Nothing in the game is "frame perfect." Theoretically, every non-buffered input must occur during the 3-5 frame window of the previous animation. However, I understand what you mean about tough timing. Really, it's tough spacing. Sometimes a jump input can happen at an uncomfortable amount of time after the previous footstep sound. It's tough to eyeball things in this game, because you can't really do pixel-perfect anything like in Super Mario Bros. The Prince's animations have too much life-like motion, and the visual foot placement at any given moment doesn't really mean anything. Really, all you can do is mimic the TAS timing until your ear receives confirmation of the correct footstep sound timing.

The same thing goes for Guard movement. Sometimes the TAS waits a few animations from a standstill, and looks for some exact timing in the Guard's movement, prior to doing a standing-jump past the guard. Also a ridiculous thing for a human to attempt. lol.

However, don't let this discourage you. If you disregard all Guard tricks, you and the other top runners are still ignoring tons of precise platforming patterns that the TAS is doing, independent from Guard areas, which can be memorized and practiced. I can say the best way is to create an alternate route with Debug Kill Button activated. Remove the Guard variables, so you can focus 100% on the rest of the levels, and record your attempts until you get really close to TAS timing in non-guard rooms. Once you do that enough, the patterns basically sink into your mind permanently, and you can go back to trying as many Guard skips as you can.

If you do all of the basic TAS platforming, you guys can easily take another 30-60 seconds off your best times, without any extra guard skips.

From first frame of level one thru first frame touching the finish line, I got 36:16.353. -- 30fps is insufficient for an exact frame count. In my experience, my 30 fps recordings were usually short by at least one frame (shorter time than they should be). If we use that rule of thumb here, I'd say the frame count is probably 130,797 = 36:16.365747.

This is assuming the twitch lag didn't affect the timing for the encoded video after the fact. Anyway, this seems to be in line with the current posted time, so all good. Just thought I'd time some of these for fun. :P

In accordance with the recent rules change to more clearly define the Start and End times for this game, I've updated my calculation for the frame count for this speedrun. Starting time from the timestamp associated with the first frame that the first level is visible onscreen, and Ending time from the timestamp associated with the first frame showing the Prince's final position as he has touched the finish line, we have an official console frame count of 126,915 frames = 35:11.772126.

There is another issue with starting the game late. Aside from the fact that previous games under these conditions have started on the First Frame of Level One, there is also the problem of floating lag frames. The SNES CPU does not operate at exactly 60.09881389744 fps. It is throttled to this output timing, but it has a floating time for processing data.

In some rooms, the CPU takes longer in a given process, and an extra Lag Frame is added, because the CPU wasn't ready in time for the next frame.

In other rooms, the CPU may be ready sooner, and the next process begins one frame early. Over the course of the game, these Early and Late processes tend to cancel each other out. However, this process will cause inconsistency in timing for Level One. The Prince will "Stand Upright" either 72 or 73 frames after the Level has begun, depending upon whether the CPU calls for a floating lag frame during the first second of gameplay.

So, some runs will be 1-2 frames faster than others, simply by having time arbitrarily start in the middle of the level, rather than on the First Frame, at the beginning of the Game.

More specifically, a "73" will likely take advantage of a 2 frame swing when it gets a Lag Save in the next room, while a "72" will likely take an extra Lag Frame in the next room, resulting in a 3 frame swing. Any other random discrepancy could result in a 4 frame difference, the equivalent of a full animation, just as the run is starting. The only way to avoid such inconsistencies is to have every player start timing when the Level Begins, on the First Frame.

I would strongly encourage the MODs to consider making the rules state that timing begins on the First Frame of Level One, as this is the Most Consistent convention in terms of Precedent with certain prior versions of the game, AND creating a fair playing field by having everyone Start at the Same Time.

The First Frame of Level One will always be The First Frame, and every run, RTA or TAS, will always be run under the same conditions from The First Frame.

If the game started on Level 10, the Prince would be falling as well, but we wouldn't be picking some arbitrary time when the A button becomes effective for grabbing the first ledge. We would say time begins on the First Frame. The same convention should be in use here, as the game has already begun!!

