Dreamcast architecture supposedly mirrors Naomi hardware, so the DC versions of CvS2 and MvC2 have always been considered the definitive console ports. Yet Evo staff switched to the PS2 version of CvS2 because it's much harder to obtain/maintain Dreamcasts. Unlike PS2 MvC2, the PS2 version of CvS2 is free of glaring differences from the arcade/DC versions.
A couple of people have expressed suspicions that DC CvS2 and PS2 CvS2 actually run at different speeds. Even if we were to prove this, it is extremely unlikely that Evo2k8 would switch back to Dreamcast. Nevertheless it would be useful information for all the players practicing right now.
My concern is that CvS2 has a lot of 3D backgrounds and fancy effects such as Capcom-Groove superfreeze and transparent shadows. Even if we verify that the two console versions run at the same speed by default on one stage, it might not guarantee that they run at the same speed on every stage or while characters are rolling around or performing supers.
What makes this more challenging still is the fact that even on the Dreamcast version, 3D background activity affects game speed. For example, sometimes you can actually feel the slowdown if you do Capcom-Groove supers while the rally cars are flying over the Nairobi desert stage. Since the PS2 is a more powerful console, maybe it's not affected by a little extra graphical processing?
How do we test whether DC CvS2 and PS2 CvS2 run at the same speed while nothing's happening and during intense activity too?
How could we determine whether DC and PS2 CvS2 speeds match?
Re: How could we determine whether DC and PS2 CvS2 speeds ma
I'm not really sure, but Sp00ky performed a simple test to compare 3S.
He set the DC version and PS2 version next to a supergun and had some one do a standard Yun GJ combo on each, in round 2. So basically in round 2 Yun started with max bar, and activated the moment the round started, then they did the standard GJ combo on all three.
Apparently the differences were so blatantly obvious there was no need for further testing.
He set the DC version and PS2 version next to a supergun and had some one do a standard Yun GJ combo on each, in round 2. So basically in round 2 Yun started with max bar, and activated the moment the round started, then they did the standard GJ combo on all three.
Apparently the differences were so blatantly obvious there was no need for further testing.
Looks like Jolly Ranchers & Baskin's Sherbet.
Re: How could we determine whether DC and PS2 CvS2 speeds ma
Hey, guys.
Through personal experience, both DC and PS2 are definitely faster than the arcade, but I don't notice any difference between DC and PS2.
Here are my ideas of testing them after 5 minutes of brainstorming.
1) While doing nothing
Run the ice stage on both systems side by side at default tournament settings (speed 3) and see if the cracking of the ice, shooting of the flags in the air, and the moment of time out match in both versions. However, that might just tell you the difference in 3D rendering timing. Therefore, we should choose the same characters and grooves in both setups and hold the up direction the entire time to see if there are any actual differences in gameplay speed. If there's a slightest difference in speed, we should see a progressively widening difference between the moment of landing on each system.
To negate the different loading times as much as possible, I can choose the same characters for both 1P and 2P, let the VS sequence run until it loads the characters and background and then press a button on both system to skip the sequence at the same time once they've been loaded (after the CD seeking sound stops).
2) While doing supers and shit
This is harder. Basically, keep the conditions the same as the above test, but do a same number of a move that seem to cause slowdown in 1 round and see if the rounds end at the same time against each other and against Test 1 for each system. Infinite meter and auto-JD/Parry might be a good way to cause lag over and over. I seem to remember a lv3 direct lightning causing lag.
This is how far I've thought of. What do you guys think?
Through personal experience, both DC and PS2 are definitely faster than the arcade, but I don't notice any difference between DC and PS2.
Here are my ideas of testing them after 5 minutes of brainstorming.
