the port (however, survival on the bosses is harder, due to the increased input lag mentioned below). Thankfully, if one plays in 480p, the slowdown accuracy improves significantly, with most of the added slowdown no longer there (several spots in stage 5 still have more slowdown compared to the PCB, though). When playing on a vertical 16:9 monitor, the best settings I have found are to set the X360 to 480p and Widescreen, then in the SDOJ screen menu turn wide mode "off", turn off all extra windows, and scale the screen to 100 x 75 (or thereabouts, depending on your screen).
PORT ACCURACY: C
The port has 4-5 frames of input lag in comparison to the PCB's 2 frames. I am not aware of any inaccuracies other than the lack of an option to choose the original low res graphics. The entire game (not just the sprites) was redrawn in high res, creating a sharp and pleasant image, but noticeably different to the PCB. We are not allowed to enter our initials in the high score board. The overflow bug is not default, but thankfully it is selectable in the options, allowing users to practice PCB scoring routes.
VISUAL OPTIONS: D.
We are not given any visual options except an option for smoothing on/off.
CONTROL OPTIONS: A.
Single inputs for each shot type and bomb are provided. Nothing else is needed for this game, however.
EXTRAS: B.
A decent selection of wallpapers is provided via unlockables, several interesting custom options are also unlockable, and there is an excellent challenge mode that actually helps you learn difficult parts of the game.

Espgaluda II
SLOWDOWN ACCURACY: C (Overall about 70-75% accuracy overall for both modes).
The game has most of the slowdown in zesshikai mode, but it runs slower there than on the PCB, making scoring a bit easier in that regard. Transitions in and out of zesshikai are also different, feeling a bit faster on the port. Outside of zesshikai mode there are several major and minor spots with missing slowdown. Major spots include the Seseri boss fight in ST5, the first large enemy in ST5 (making the high-level scoring pattern here very difficult to do), and the first phase of the ST6 boss. Minor spots include, for example, the ST1 boss, which has some stuttery slowdown in the first phase patterns, which run perfectly smooth on the X360 port (but these patterns are easy to dodge, and thus the PCB's slowdown really isn't needed).
PORT ACCURACY: B.
The port has 3 frames of input lag in comparison to the PCB's 2 frames. In both Arcade and X360 modes, the final bullet pattern of the last boss is slightly different from the original arcade PCB (they patched the game to improve this, but they still didn't get it quite right). The bullet movement in zesshikai mode also seems a bit different to me, but this is a minor point. I am not aware of any other inaccuracies in Arcade mode other than the lack of true 240p graphic output, which is common to all X360 ports. When we turn to X360 mode, only one difference pops up: they removed the tutorial glitch that lets you start ST1 with extra gems (this is still present in Arcade mode). Thankfully both modes allow us to input our initials in the high score board, true to the PCB.
VISUAL OPTIONS: C.
We are not given any visual options except smoothing on/off. No option for scanlines, etc. Screen sizing and orientation are handled well, however, and it is easy to set up the screen to display without any vsync tearing in the background, providing a clean, crisp image in horizontal or vertical orientations.
CONTROL OPTIONS: A.
We are able to map the standard four buttons however we like. No other options are present, but none are needed for this game.
EXTRAS: B.
An excellent selection of wallpapers is provided, but nothing more.

