Yeah, I like having everything on the same board to start with. If it turns out that the out-of-order runs start completely overshadowing the rest there's no reason we couldn't split them up then. I think someone would have to really take full advantage of alternate ordering beyond what we've already thought of for this to happen.

I think it's essentially the same story for runs from the same runner obsoleting each other. If we get to a point where people are actually saying "I think the run I did in order is important enough that it should appear on the leaderboard, not just the other one" I see no reason to think we couldn't talk about changing it again then. But until that happens, I'd probably keep it at one time per runner, since having unranked times mixed in with other ones really does make it harder to interpret the rankings.

Yeah, All Visions definitely seems like it should require playing all visions, and not just because of the name. Even if we wanted to re-title the category to place the focus to be on rescuing everyone and not completing the stages, we'd also have to think about whether it should require actually playing through the extra vision, since that's really just the reward you get after completing the rescue objective. To me neither of these things seem worth the change (including breaking compatibility with existing runs) unless there's actually demand for a Nahatomb-free category.

For 100%, well, it also obviously really wouldn't be 100% anymore, but on the other hand, renaming it something like "All Post-Game Icons" (or ideally something a bit catchier) could also help to clarify why it doesn't require playing through the entire game twice.

In the end, though, I think part of the appeal of these categories (maybe most of the appeal) is the completionist aspect of them. For that reason I imagine most people simply won't be especially interested in skipping the final boss, even if there's ultimately not a strong objective reason for requiring it.

As for the overall issue at hand, I didn't have a strong opinion on it at first, but based on the discussion so far, I am absolutely convinced that playing stages out of order for 100% (I'll get to All Visions in a moment) SHOULD be allowed, even if it doesn't really seem to make too much difference from a practical standpoint. Ultimately I think using a developer-intended route to achieve the goal for a category should never be forbidden. After all, almost all categories in almost all games (including No Skips / Glitchless categories) permit at least some strategies NOT intended by the developers; here, we're talking about banning something that IS intended to be allowed (playing the stages out of order) for no concrete reason, which seems completely nonsensical.

And in the case of the 100% criteria, it is overwhelmingly clear that the developers deliberately allowed the stages to be played in any order, because the 150 dreamstone achievements can only be accomplished from the level select screen (NOT a first playthrough), and the game additionally doesn't reward collecting these in the "correct" order in any way. The fact that this is fundamentally different from how the criteria for the other categories function (unlocking the Extra Vision or simply completing the game) further emphasizes that it's an intentional choice from the developers. The game doesn't just allow you to use the level select screen, it forces you to use it. Defining the category in a way that's basically "yeah, but just pretend you're not actually allowed to use the level select screen" would need to provide some substantial practical benefit to be worth considering, I think.

It's worth noting that there is an argument to be made that a "true" 100% category would require first playing through the entire game once (in order, from a fresh save file) to unlock the level select screen BEFORE collecting all dream stones by replaying each level from the level select screen, but there would still be no reason at all to play the visions in a particular order the second time through. It's also not at all clear why anyone would want a category like this; it would essentially just be "do an Any% run and then also do the current 100% NG+ category," and that really doesn't seem to add a lot of value over completing those categories individually as they currently exist.

As for All Visions, it's definitely less clear whether this should be allowed, at least in principle. The game does at least allow you to rescue all Phantomilians the first time through, before unlocking the level select screen. On the surface, that might actually be a more sensible way of defining the category than rescuing all them from the level select screen, but as Balneor addrressed above, requiring that would both require watching long unskippable cutscenes and also introduce a drastic timing difference between game versions that otherwise doesn't exist (with different cutscenes being skippable depending on version). It would also render all existing runs invalid (or at the very least make them hard to meaningfully compare to runs completed on a fresh playthrough). In light of all that, I do understand thinking that the All Visions category should be done in order such that it roughly simulates a fresh playthrough despite not being one, with the level select screen being a technical convenience and not something the game itself imposes on the category. To me this all feels like a lot of unnecessary baggage, though, compared to what could otherwise be a very simple definition for the category: just play through all of the visions from a completed save file with nothing unlocked. It's easy to understand and it should also allow timing methods, etc. to carry over from the 100% category.