1) While doing nothing
Run the ice stage on both systems side by side at default tournament settings (speed 3) and see if the cracking of the ice, shooting of the flags in the air, and the moment of time out match in both versions. However, that might just tell you the difference in 3D rendering timing. Therefore, we should choose the same characters and grooves in both setups and hold the up direction the entire time to see if there are any actual differences in gameplay speed. If there's a slightest difference in speed, we should see a progressively widening difference between the moment of landing on each system.
To negate the different loading times as much as possible, I can choose the same characters for both 1P and 2P, let the VS sequence run until it loads the characters and background and then press a button on both system to skip the sequence at the same time once they've been loaded (after the CD seeking sound stops).
2) While doing supers and shit
This is harder. Basically, keep the conditions the same as the above test, but do a same number of a move that seem to cause slowdown in 1 round and see if the rounds end at the same time against each other and against Test 1 for each system. Infinite meter and auto-JD/Parry might be a good way to cause lag over and over. I seem to remember a lv3 direct lightning causing lag.
This is how far I've thought of. What do you guys think?
CC that shit
Re: How could we determine whether DC and PS2 CvS2 speeds ma
I haven't noticed a glaring difference either, so it's gotta be subtle if it's there at all. If i do any testing, i'm not at all comfortable with announcing an estimate. Anything i claim has to be backed with some empirical evidence.
Parry/JD don't seem to cause slowdown on their own. It really does come down to Capcom-Groove superfreeze and transparency effects on shadows. In fact SNK-Groove supers are way more consistent in this regard. I just remember the game slowing to a crawl whenever two fat characters rolled on the fire stage in CvS1.
Blanka's Direct Lightning used to cause a ton of lag in CvS1 but it's not as bad in CvS2. Still, it probably does happen when the crazy 3D lightning sparks cross other intense graphics. The difficulty is in consistent timing because if our test sequences are one frame off, differences in background states could totally screw up the results. The brute force solution would be to run so many tests that discrepancies average out, but there's probably a better way using program pads.
Another issue is the measurement method. Is it ok to use a capture card or is there some reason to use a camcorder instead?
Parry/JD don't seem to cause slowdown on their own. It really does come down to Capcom-Groove superfreeze and transparency effects on shadows. In fact SNK-Groove supers are way more consistent in this regard. I just remember the game slowing to a crawl whenever two fat characters rolled on the fire stage in CvS1.
Blanka's Direct Lightning used to cause a ton of lag in CvS1 but it's not as bad in CvS2. Still, it probably does happen when the crazy 3D lightning sparks cross other intense graphics. The difficulty is in consistent timing because if our test sequences are one frame off, differences in background states could totally screw up the results. The brute force solution would be to run so many tests that discrepancies average out, but there's probably a better way using program pads.
Another issue is the measurement method. Is it ok to use a capture card or is there some reason to use a camcorder instead?
Re: How could we determine whether DC and PS2 CvS2 speeds ma
Might want to contact Nicholai for some pointers, since he had to deal with the brunt of CCC2 testing scrutiny.
Looks like Jolly Ranchers & Baskin's Sherbet.
Re: How could we determine whether DC and PS2 CvS2 speeds ma
Finally got ahold of everything i needed to run some preliminary tests:
CvS2 Default Game Speed Test 01 | CvS2 Default Game Speed Test 02
CvS2 versions compared:
American Dreamcast console + original Japanese CvS2 disc + Utopia bootdisk v1.2
American Dreamcast console + burned selfboot Japanese CvS2 disc
American PlayStation 2 console + original American CvS2 disc
Results:
All three versions appear to run at the same speed, although loading times vary.
Obviously the observed results have an error margin of ±1 frame, since the game runs at 60 frames per second while the capture card operates at 30fps. The first clip shows one full round of three versions of CvS2 on the same stage with both characters chillin' in neutral. The second clip shows the same sequence of attacks performed (clumsily) by hand, containing the same number of superfreezes. In both cases, time runs out at the same exact moment. However, there are noticeable differences in loading times between versions.