Muchi Muchi Pork
SLOWDOWN ACCURACY: D- (Overall about 60% accuracy)
One of the worst Cave ports out there in this regard. There are several important parts that are missing slowdown, including (but not limited to): the formation of ships in the second half of stage 1 (making scoring here more difficult); the stage 1 boss milking (the port feels like it is on fast-forward here); the second half of stage 3; the stage 3 boss milking; the second half of stage 4 (large amounts are missing here, making it much more difficult); several swarms of ships in stage 5 (making scoring more difficult). In addition, several parts with slowdown do not slow down as much as on the PCB. The end result of all these inaccuracies is a situation in which it is essentially impossible to score as optimally as one can on the PCB.
PORT ACCURACY: C-.
The port has 4-5 frames of input lag in comparison to the PCB's 2 frames. A large enemy formation in the second half of stage 5 appears in the wrong location, requiring the player to alter their scoring pattern here, as compared to the PCB. They needlessly changed the font for the medal value, and it is better on the PCB (IMO). We are not able to enter our initials for the local scoreboard in this port, and saved scores randomly delete (CAVE never patched this bug).
VISUAL OPTIONS: D.
We are only given the option to have smoothing on or off. No option for scanlines, etc. Screen sizing and orientation are poorly implemented making it difficult to figure out and cumbersome to adjust.
CONTROL OPTIONS: A.
Single inputs for shot, lard and bomb are provided. Nothing else is needed for this game.
EXTRAS: D.
A very meager and lackluster selection of wallpapers is provided, and that's about it.

Mushihimesama
SLOWDOWN ACCURACY: B (Overall about 80-85% accuracy)
The important slowdown is mostly there, but there is good deal of slowdown missing in less critical parts. It also does not slow down as much in certain spots (i.e. slowdown speed difference), such as in the first half of stage 3 and the last phase of the stage 4 midboss. Bullet speed in Original mode feels a bit faster in many parts.
PORT ACCURACY: B+.
The port has 3 frames of input lag in comparison to the PCB's 2 frames. There are no known discrepancies in enemy placement or bullet patterns. However, the font for the score as well as for the +value has changed. I really dislike such needless changes, especially for the +value (chain value) font, which looks a lot better on the original PCB in my opinion; it is a larger, rounder font. Furthermore, there is no option to play with the original graphics, rather we can only play with hi res sprites over low res backgrounds. Scores save without issue and you can enter your initials in the local scoreboard, just like on the PCB.
VISUAL OPTIONS: B.
We are only given the option to have smoothing on or off. No option for scanlines, etc. Screen rotation and size adjustment are handled well, however.
CONTROL OPTIONS: A-.
Overall very good, with single adjustable rapid fire buttons for both shot and autoshot. The inability to create multiple rapid fire buttons (e.g. more than one rapid-shot at different speeds) is a disappointment.
EXTRAS: B.
A good selection of wallpapers is included, as are side gadgets for the display of extra scoring info, etc. A gallery of artwork is missing, but would have been nice, as would a music jukebox and scans of the arcade artwork.

Mushihimesama Futari Black Label
SLOWDOWN ACCURACY: C (Overall about 75% accuracy in Original and God modes, 85-90% accuracy in Maniac mode)
Unfortunately this is slightly less accurate than the 1.5 port overall. There is a large amount of extra slowdown added in the second half of stage 3, and in general A-Shot slows down more and more often in the port. In God mode, we find the following differences, in addition to a general difference in slowdown speed: added slowdown right before the stage 1 boss; slowdown missing on the large plant enemies in the first half of stage 3; slowdown missing in the first section of stage 4; additional slowdown added on Larsa with Reco's A-Shot (making her difficult last pattern MUCH easier on the port). There is also a very annoying slowdown "stutter" that occurs randomly, particularly in stage 2. This causes the entire screen to freeze for a split second, and is quite jarring. Original and God are similar in their accuracy, but for some reason Maniac mode is comparatively more accurate, with only a small number of minor differences compared to the PCB.
PORT ACCURACY: A.
The port has 3 frames of input lag by default in comparison to the PCB's 2 frames, but if you set the frame delay to 0V (Tearing) you can get 2 frames of lag on the port (with the distracting side-effect of screen tearing). There are no known discrepancies in enemy placement or bullet patterns. We can choose between original graphics (Arcade mode) and hi res sprites over low res backgrounds (360 mode). Scores save without issue and you can enter your initials in the local scoreboard, just like on the PCB.
VISUAL OPTIONS: A+.
A wonderful array of visual options are present, including scanlines. It is hard to find any fault here. Screen sizing and orientation are handled well,  and it is easy to set up the screen to display without any vsync tearing in the background, providing a clean, crisp image in horizontal or vertical orientations.
CONTROL OPTIONS: B.
Nothing beyond the three standard buttons is offered. The inability to create multiple rapid fire buttons (e.g. more than one rapid-shot at different speeds) is a disappointment.
EXTRAS: B.
A good selection of wallpapers is included, as are side gadgets for the display of extra scoring info, etc. A gallery of artwork is missing, but would have been nice, as would a music jukebox and scans of the arcade artwork.

