SML forensics
8 years ago
Germany

Due to a recent emergence of a new run whose runner (KnucklesMaster) now claims WR, I have done some deep insight. I compared my run, Tycon's run and KnucklesMaster's run.

http://i.imgur.com/nBU9D8t.png

Each run has been downloaded from Youtube, fps was changed to 60 and they have been trimmed to start at the same time. For the record, Knuckles' run was screen recorded off of the emulator and my run was recorded with Debut Capture software and ezcap, at PAL60 (I.e. 60 hz) into avi (but I used the youtube video for this test).

I will make this video available on request.

http://i.imgur.com/LbuPG28.png (Frame 28)

It is a bit odd that tycon falls behind. I have noticed that the ingame timer in tycon's run is not the same as the other runs after a while. This would mean the framerate in tycon's run is not true to real time but slower.

The calculated time above in red text is therefore not true. (TODO: Adjust framerate and recalculate time)

The ingame timer changes on the same frame up until the timer says 394, at which point it becomes noticeable that tycon's run is slower than real time, apparently. (This would mean ~3 frames slower every ~6 seconds.)

There is no way for me to tell why this happens but I will note here that both his and my youtube videos have different framerates, and it is unknown how Tycon processed or encoded his video. As long as it doesn't suddenly catch up or I see other inconsistencies, there is no further problem.

http://i.imgur.com/jBqejKX.png

I reach the goal in 1-1 at Frame 1576 (0:00:26.267). Knuckles also at Frame 1576 (0:00:26.267). Tycon at Frame 1589 (0:00:26.483).

http://i.imgur.com/jnavKlp.png (Frame 1576)

Although Knuckles' scoring finished 2 frames sooner than mine, the fade to level 1-2 seems equally fast and we start running at the same frame (2288). Tycon starts running at 2306.

Throughout 1-2, there is nothing worthwhile to note. My run and Knuckles' are pretty much equal, while Tycon falls further behind due to the framerate problem.

I will not go into detail for the remainder, but I will point out other things:

  • It is not possible to tell for sure if the input file in Knuckles' run was tampered with (so that it's actually a TAS and not just an emu RTA). This suspicion may have been avoided if the runner played on gambatte or bgb.

  • KnucklesMaster has paused in his run in boss levels. It is not possible to luck manipulate on console because the outcome of the minigame changes every frame and waiting even 1 frame on the title screen or delaying anywhere throughout the run would mean, the RNG can't be predicted. Yet the runner pauses right before the minigame in 2-3 and 3-3 (also in 2-3 for unknown reason). Paired with the oddity that he pauses and unpauses inhumanly fast there is no other way for me but to deem his run illegitimate (a cheated run).

Besides, I don't know what everyone else's stance is, but because ¤ there is ever so slight differences (framerate, timing, lag) in emulators ¤ emulator runs can be cheated (input file modification) ¤ non original hardware is used (instead of a GBA linked to a gameboy player, or a snes controller linked to a sgb2, you can play on keyboard or third party usb controllers) ¤ macros can be used, and you can't detect them I would urge to not accept emulator runs as world records in gameboy games in general.

Germany

Same thing in Mario sunshine. Runners can use an emulator but their times will only be added if they're not below a certain point. They are strongly encouraged to get a real console and recording equipment.

That's just my opinion. Emu runs just aren't the same thing as console runs. They are different. Instead of looking into a CRT and playing on original GBA or SNES controller, they have the advantage of playing on whatever they want and on an LCD, at whatever resolution, at whatever colors, etc. Plus the other things I've listed (possibility of using macros undetected, timing differences). I always liked the idea of seperating the two (so there can be emulator WR and console WR).

If the runner gets so good at the game that they have a shot at WR, they are encouraged to get the real console before claiming it. Same thing happens in SML2: oh_deer was on emulator for a long time. Now he got a SGB2 and has a shot at WR.

The money argument, I can't buy that, really... It doesn't cost that much to get a gamecube, gb player, the game, and recording equipment.

cube $35 gb player $13 game $5 CRT $5 ezcap $30

Some people have already some of the things. Without CRT and recording equipment, that's a little over $50. That's not an argument for a runner, who has a chance at WR, to not go for it. Now for Trip World and Earthbound on the other hand, I can understand it.

But you know, I haven't really been involved in the speedrunning scene for gameboy games for a while, so... If the communities come up with new systems, maybe I'm just being an oldtimer.

Germany

Since Mugg already brought up my example I might as well add my 2 cents to the discussion.