Personally, I would have preferred to set the timing to begin on the first frame of the first level.

If we consider consistency across the Series, we can take note of the original 1989 section. There are several versions of this game (including the NES and Mega Drive) which do not allow for In-Game-Timer, and instead direct time to begin "On the First Frame of the First Active Screen." In some of these versions, the Prince also falls down before "gaining control."

For the sake of consistency with other versions of this game, I think it would be best to begin timing on the first frame of the first active screen, before the Prince falls. You may have noticed that there has already been some inconsistency on the leader board, with some players starting time when the Prince Touches the Floor, and others when he Stands Upright. So inevitably, there would need to be some adjustments.

Hopefully, you would not be too upset with having to adjust the times again, if we agree on the other option.

yep. Speedrun.com defers to mariokart64.com on this point. It would be considered a duplication of efforts to add all those times to this board.

Yes, under normal gameplay conditions, the final pursuit will not end unless you are caught, or jump the bridge.

There are many places that seem like a good idea at first, but all of them eventually fail. Even the best cool-down spots don't work for long-term hiding, because it's like the cops can see you through the walls, and eventually creep around the corners to find you.

The place I found is not 100% perfect. I can remember now that I did try this a couple more times after I got the 18 hours, and sometimes I would would turn the TV on the next day, and find that I had been caught. Lol!

I would like to have a discussion with the MODs about using more precise language in the RTA rules section, specifically regarding timing.

All of my submitted runs have been timed from the First Frame the Prince is visible on-screen, because the game has started, and the Prince is in motion. I think it is vague to say time begins "when you gain control."

The Most Recent #1 record for Any% No Exit Glitches, 35:10, shows a good example. The guy programs his timer to begin 1.2 seconds after First Frame. So he is not timing his run until the Prince is Standing Straight Up, 73-74 frames after the Level begins.

I think we need to say exactly in the Rules when the timing will begin, and when exactly it will end.

For example:

1. Timing begins on the First Frame the Prince is Visible Onscreen. -OR-

2. Timing begins on the First Frame the Prince is Standing Upright (1.214 seconds later)

My proposal for ending time language:

Timing Ends on the First Frame of the Prince's final frozen position at the Finish Line on Level 20, OR the First Frame when the Prince is no longer visible, behind the curtain.

I have been ending time when I pull up the Select Menu after the Prince crosses the finish line. But theoretically, we could say the run is over a few frames before that, when the Prince stops moving.

I think we should discuss these options, so everyone knows exactly how we time runs, and how MODs might re-time runs if it's very close to a 1 second barrier.

To further clarify, you must start a run with the "Game Start" option on the Main Menu. This will give you the default Level One pre-loaded password with 0 time elapsed and 3 bottles hit meter at bottom.

You cannot input a new password with different parameters.

I have bumped this old discussion, because I want to address the RULES.

See below --

tasdasd, it really comes down to what the community agrees upon as the definition of 100% for a given game. Even if the game doesn't properly track progress or obtainable objects on-screen, we as a community can still make rules dictating what must be done.

Under your definition of 100% being to drink the minimum number of LPs to reach 15 health meter on-screen, the minimum number is Zero, because you can do an exit glitch on Level 8 to gain 15 bottles + Invincibility. lol.

Anyone interested in referencing that glitch can look up my obsoleted run in the Any% Standard category, 20m15s, and go to the 5:30 mark in the video.

I used to play this game on the Xbox 360.

On the final chase, there is no cool down, so you can't "accidentally" get away and end the pursuit time when all of the cops crash themselves and disappear.

The bus station mentioned earlier is not good enough for long-term record attempts, because cops will eventually come up from the station and get on top of the buses with you, forcing you to constantly monitor the game, and perform a risky escape (without landing on cops and getting caught) before returning to the top of the buses to wait again.

The only feasible long-term strategy is to hide from the cops during the final chase, so they can't find you, but also the pursuit never stops.

I won't spoil it here, because others should have fun trying to find the right spot for this. But I can say that I did find such a spot. I would use a wired controller, and leave a heavy book resting on one of the analog sticks, so the game & controller would never time out, and I was able to let the pursuit timer run forever.

