Posts for Alyosha

Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
Let's play Intellivision! I've started work on this core. Fortunately, it was already in a pretty advanced state before I picked it up, so a lot of the complicated stuff was already done. So far I have controls and the basics of the sprites programmed in, but there is a very long way to go. There is plenty of high level information about how the system works, so getting it into a usable state should be pretty easy. Unfortunately I can't find any info about low level things like cycle timing, so for now this will be high level emulation only and lack the real hard core accuracy of NESHawk and AtariHawk. Right now there is no sound or hit detection, but CPU emulation seems in order and the characters react to the controls, so that's about it for now. Stay tuned!
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
Cool more A2600. This run looks pretty good and the graphics are impressive, too bad it's never released. Voting yes!
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
Oh yeah, you still have to use player 2 controllers since controller registers are backwards on VS. I'll probably change this in the controller definitions to make it seemless for users, it is kind of annoying. EDIT: OK I fixed it. Now you shouldn't have to worry about configuring the controller. But, some games like Super Mario Bros still require you to hit start on player 2 instead of player 1, for unknown reasons (maybe some kind of security feature.)
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
Update on VS System. Now all but one VS UniSystem games can be played! Contorls are are implemented in EmuHawk and can be used in TASing (although I'm not sure of the TASing potential of any particular game.) The one notable exception is Hogan's Alley. The game boots and runs correctly, but light gun detection for that game is broken because the palette color appears to be slightly different then expected. I'm not sure if this is a problem of the emulated palette or a weakness in light gun detection emulation, but either way it doesn't work. Dual System games also don't work, and it's not really obvious how they could be made to work, but maybe I'll work on it someday.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
Nice work! You're making good progress on this, keep it up! Just like PoP 2 this run really needs the update just by virtue of emulation advancements that have happened since 2005. Hope you stay motivated to make it to the end!
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
^ wow that would be really cool to see this console verified! Why is bizhawk behaving differently then lsnes though? Don't they use the same bsnes core? Where is the discrepency coming from?
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
As a side note to what patashu said , the site already does have runs that rely on the initial ram state to sync , specifically monopoly and clue . Monopoly in particular is not very likely to sync on console due to nature of fceux initial ram pattern. On the other hand , there are some games that won't boot properly assuming a ram state of all zeros , which I guess would be the nominal state. These are a couple of camerica multi-cart games. As someone who likes things to be console verifiable I admit this is messy, but I guess that's the physical world for you.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
nice run with some cool tricks, voting yes! In the encode the time diff list for 5-2 has : 102.7 (new) , 102.8 (old) , diff=0, what's happening there? Rounding?
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
^No worries. Good Luck True. I'll try to get some hopefully definitive tests ready in the mean time. VS system is moving along nicely. Now you can insert coins and play games requiring the zapper. Should have full compatibility soon.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
feos wrote:
Alyosha, are there any runs that blew your mind?
http://tasvideos.org/2513M.html Only the SMW ACE one from AGDQ. (It was also the first TAS I ever saw, I didn't know TAS existed before seeing it.) Also the SMB3 DMC glitch run. 8D
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
moozooh wrote:
In my opinion, and I think I can back it up sufficiently, a newcomer rec should first and foremost be newcomer-friendly. Which means they should look and sound good, have intuitive mechanics and the run itself should demonstrate easily discernible level of precision and planning without boring nor alienating the viewer. What is friendly for a pro isn't necessarily friendly for a newcomer. We already have stars for the pros.
As a relative newcomer, my experience is that: -I wasn't even aware of the newcomer recs until several months after joining -Once I did see the list, I was underwhelmed. Maybe it would be worth it to ask the newer TASers about their experiences and see if the Newcomer recs are having the desired impact? Or are even being noticed? That might help guide decisions about what to put there and how to promote it. (Or maybe it's not relevent to newcomers at all and can be re-purposed.)