About the claimed WR in SML1:

The case seems obvious. No real need to go into further detail. The TASer seems to be very young. Maybe he'll learn his lesson. At least he seems to show some general interest in the matter, right?

About speedruns on emulator:

Now this is where I'm a little conflicted about how to phrase things. I had been running SML2 on emulator (BGB 1.4 & 1.5.1) for almost a year and had managed to get competitive times after a while. I'm a competitive person. Speedrunning in general has a strong competitive aspect. It is fun to compete against one's own personal best as well as against other's records. And if it hadn't been for the fact that runs on certain emulators are currently accepted by the majority of the small SML2 community I am not sure if I would have had the motivation to keep going. If a person is willing to put in hundreds of hours into this hobby then they are much more likely to walk the extra mile and get "proper" equipment, eventually. So one important thing is to make it easier for people to stick around for a longer time.

While the financial hurdle is not that high for a GB/SFC/SNES setup (my numbers: SFC: €40, SGB2: €20, Game: €3, CRT: €20, Dazzle: €4, Cables: ~€20) it is still an effort that most people wouldn't take easily - most likely not when starting out! On top of that individual cases might differ significantly (availability of hardware in certain countries, personal financial situation, no/too bad internet connection for streaming, certain cartridges are rare/expensive, etc.)

This is where I think that we - as a community - should welcome ANY new runner and try to carefully give leads into the right direction. Once someone feels welcome they might be more open to advice, anyway.

That said, I DO recommend playing on original hardware for very pragmatic reasons:

  • significantly lower input delay on CRT
  • potentially lower input delay with original input devices compared to adapter combinations
  • usually better quality of original input devices (this might be subjective) (- accuracy/better comparability)

Practice on emulator is very convenient. For some games savestates make practice so much more worthwhile. However, I got to the point where I felt that playing on emulator actually crippled my progress; so to me switching to my new setup came quite naturally, but mind you, this was after almost 1 year of playing on emulator! Not everyone will or will be able to choose that path - and that is perfectly fine in my opinion.

This is where I think we as a community should put in some effort by analyzing and evaluating emulator accuracy with hard numbers (Mugg already put in some effort and I'm willing to help as much as I can). New runners usually don't have the knowledge and understanding of the matter so we should make things as easy as possible to grasp and not scare them away. If we find BGB and/or Gambatte to be accurate (within a few frames, mind you that there are different revisions of chipsets in original hardware, too. Do we ever question those?) then there is no issue, to begin with. If we find inaccuracies we should try to find a way to quantify them (compare: https://kb.speeddemosarchive.com/Super_Game_Boy_timing , where a coefficient is determined that is used to recalculate run times).

In my opinion community growth is very important so we shouldn't just take the easy way out by demonizing emulators in general. Different people might share the same passion, but simply not have the possibility to adhere to all of the recommended factors - at least not immediately (personal experience).

So, to summarize my standpoint:

Allow emulator runs, but evaluate accuracy and point out possible issues generally & individually. No separate emulator category as that would divide the community and create tension. Be open as a community. Be considerate of different individual circumstances. Don't be elitist.

I roam Twitch and visit basically every casual SML2 stream hoping to find people who share the same passion and I'm sure many others do the same for their respective games. I've been welcomed to the community very warmly and I think being open and welcoming is more important, initially, than to immediately list the "do"s and "don't"s and potentially scare someone away with a load of specifications, version numbers, coefficients, etc., but we should have those handy if someone is interested and wants to look into the matter more seriously.

Since I am bad at finding an appropriate ending to this article I will just stop here.

Arkanias566 likes this
Jamaica

I don't think it would help sml as a speedgame if we were to ban/put in a different category all emulators. Especially if this one god awful cheated run is the triggering factor, the community is small and knowledgeable enough for cheated runs not to be an issue. I think we/you should allow the most accurate emulators for the leaderboards (i don't even have a time on there lol), as a way to let the doors of sml speedrunning open for anyone serious enough. RTA video proof should be necessary for any unconfirmed runners though. And if multiple great emulator runs show up in the future, maybe put the emulator ban in place, but i really don't think it's necessary right now, as the game is not that prestigious and a higher number of runners would revitalize it.

I'm barely in the community despite having spent thousands of hours playing this game, so do whatever you feel is necessary.

About Tycon's framerate, could it be that Gameboy player EU and US are different in some way? could it be the cartridge?

Game stats
Followers
570
Runs
571
Players
184
Latest threads
Posted 5 years ago
1 reply
Posted 3 months ago
2 replies
Posted 1 year ago
3 replies
Posted 2 years ago
2 replies