Nice comparisons dood! Is it possible to use Super Metroid for one of these videos?
Edit: if anyone is going to use SM for a comparison video, remember that if you beat the game there are some different demos, like the gauntlet shinespark and the crystal flash in the landing site.
Kirby Super Star Deluxe (Japan):
Link to video
I don't know which version of the game my cart is. There is a v1.0 v1.1 and v1.2. I used v1.0 for the rom, not sure if this would affect anything.
It seems Bizhawk's BSNES core doesn't emulate the lag correctly or timing is wrong for SA-1(or both), it seems to desync during lag frames in the fight scenes from my observations.
EDIT:
Link to video
Here's console and BSNES v89 (taken using retroarch). It appears any timing changes done to the bsnes core in bizhawk to allow savestates to sync properly did not affect the timing (and if they did, it wasn't significant). BSNES without these patches does not sync correctly with the console as well in Kirby Super Star Deluxe (JPN).
Regarding the Super Metroid comparison, something i never noticed before and i don't know if it's relevant or not, but here it goes:
The title screen apears tons of times in this comparision. Note the green computers in the left and right of the capsule with the metroid in the title screen and you'll see that sometimes they have a different tone of green, as if sometimes they were turned on and sometimes turned off. In the first time it appears, for example, the console version seems to be turned on, but all emulators seems to be turned off. Most of the time there's differences and it's rare for them to have the same tone of green. I don't know why these computers act like that, changing the tone of the green, and i have no idea if it makes any difference too, but considering that bizhawk emulates the console with a high accuracy, shouldn't it match the same tone of green with the console?
I'm not sure beholder, but that's an interesting observation, I hadn't noticed that before. At first I thought maybe my capture card got the colors wrong, but I noticed the colors alternate between scenes and when Bizhawk is one color the console was the other.
EDIT:
I see what it is. I didn't apply a deblinking/deflickering function to the video before uploading to youtube, the colors actually alternate during that screen (you can verify this by frame advancing the title screen, you'll see the different tones on the monitors frame by frame).
That's what it appears to be to me anyways.
Yeah, it flickers at 60Hz. Actually a great way to spot a timing difference of an odd number of frames...
Re: special chip games: bsnes is ~99.999% accurate only with regard to the base unit. There's still stuff to discover, like timing, cache behaviour and strange edge cases.
Re: Super Metroid, it appears that Snes9x 1.52 is actually more accurate than Bizhawk. o_0 If you fastforward to the last minute or so you can clearly see how BH gets a couple frames behind by that point, while 1.52 remains spot-on with the console timing.
Warp wrote:
Edit: I think I understand now: It's my avatar, isn't it? It makes me look angry.
Link to video
Something interesting to note in this video. At around 5:17 yoshi turns a corner and gets hit in the BSNES clip, but in the Snes9x 1.52 (and the 1.43 which happens quite a few seconds before the other 3) and Console no hit is taken. There may be more of these occurrences, but I didn't look thoroughly.
Bsnes also missed the flower powerup just before the one minute mark. The powerup at about 2:50 comes from the wrong direction too.
For some reason this game causes severe desyncs. And it's not even a game with save ram. It also lacks special chips. Nintendo doesn't allow special chips and special controllers at the same time, they are scared of power consumption.
Did you have equivalent controller setups? That's the only thing I can think of that would cause this level of sync failure.
Re: Super Metroid, it appears that Snes9x 1.52 is actually more accurate than Bizhawk. o_0 If you fastforward to the last minute or so you can clearly see how BH gets a couple frames behind by that point, while 1.52 remains spot-on with the console timing.
Either bsnes really has an elusive bug somewhere, or... the only alternative possibility I can think of is that the timing crystals in all systems involved in the capturing have slight differences due to their inherently analog nature. Something like production inaccuracy and temperature effects adding over time.
Either bsnes really has an elusive bug somewhere, or... the only alternative possibility I can think of is that the timing crystals in all systems involved in the capturing have slight differences due to their inherently analog nature.
The video capture should lock into the SNES output signal, which in turn is locked into the same clock S-CPU and S-PPU is. So the video capture should not introduce timing differences.
Also if it did, it would have been seen in the earlier videos already.
This game doesn't use enhancement chips that could have their own clocks, so the only possible source of interference from other crystals would be the possible 2nd crystal (I think there is one, but haven't seen any definite information to confirm or refute existence of such crystal) that would clock S-SMP and S-DSP.
Either bsnes really has an elusive bug somewhere, or... the only alternative possibility I can think of is that the timing crystals in all systems involved in the capturing have slight differences due to their inherently analog nature. Something like production inaccuracy and temperature effects adding over time.
I'd think the former, because this behavior seems consistent with other videos.
Warp wrote:
Edit: I think I understand now: It's my avatar, isn't it? It makes me look angry.
Well, it seems to be that bsnes core's inaccuracy remains fairly stable at a couple frames per ten minutes, while Snes9x 1.51 fails much more prominently in this case.
Warp wrote:
Edit: I think I understand now: It's my avatar, isn't it? It makes me look angry.