I remember the first time I successfully did this back in 2008, I let it sit for 18+ hours, and then finished the pursuit to make it official and get a fun screenshot (photograph of the TV).

I can't remember if I did it again to get a longer run. I think if I go back through my personal photos, I can see if I ever did one longer than 18 hours.

I had some quick thoughts about 100% in the new category labels stickie thread. I'll mention them here real quick. I think we could eventually sub-divide 100% into 2 sub-categories, the same way they've set up Any%. That would be "No Exit Glitch" for the existing run, and "Standard" for a glitch category, which would theoretically include kill button & gate skip, while also having to track down all guards, and all bottle types, and any other collectible objects, like Sword.

For further symmetry of categories, since "All Levels" has been set up as a separate category for the non-%-related condition, another separate category could be made for "All Life Bottles," which would be a less strict "completist" category, involving only the 20 Life Bottles, while avoiding other unnecessary miscellaneous Big bottles, Guards, and Small Bottles, wherever feasible. This one could also be sub-divided by glitch vs no glitch.

I actually just started testing a new glitch sprite that Challenger found earlier this year, which might be perfect for routing "All Life Bottles." I will actually go ahead and start doing some preliminary work for this!

ah hah. Looks good!

I was wondering about other ways of organizing them, like bundling the glitch categories together, or bundling N.E.G. with A.L., since they both naturally do all 20 levels, or bundling all 3 together under Any%.

But after thinking about it, I think I like it best the way you did it. Any% is beating the game as fast as possible, with or without glitches. Whereas, All Levels is a separate arbitrary condition.

So I agree! Good job. And thanks for all your hard work!

I'm guessing whenever more runs appear for 100%, we might possibly see a similar breakdown there, having N.E.G + Standard sub-categories, and a separate category for "All Life Bottles" as a similar label to "All Levels," with this one only focusing on the 20 big Life bottles, and disregarding miscellaneous Big bottles, Guards & Small Bottles.

On the subject of memory for guard status, I'm curious to hear about that, because it's probably much more complicated. Like potions, guards will probably have a location related to their spawn placement; but from there, they can move quite a bit, and into different rooms, and end up dead or alive in many different places with pixel precision. Furthermore, if the Prince leaves the room while the guard is in the middle of his death animation (crumpling to the ground), the guard will come back to life after you return to the room 2x.

The Gold Skeleton at the end of Level 12 will always come back to life after 2x revisits.

It's also possible to use select menu to skip the Woman's intro, and move around while she's still jumping in, causing her to freeze in mid-air. You can manipulate this further by exiting and entering the room many times, until you get her close to the ground, then she can fall through the floor and die without activating the gate. lol.

Also possible on Level 15 to get 2 active guards in the same location, and which one is loaded depends on which room you're in. Each time you switch rooms, the previous guard reloads with full health.

Just some examples of how there must be a complicated set of memory points to cover these things.

==================

On the subject of the kill potion again, it was mentioned by GMP that the purpose of drinking detrimental potions was to clear the level map more thoroughly. On this point, we could also argue that the kill potion is in the same room as a target Life Potion. So the room itself is a cleared diversion anyway. Really, the Kill Potion there serves as a trap, the 50-50 chance of choosing the right or wrong drink. If you drink the kill potion, it is only 1 room past the checkpoint, so you would continue only 1 room back, but then again, the memory for all previously drunk bottles will be reset, such that they could be respawned if you backtracked into the first half of the level with Gate Skip & Invincibility.

I think I also mentioned previously that if you grab Invincibility during level 8, you can drink the Kill Potion and you will not die. That trick costs about 30 seconds though.

No, this game doesn't track any items in memory. It loads levels based on the stored password. The stored password tells the game what level is loaded, how much health is available, and how much time has elapsed.

The only temporary item that might be stored in memory is the sword, during level 1. If you start level 1, you do not have the sword until you pick it up. If you start any other level, you automatically have the sword.

We have established here, and on the TAS forums, that you can legitimately use Exit & Continue as a strategy in this game for saving time. For the 100% category, it would make sense to use it on Level 11 to quickly restart and move to the secondary route.