These tests aren't absolutely conclusive, but at least there aren't any glaring speed differences. Who knows what'll happen if there's enough clutter on the screen to actually cause the consoles to slow down. But since there's no way of duplicating an identical situation across three versions of the game, i'll stop here. There's enough variance in superfreeze timing within any one version of the game that i don't think there's any point to worrying about console differences under extreme conditions.
CvS2 Default Game Speed Test 01 | CvS2 Default Game Speed Test 02
CvS2 versions compared:
American Dreamcast console + original Japanese CvS2 disc + Utopia bootdisk v1.2
American Dreamcast console + burned selfboot Japanese CvS2 disc
American PlayStation 2 console + original American CvS2 disc
Results:
All three versions appear to run at the same speed, although loading times vary.
Obviously the observed results have an error margin of ±1 frame, since the game runs at 60 frames per second while the capture card operates at 30fps. The first clip shows one full round of three versions of CvS2 on the same stage with both characters chillin' in neutral. The second clip shows the same sequence of attacks performed (clumsily) by hand, containing the same number of superfreezes. In both cases, time runs out at the same exact moment. However, there are noticeable differences in loading times between versions.
These tests aren't absolutely conclusive, but at least there aren't any glaring speed differences. Who knows what'll happen if there's enough clutter on the screen to actually cause the consoles to slow down. But since there's no way of duplicating an identical situation across three versions of the game, i'll stop here. There's enough variance in superfreeze timing within any one version of the game that i don't think there's any point to worrying about console differences under extreme conditions.
Re: How could we determine whether DC and PS2 CvS2 speeds ma
Good test, and I enjoyed watching the vids.
However, there could be a problem with this method. When capture cards capture at 29.97fps, I believe the resulting video is forced to 29.97fps even if the actual video feed is running at a slightly different frame rate (like 59.94 vs 60.00). But then, I'm pretty sure the consoles are running at NTSC standard frame rate of exactly 59.94fps to be compatible with absolutely any TV out there, so even if my theory is right, they might still be running at exactly the same speed.
Now that we know all 3 versions played the exact same frames for the exact same amount of time, so I think the best way to check for the speed differences for good is to use a frequency meter and feed the vsync from each console while running cvs2 and compare the values.
I still think inter-console netplay (DC and PS2) is the proof that both cvs2 versions run at the same speeds.
BTW on a random note, did any of you notice the difference in meter bar heights between DC and PS2? I always used it to determine the version of the game if it wasn't noted. Arcade has it the lowest, then PS2, then DC on TV. That's why I hated playing Claw on the DC version cuz it's really hard to see dropped claws because of raised meter bars. However, I recently found out that if you use a VGA box for the DC and run cvs2, the game actually knows that you're using VGA and lowers the meter bars to what seems like arcade cvs2 height. I thought that was cool.
However, there could be a problem with this method. When capture cards capture at 29.97fps, I believe the resulting video is forced to 29.97fps even if the actual video feed is running at a slightly different frame rate (like 59.94 vs 60.00). But then, I'm pretty sure the consoles are running at NTSC standard frame rate of exactly 59.94fps to be compatible with absolutely any TV out there, so even if my theory is right, they might still be running at exactly the same speed.
Now that we know all 3 versions played the exact same frames for the exact same amount of time, so I think the best way to check for the speed differences for good is to use a frequency meter and feed the vsync from each console while running cvs2 and compare the values.
I still think inter-console netplay (DC and PS2) is the proof that both cvs2 versions run at the same speeds.
BTW on a random note, did any of you notice the difference in meter bar heights between DC and PS2? I always used it to determine the version of the game if it wasn't noted. Arcade has it the lowest, then PS2, then DC on TV. That's why I hated playing Claw on the DC version cuz it's really hard to see dropped claws because of raised meter bars. However, I recently found out that if you use a VGA box for the DC and run cvs2, the game actually knows that you're using VGA and lowers the meter bars to what seems like arcade cvs2 height. I thought that was cool.
CC that shit