As for how runs should be timed, I haven't given this much thought yet. Requiring the run to start from Vision 1-1 just for timing purposes obviously is a bit arbitrary, so I would be fine with any alternate method as long as we can be sure it can still be safely applied even in cases with poor capture quality and inconsistent brightness (e.g. as from a run recorded via a camera pointed at a TV screen).

I was a little bit confused at first too. You're saying you approve of playing the stages out of order if it IS faster, yes? This seems to run a bit counter to your earlier concern that doing so seems like cheating (which I still think is certainly not the case for the 100% category, for reasons I'll reiterate in my next post), but it's a valid stance I suppose. It's also just a bit surprising generally; normally you see people wanting to ban certain methods of play because they do give some measurable "unfair" advantage over playing in a more traditional way, rather than wanting to allow them for that same reason.

OK, I'm planning to update the category-specific rules to what's listed in the top post here tomorrow. I'll leave the sections that are shared between categories unchanged for now, since that would need to be coordinated between categories and probably should be part of a larger effort.

Nothing about the change is irreversible, so don't hesitate to raise concerns even after the rules are updated.

For reference (especially for anyone reading this after the change has happened), the OLD rules are the following:

Formerly known as Glitchless. Beat the game without using skips, or certain tricks coming from these skips.

Runs requires a video to be deemed verifiable.

Restrictions : =1= Turbo/Autofire, and left+right/up+down button presses based tricks are banned (e.g fast waterfall walking). =2= No wall-clip tricks use or skips. =3= No pulling enemies through unintended doorways (mainly concern the 4-1 clock skip). =4= No out of bounds. =5= Remove moving platform speed tricks are not allowed. =6= Skips between different rooms are not allowed. =7= Skips between parts of the same rooms are not allowed under the condition that : the place the player would arrive at from the skip is, under normal gameplay circumstances, either not visible (e.g : 6-2 end skip / 4-1 end room door skip), or unreachable (e.g : old 6-1 skip before wall-clips were a thing / 6-2 giant Moo block bypass). The only exception to that is the medium plane jump situated at the far right end of the final room in 4-1.

If you're running on the emulators that are allowed to be used with a modified disc speed, it is very recommended to show proof of it somewhere in the footage of the run. Maximum allowed values are : - 2x for duckstation ||| - 4x for retroarch/beetle. If you've played on a PS2, please indicate on the same field if you've used the Fast Disk Speed feature. If not, just add --- as a value.

Timing rules :

• Timing starts upon the first frame the player starts moving, after getting past the beginning cutscene of the first Vision.

• Timing ends upon the first frame of Nahatomb's last health bar breaking up into pieces, after decreasing to zero.

I've added the word "grab" to rule 3 to make it more clear that it bans the 5-2 gate skip.

Thanks for both your feedback and concerns! I'm going to keep this thread open through (at least) the end of the week to make sure most people get a chance to look over it and give any feedback they have (and also to give me a chance to notice any major mistakes).

Also, I wanted to keep the section addressing when accidentally doing a skip invalidates a run, mainly because I've watched a lot of glitchless Sonic the Hedgehog tournaments, and it's a question that comes up multiple times in every single one.

And honestly, I once even reset a run after accidentally doing the 2-1 door skip even though I probably could have salvaged it (time-wise) by going back and grabbing the key. I wasn't sure I wanted to submit a run with a banned trick even if it was an "obvious" technicality, but having it be explicitly allowed would have made me more comfortable with it.

I did consider heavily that it might make it scarier because there's "more" to it, but I ultimately this might be balanced out by the fact that it will make people LESS afraid of unintentionally doing something against the rules, especially if they don't know what all of the particular examples of banned tricks referenced in the current rules refer to.