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
Oh nice it's submitted. Voting yes since I think this is one of the better a2600 games out there. The music and gameplay are both really good. As a side note this game also happens to be one of the hardest to emulate , as it relies on cycle accurate timing of TIA instructions to get the graphics right.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
I'll add that to my list of things to fix. I'll probably look at it after I get VS system in working order.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
@feos: Well having read that over, to me it looks more like a retrospective thread, and not so much a 'this is the best we've got' thread. Even at the time of it's creation (2012) it seems to be looking backwards. Now in 2016 when TASing has matured considerably and standards for runs have risen very high, the things that were eye catching before aren't so much anymore. I think it's really apparent with the amazing recent NES run of Legend of Zelda. Surely that run is of very high quality, but really it's now what we expect on average from a good NES run (currently it isn't even starred.) If Mega Man II were published today for example, would it even stand out at all? Personally I don't think so. I guess this is what I mean when I say the list is dated. TASing is just so much more developed and the list hasn't kept up well. Mario suddenly getting powerups from Super Metroid is simply on a different level then 25 minutes of Mario walking casually to the right. For your own reason for Gimmick:
I suddenly think Gimmick must be in that list. Though the game is not famous at all, it appears as a TAS gem and it's unique features get abused insanely well. It has a deep and complicated engine and really conduct the sense of TASing the gameplay way.
For NES maybe this is true enough, but it's not really that complicated when you have runs like F-Zero Jack Cup and Super Monkey Ball in stars for comparison. I would say a similar thing holds when comparing Ikaruga or even Pocky and rocky to Gradius, the level of play is just that much higher. I guess to sum up I would say that newcomer rec should keep up better with what the State of the Art in TASing is, because that frontier has gotten pushed very far forward recently and is still ongoing. EDIT: I should also say that some runs do hold up well over time, that Superstar Soccer run is just really funny.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
I liked this run way more then the other X series runs. The movement tech makes it feel really fast, and I like the category too. Voting yes! I also think it deserves a star.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
Samsara wrote:
I'd vote for Gimmick to stay. We honestly don't need to throw it away in favor of more Mario. There's nothing wrong with having both co-exist, except for "rules" that hardly anyone agree on in the first place.
What are the rules exactly? Are they listed somewhere? (If they are I couldn't find them.) Personally I don't think Gimmick really stands out as being exceptionally good that it warrants a newcomer rec, but I would also say that about mega man 2, mario walkathon, Family Feud, and even Gradius as well. Maybe it's just me, but the newcomer rec list seems really dated. I think Ikaruga for example would serve better then Gradius in there, and one of the recent Mega Man X series runs would be better then mega man 2.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
I've made some progress on VS System, and now you can play some VS system games on NESHawk. There is no 'insert coin' feature yet, but you can play in free play mode by setting dipswitches 1,2,3 to '1' in the NES->advanced settings tab, then reboot core. This is a work in progress, so some games will still be incompatible, but I should be able to make rapid progress and have it pretty well finished by the next release. Right now you can play VS Super Mario Bros and several other games from the dev build. *NOTE* Controller registers are reversed in VS System, so EMUHawk player 2 is VS System player 1, so make sure you set up player 2's controller. I might end up changing this to normal later but for accuracy right now that's how it works.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
Very impressed with the technical work that went into this run. So many surprising improvements on NES games lately, and with a bonus of being verified on console, really cool!
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
True wrote:
The difference is that of one settable register on hardware that is on the cart (8-bit register?), to three specific bits set and all others cleared out of 16384.
Actually half the RAM is explicitly cleared (0-0x400) at startup, and most of the rest is properly initialized by loading values from ROM, so it's not 3 out of 16384, more like 3 out 16-20 (for whatever RNG bits that I haven't had time to track down.)
Personman wrote:
The default branch always begins from zeroed out RAM. On a case-by-case basis, "arbitrary starting RAM state" submissions can be accepted to Moons, much as we do with "game end glitch" runs. We never draw a distinction between runs based on the number of bits they flip, unless that number is 0.
Personally I don't like framing anything in terms of moons-vault since I would hope to see that distinction removed someday, so I hope whatever decision is made can avoid that, but that being said, doing things on a case by case basis seems fine to me, not much point trying to frame a sweeping policy on what might amount to a small handful of runs overall. Also please replace the submission file with this one: http://tasvideos.org/userfiles/info/34501518718338775 I'll also updates the submissio ntext a bit to clarify what RAM is set to.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
Invariel wrote:
My question, entirely unrelated to the setting of RAM beforehand, is why you use abilities with no result early in the run (0:24, 0:35), and why you fire one Bomb too many at around 4:08 - there's nothing in the way, you just launch an extra Bomb into the wall for no apparent reason.