Mushihimesama Futari ver 1.5
SLOWDOWN ACCURACY: B- (Overall about 80% accuracy)
The important slowdown is mostly there, but much of the slowdown is slower in the port, and there are several sections of added slowdown, especially in the second half of stage 3, parts of stage 4 and parts of stage 5. Overall this makes the port noticeably easier for both score and survival, compared to the PCB. There is also a very annoying slowdown "stutter" that occurs randomly, particularly in stage 2. This causes the entire screen to freeze for a split second, and is quite jarring.
PORT ACCURACY: A-.
The port has 3 frames of input lag by default in comparison to the PCB's 2 frames, but if you set the frame delay to 0V (Tearing) you can get 2 frames of lag on the port (with the distracting side-effect of screen tearing). There are no known discrepancies in enemy placement. The only difference in bullet patterns occurs on the fifth bullet pattern of the first TLB, where the bullets aim at you, whereas on the PCB they don't (this is only a problem with Reco, as Palm can deal enough damage to Larsa to skip this phase entirely) -- thanks to SMC for telling me about this. We can choose between original graphics (Arcade mode) and hi res sprites over low res backgrounds (360 mode). Scores save without issue and you can enter your initials in the local scoreboard, just like on the PCB.
VISUAL OPTIONS: A+.
A wonderful array of visual options are present, including scanlines. It is hard to find any fault here. Screen sizing and orientation are handled well, and it is easy to set up the screen to display without any vsync tearing in the background, providing a clean, crisp image in horizontal or vertical orientations.
CONTROL OPTIONS: B.
Nothing beyond the three standard buttons is offered. The inability to create multiple rapid fire buttons (e.g. more than one rapid-shot at different speeds) is a disappointment. 
EXTRAS: B.
A good selection of wallpapers is included, as are side gadgets for the display of extra scoring info, etc. A gallery of artwork is missing, but would have been nice, as would a music jukebox and scans of the arcade artwork.

Pink Sweets
SLOWDOWN ACCURACY: D (Overall about 60-70% accuracy)
Similar to the MMP port, but just a tad better in this category. There are several important parts that are missing slowdown, including (but not limited to): the first boss, the ST2 midboss, the snakes in the middle of ST4, the ST6 midbosses, and several in-stage sections. In addition, several parts with slowdown do not slow down as much as on the PCB. Naturally, this makes the port harder to clear than the PCB, but if you can clear the PCB without the infinite lives trick you should be able to clear the port as well (similar to MMP and the case of Ibara PS2 vs PCB). Harder mode is a bit worse in this regard, as it runs even faster overall and the slowdown is really needed in this mode due to its extreme difficulty.
PORT ACCURACY: D.
The port has 4-5 frames of input lag in comparison to the PCB's 2 frames. There are two main problems in regard to port accuracy. First, the "press start on the Harder mode high score screen" glitch was removed (this glitch gives the player Harder mode bullet patterns in Normal mode without the Harder mode suicide bullets). The second problem is the infinite lives trick was removed. Both of these aspects of the original PCB version are required to get anywhere near the world record scores, so players on the port cannot compete with PCB scores. However, CAVE did release a short-lived (and now quite rare) arcade revision in May 2006 that removed the infinite lives trick, so the Xbox 360 version can be considered a port of that version, rather than the standard, original April 2006 version. Otherwise the port is faithful in terms of rank and other aspects. The most arcade-accurate mode in this port is Score Attack. We are not able to enter our initials for the local scoreboard in this port, unfortunately.
VISUAL OPTIONS: D.
We are only given the option to have smoothing on or off. No option for scanlines, etc. Screen sizing and orientation are poorly implemented making it difficult to figure out and cumbersome to adjust.
CONTROL OPTIONS: C-.
Single inputs for shot and formation change are included, but we are not given the ability to map one input to more than one button, and there is no option for a separate speed change button. Another major omission is the lack of rapid fire buttons (especially rapid formation change).
EXTRAS: D.
A very meager and lackluster selection of wallpapers is provided, and that's about it.