Any checklists would be strictly outside the game, and subject to peer review, as to whether or not all items were acquired. Therefore, I don't think it necessarily matters whether they are acquired once, or multiple times, depending on unintentional deaths. Seems like an unnecessary obstacle to require re-grabs.

If I had a vote, I would say the rules could clarify that every bottle and guard must be checked off just once, regardless of when any unintentional deaths occur. So if a death causes you to go back, you don't need to redo everything from the previous life.

However, I suppose we must defer to redos if that is standard practice for other games. (I know consistency across the community is important.)

Bumping for MODs! :O

Ah, ok. That makes sense. I think that's a bit more accurate too, since Guard skipping might be considered a "minor" glitch.

• "Any% (NMG)"
• "Any%"
• "Any% (Warpless)"

That looks good.

For the rules, the main distinction between "Any%" and "Any% (Warpless)" would be that "Any% (Warpless)" requires the player to complete each of the 20 levels during the run. Otherwise, both categories can use any Exit Glitches for Sprite Manipulation & Kill Button activation.

Also for "Any% (Warpless)," Warping is not specifically banned in the sense that you can temporarily warp as part of a glitch combo. It just won't be used to skip levels, since each individual level must be completed during the run.

And as I mentioned previously, I think Skinomarsh & Zotmeister's Warp-8 runs should be merged into the "Any%" category, and the distinction between gate skipping or no gate skipping should be discontinued. Especially now that Challenger has demonstrated that Gate Skipping can be done in the context of "minor" glitches (using a Guard Fight to bump you through).

Hello all. So, if you guys have been following this particular thread, you know that I've been working for a long time on this concept for a "Warpless" Any%, where any glitches can be used, as long as all 20 levels are completed. As my individual level runs have been perfected, I've been steadily bringing down the projected completion time for a theoretically perfect human run of this proposed category. The one hold up was that, for a long time, I never did put together a complete run that I was satisfied with. Part of it is my perfectionist mentality, and also having wanted to demonstrate all of the main important tricks in a single run.

However, today, I came across a perfectly suitable run that I performed in April of 2020, some 3 months after I originally proposed the category in another thread.

I think that for the time being, I'm no longer majorly concerned about waiting until a flawless run comes along before submitting it, mainly because all of the important tricks are now well documented here in this thread, especially with the new Compilation video I posted here in the previous post, above.

So, I would like to submit this 2020 speedrun for the new category. My previous category recommendations had been based on some obsolete ideas. Now, I think it's best for the two main glitch categories to be distinguished by Warp vs No Warp. (Warpless).

This translates the best with other games -- trying to reach the end as quickly as possible, vs. trying to beat every level as quickly as possible.

At present, this can all be executed by moving all of the "no gate skip" runs into the main glitch category, which is called "Any%." This would allow my 11 minute run to be obsoleted, in favor of my 7 minute run. I would change the main competitive category name from "Any% (No Warp)" to "Any% (NMG)." [No Major Glitches, as per GMP's suggestion.] I think this label is more clear, as the intention is to bar any type of exit glitch, regardless of its effect. Not all glitches are specifically used for "warping," so calling them all "Warp Glitches" isn't exactly accurate. And "NMG" allows for acceptable minor glitches, generally associated with guard fights.

In short, I think these should be the current categories:

• a main Glitchless category called "Any% (NMG)", for all of the classic competitive speedruns;

• an "Any%" glitch category for beating the game as fast as possible with the use of Exit Glitches (including warping to skip levels);

• an "Any% (Warpless)" category for using Exit Glitches to beat All 20 Levels faster than the classic glitchless category.

I think this would clear things up considerably, and is the best focus for the various speedrun types. Of course, once the 100% routes are figured out, those categories can be added as well, as variations of the "100%" label.

Here's a link to my "Any% (Warpless)" Speedrun, recorded on April 22nd, 2020, and completed in 84,832 console frames = 23:31.54200055873. A brief level by level commentary will be provided in the YouTube description.

If the MODs will consider my category proposals, please let me know when things might be updated, so that I can go ahead and submit this run for the new "Any% (Warpless)" category.