Oh yeah I didn't explain this before. In order to get the transformation balloon to appear, you have to 1.) use up all your bouncers and 2.) shoot the box on the left side of elevator. So that is why I randomly used some bouncers. I used them a bit more carefully in the updated submission file and it looks less sloppy, so perhaps I should update the encode as well. At 4:08 I just missed that extra shot, oops! Also fixed in the updated submission I think.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
Ok I just commited a fix to MMC3 IRQ clocking that fixes Recca, Burai Fighter, and Mickey's Safari in Letterland. All 3 of these games are known to require precise IRQ clocking behaviour and didn't work properly before this patch. Most of the relevent info can be found on this thread: https://forums.nesdev.com/viewtopic.php?f=3&t=11361 However, there seems to be some disagreement over exactly what is happening here on a hardware level and different emulators seem to be dealing with the edge cases a little differently. New hardware tests would likely be needed to resolve them. All the MMC3 test ROMs still pass with this patch though, and I have noticed no deficiencies with any other MMC3 games. (The changes would be invisible to most games anyway as they only really effect edge cases.) Still I would greatly appreciate anyone willing to test their favorite MMC3 game and let me know if anything is wrong.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
I can certainly see where the initial vector for initializing RAM I used is non-physical based on what True said, so I will re-sync this run to one where all of RAM is 0 except the 3 bits I purposely set (well it's actually 2 bits that are set to 1) RNG is usually just a matter of pressing select until you get what you want, so it should be done later today. I'm a bit surprised the run de-synced since I was quite certain I looked at RNG before and concluded it wasn't impacted by uninitialized RAM, but it's possible I just made a mistake since I didn't have the tools back then to easily check like now. Anyway I guess my opinion would be that it's fine to deviate a bit from 'ideal' NES behaviour so long as it doesn't venture into the realm of statistical impossibility. Where to draw the line? Only a certain number of bits can be set maybe? Maybe that can only be decided when more such runs are submitted. EDIT: And here it is: http://tasvideos.org/userfiles/info/34501518718338775 Please replace the movie file with this one. As stated it only sets 0x523 and 0x5FB to nonzero values. I also saved a few frames along the way with another death abuse, so that's good too.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
@True: http://tasvideos.org/userfiles/info/34498057663840024 Here is a new NG run that syncs on 1.11.8 up to the first ladder in level 2. 1.11.8 should have a start up state that is very close to a real console, given that it is anchored by your previous Battletoads tests and conforms with NESDEV info about VBlank timing. The only variable left SHOULD be the state of the DMC timer at power on, which I don't have any info on but I should be able to infer from where the run (hopefully eventually) desyncs on console. @fsvgm777: I tried the game on FCEUX but I couldn't get the level select feature to work in the debug screen. It worked just fine in BizHawk so not sure what's happening. Do you have a savestate or movie file for FCEUX that gets to the title screen in new game plus mode?
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
Derakon wrote:
Having played this game before, I had no idea it had music! How bizarre. The version I'm used to just plays a droning sound wave. If I understand correctly, music is only enabled if an uninitialized value in memory happens to be set to 1?
Well you can enable music manually by pausing with Start and then toggling between music and sound effects with A. But yes the game never writes to the address before reading from it to chekc what mode it's in, so if you didn't know you can change the setting you'll be stuck with the droning sound (if that's what your console favors.) I just tried to initialize only the 2 variables I initially mentioned and set everything else to 0, but the run then de-synced and I don't know why. So, looks like there are other uninitialized factors at play here that I wasn't aware of.
Alyosha
He/Him
Editor, Experienced Forum User, Published Author, Expert player (3560)
Joined: 11/30/2014
Posts: 2744
Location: US
What are the actual values written in this run?
The initializing values are: FF,FF,FF,02,FF,FF,FF,FF,FF,FF,FF,01,FF,FF,FF,FF Repeated across all of RAM. The other bits at the actual addresses at issue, 0x05FB and 0x0523 don't matter since the game explicity ANDs them with 1 and 3, so this run is basically setting 3 bits. Now if you say I should be using 00 instead of FF I can certainly do that too, or even set all of RAM to 00 except the 2 addresses I explicitly need to, maybe that would be less controversial.
Right, I remember discussing this with adelikat on IRC some time ago, when he was about to add this feature, and his opinion (and mine too) is that such runs shouldn't be obsoleting the ones that don't manipulate startup RAM. Because it won't be a fair competition, after more than a whole decade of not using this in TASes.
Why would it not obsolete the existing run? Start up RAM is already manipulated with the default RAM setting, it's just that that manipulation was consistent until now. To me it seems like just adding another variable to optimize (while still being physically reasonable of course, as True points out.)