[EOJ]

Reserved.

[EOJ]

Added Futari 1.5 and BL.

[zak]

Nice thread EOJ!

[EOJ]

Thanks! It's overdue for an update. I need to revise the Futari 1.5 data with some new info, and I think I'll add SDOJ and Akai Katana next.

[EOJ]

Here is a link showing the input lag of the CAVE ports (It seems the tester may have about 1 frame of lag in this setup):

http://electricunderground.io/shmup-input-lag-database/

It shows AK, DSII, SDOJ, Guwange, MMP and PS all have 4-5 frames of lag. It is unsurprising that they are the same in this regard because they all use the same menu system (well, for AK the connection to the other ports is less obvious). Unfortunately it is clear CAVE had two port-code systems in place and one was really good (Deathsmiles, Mushi, ESP2, DDPDFK, etc), and the other was pretty bad. It's baffling that they didn't just use the good one every time.

[EOJ]

Updated with DSMBL! A second, more in depth comparison over the past year has given me a much better impression of the Japanese X360 port. For all but the very top tier players who are going for a WR with certain characters, you'll be able to score just as well on the port as on the PCB (also note the highest score with Sakura was obtained on the 360 port's Arcade mode, even with the missing Ice Palace slowdown). I really enjoy 360 mode the most (even more than the PCB!) -- I think because the clearer, high res bullet sprites make it easier on my eyes to survive Death Mode. If they would just tweak a few spots of slowdown it would be the perfect version of the game IMO.

BTW, although Tyjet's guide is excellent, his account (in section 15.0) of the slowdown differences among the modes in the port and the PCB is almost entirely wrong, at least in regard to the Japanese port (though maybe he is talking about the North American version, which I have little experience with in general and no experience with post-patch).

[EOJ]

Added Akai Katana. As you can see, this is a very poorly done port, and the worst there is (on X360) in regard to slowdown emulation. Deathsmiles II is a not-too-distant second though!

[EOJ]

Added DDP SDOJ! Another poorly done port.

[Plasmo]

Added DDP SDOJ! Another poorly done port.

When you speak of "one additional frame of input lag" for SDOJ, does that mean that the PCB has 3-4 frames of inout lag already? I'm referring to Mark's finding of 4-5 frames of input lag on the SDOJ port (http://electricunderground.io/shmup-input-lag-database/).

Highly interesting topic overall, thanks for doing this!

[EOJ]

The PCB has two frames, like all SH3 PCBs. I was going by this:

http://cave-stg.com/forum/index.php?topic=1663.msg31325#msg31325

I wrote "at least one additional frame of lag", because I have seen varying results, and I have yet to see a good test done on a CRT via RGB output on the X360. Seems to be the same amount of lag on SDOJ, MMP, Pink Sweets, DSII and AK (they all run on the same porting engine).

In that link you posted with the input lag tests, it looks like the guy's setup is adding at least one frame of lag above the actual lag.

[MarkMSX]

Hey there,

I'm Mark of the Electric Underground. It sounds like you are skeptical of my input lag results. I've found that, when people test input lag and get different results, the discussion that follows is always just a delightful experience.

So what I think the best way to go about things is to thoroughly explain my methods, my approach, show my results, and let you compare it to your method of measuring lag.

Reading through your post, I don't see an explanation of how you came to your input lag results, so it'll probably be useful when we compare our findings to each other.

To clarify, I do use native analog video from the Xbox 360. I use the stock CRT VGA cable. I also own the 15 kHz RGB cable you describe, and have found that it is no more responsive than analog VGA. In fact, I would argue that the native 31kHz cable for the Xbox 360 is more responsive than the RGB cable, due to the higher horizontal refresh rate (though I don't think it matters). I am NOT using HDMI.

The arcade stick that I use to test with is the Hori Real Arcade Pro VX-SA, which has been shown by my own tests, as well as on Teyah.net, to be the most responsive input method on the Xbox 360.

http://www.teyah.net/sticklag/results.html

This is the most responsive controller for the Xbox 360 that I have come across or heard of.

On the input lag index, I have linked this page which explains my testing methods in-depth:

http://electricunderground.io/input-lag-test-setup/

I have also added this page, which shows examples of frame by frame pictures of my testing results. I have included pictures of Futari with vsync to 0 and to 1, as well as results from SDOJ, Mushi, and DFK.

http://electricunderground.io/shmup-input-lag-results-examples/

I may not be part of the OG shmup player club, but I do take input lag testing seriously and would welcome a comparison with your results and where you identify my extra frame of lag is coming from exactly. Also, I don't think it is possible for a game on the Xbox 360 to have zero frames of lag, as the lag from the controller and the usb ports alone will at least add a frame, in addition to the game processing the input as well. Futari's ability to turn off vysnc is the coolest setting I have ever seen in a shmup port and it is the reason why the Futari port is able to achieve PCB levels of input lag, but the other Cave 360 ports are 3 frames or higher.

I do agree that the main issue with the SDOJ port is how funky its slowdown acts, in relation to the input lag, but it still does have an extra frame of baseline lag in comparison to DFK.

[EOJ]

Thanks for your detailed reply. I have not conducted any input lag measurements myself. Here are a few questions and comments for you.

1) Ports like Deathsmiles have 3 frames of lag and that is as good as it gets by default on the X360, according to your results. However, Futari X360 in the options is set to the standard PCB input lag by default (2 frames), so shouldn't the default Futari lag on the X360 port be 3 frames as well, and changing the setting to 1 will reduce it to 2 frames (and setting it to 0 will reduce it to 1)? Or are you claiming the Futari port has 4 frames by default?

2) When I changed from 2 to 1 frames in the input lag settings in the Futari port, it felt more responsive to me on my RGB monitor than my PCB, which I played back to back. I still play them back to back through my OSSC and low-lag LCD monitor (5ms total input lag), and the Futari port still feels more responsive than the PCB on this newer set up. (I admit I cannot discount the possibility this could be some placebo effect.)

3) I feel no difference at all between the ports like Deathsmiles, DDP DFK and Espgaluda 2 and their respective PCBs when I play them on the same monitor. However, I do feel a difference between those ports and the SDOJ, MMP, AK and PS ports with extra lag. So, based on your results I would have to conclude I cannot feel a difference between 2 and 3 frames of lag, but I can feel a difference between 3 and 4 (and possibly 1 and 2, in regard to the Futari port/PCB)? That seems a bit strange to me, but sometimes strange things are true.

[MarkMSX]

Hi there, I can clarify your questions:

1) For the most part, it is correct that 3 frames is the amount of input lag most 360 CAVE shmups have (DFK, Deathsmiles, Mushi, etc.) In regards to Futari, it also has 3 frames of input lag by default. In the settings menu the "1" setting is default. However, if you turn off vsync, by changing the setting to "0," then the game runs with 2 frames of input lag. Yes, I am aware that the settings menu claims that the "1" setting is the same input lag as PCB, so there are two possible conclusions. The first is that M2 are not accounting for the additional lag that vsync causes, and are incorrect in their description (which is totally possible, since all of their ports are 1 frame more laggy than PCB -- see my Ketsui Deathtiny and PGM lag test results). Or the second conclusion could be that the Futari PCB itself is 3 frames, not 2. Not owning the Futari PCB, I cannot say what its input lag is like, but I am sure of the tests I have performed on the 360 port and the results of 2 frames with vsync off and 3 frames with it on.

2) I'm not sure what the original PCB is like, in comparison to the 360 port, but having done a ton of input lag testing, it is actually really tricky to decipher a frame of difference between two different setups by feel, I have had trouble with this in the past as well, which is why I like to use an objective testing method.

3) Yeah, on most of the xbox 360 Cave ports, their input lag is really close to PCB and is extremely consistent (which is a good thing), so it's tricky to tell what the input lag is like based on feel. However, with SDOJ and the other laggier ports, not only is their lag high, but their lag is all over the place so it becomes very noticeable by feel.

Hope my responses have helped :-)

[EOJ]

Thanks! That has cleared up a lot of things for me. I mistakenly remembered the Futari port defaulted to 2V (PCB) for the input lag, but you are right, it defaults to 1V (Fast). So 1V is 3 frames, and 0V (Tearing) is 2, like the PCB. I wish I could use 0V, but the screen tearing looks hideous so I could never play like that. It is interesting that 2V (PCB) in the port equates to 4 frames of lag. This seems to indicate the X360 is adding 2 frames somewhere in the output. We could conjecture the "good" Cave ports (Deathsmiles, DDP DFK, Espgaluda 2, Mushihimesama), like the Futari port, default to 1V (1 frame) in their coding, then the X360 adds 2, making the output 3. This seems to be the lowest input lag you can get on the X360 without tearing. However, the "bad" ports (Akai Katana, SDOJ, Pink Sweets, MMP) default to the PCB's 2 frames, then the X360 adds another 2 frames, providing 4 frames in the output (and a few ports get even worse than this, perhaps due to sloppy coding).

Anyway, I will need to go back through the comparisons and more accurately describe the input lag differences, using your results as my guide.

[Oniros]

EOJ, will you rate the CAVE M2 STG ports (Dangun Feveron, Ketsui Deathtiny, ESP Ra.De. Psi)?

[EOJ]

I could do Ketsui, but I can't do the others because I don't own the PCBs nor do I have any good experience with them. I hope someone else will be able to do these at some point.

[EOJ]

Long overdue, but I added Espgaluda II, DFK 1.5 and DFK Black Label today.

I just have to do Deathsmiles and Pink Sweets and then I'll be done with the SH3/CV1000 games.

[EOJ]

Added Pink Sweets.

[EOJ]

I updated the SDOJ listing to describe how playing in 480p makes the slowdown much more accurate.

When playing on a vertical 16:9 monitor, I recommend setting the X360 to 480p with Widescreen, then in the SDOJ menus turn wide mode "off", turn off all extra windows, and scale the screen to 100 x 75 (or thereabouts, depending on your screen). Turn smoothing on or off to taste (I keep it on).

Doing so gives you a rather accurate port, outside of an extra frame or two of input lag compared to the best X360 ports. For reference, I did a run today on the port with these settings and 1CC'd it with 6 billion (A-Shot), a typical score for me on the PCB. Was pretty much the same difficulty as the PCB. Stage 5 still has several spots with more slowdown than the PCB, but it's not nearly as bad as when playing on higher resolutions, and it's certainly no worse than the added slowdown in the Futari ports.

Obviously anyone playing competitively should only play on 480p, any scores done in 720-1080p should be removed/disqualified.

[Oniros]

4gamer just published an article discussing most modern CAVE ports (with the exeption of Steam I guess)

https://www.4gamer.net/games/578/G057864/20211215029/

[EOJ]

A few amendments in regard to DFK 1.5 in Stage 1-5, Ura route (with C-Strong, at least).

1) The end of the "tateana" section has slowdown on the PCB, but not on the X360 port.
2) The second midboss has slowdown when you use your hyper (shot) in its first phase on the X360 port, but not on the PCB.

These sort of cancel each other out in terms of the overall difficulty of chaining the stage, but are still worth pointing out.