I've specifically seen people be afraid of doing things like the double jump near the beginning of 2-1 to skip the small platforming section, the door animation skips, certain damage boosts, etc. so my goal was to let people find clear answers to those questions.

I expect people will mostly learn the category by watching existing runs, since those are guaranteed to be "safe," and then consult the rules once they want more detail.

UPDATE: Rules have now been updated.

ORIGINAL:

I am working on an update to the rules for the Any% (No Skips). The goal is not to actually change the rules, but rather to add clarity to the existing rules. Ideally, this will help to both make the category more accessible to new runners and make the category easier to moderate.

I'd like to get feedback about all aspects of this! Does it actually make the rules clearer? Are there any rules that still don't make sense? Is it too wordy? Have I overlooked something?

I'd especially like to make 100% certain that the modified rules 1) don't accidentally ban anything used in any existing runs and 2) don't accidentally ALLOW anything that was previously banned (even implicitly). Thanks for your help!

I'm also compiling a list of specific examples (with video clips) of tricks that are and are not allowed under these rules, but I haven't finished that yet.

Without further ado, here is the proposed re-write:

Formerly known as Glitchless. Beat the game without using glitches or other tricks to skip meaningful portions of the game. Accidentally using a skip or glitch is acceptable ONLY IF no unfair advantage is gained; any accidentally-skipped section MUST be replayed correctly (e.g. by dying intentionally) for the run to be considered valid.

Runs require a video to be deemed verifiable.

Control rules:

1. Turbo/Autofire and tricks that require pressing opposing directional buttons simultaneously (up+down or left+right) are banned. These can provide an unfair advantage over using an official controller.

Category rules:

2. Do not use glitches/exploits to travel through solid walls or barriers. These exploits include (but are not limited to) the "single-frame tap" method, preserving velocity from moving platforms to carry Klonoa through walls and barriers, and using giant moos to squeeze through nearby unintended gaps. NOTE: This does NOT apply to "solid" objects that can simply be walked through normally, including doors DURING their opening animation.

3. Do not carry, squeeze, or grab enemies through unintended doorways or barriers, i.e. those where enemies normally do not fit.

4. Do not throw enemies through unintended geometry to hit targets earlier than intended, including the vine walls in vision 3-1.

5. Do not enter rooms blocked by key doors without first using a key to open the door.

6. Do not travel between separate rooms except through intended doorways.

7. Do not travel to "unreachable" areas of the SAME room earlier than intended. "Unreachable" areas include paths/planes that are entirely off-screen and can only be reached by unintended triple jumps, require heavily exploiting enemy re-spawn positions to reach, or can only be reached directly by checkpoint/death abuse. NOTE: This rule does NOT generally apply to using smaller unintended DOUBLE jumps between paths/planes, jumps where the target path is visible onscreen, or damage abuse that would be readily understood by a casual player.

8. Do not traverse "out-of-bounds" regions (areas that are not intended to be reachable at all).

9. Please contact the moderators/discord before submitting a run that uses ANY method to skip a meaningful portion of intended gameplay (especially something not used in other runs in this category), even if it's not covered by the above rules. This is simply to help avoid the use of newly-discovered skips or "loopholes" that are likely to violate the spirit of this category.

Emulation rules:

10. For CD loading speed, maximum allowed values are : - 2x (Quad Speed) for DuckStation ||| - 4x for Retroarch/Beetle. If you've played on a PS2, please indicate on the same field if you've used the Fast Disk Speed feature. If not, just choose --- as a value. NOTE: If you're running on the emulators that are allowed to be used with a modified disc speed, it is strongly recommended that you show proof of the setting used somewhere in the footage of the run to make verification easier. Runs with allowed loading speeds will not be rejected simply for forgetting to show this, but runs with loading times faster than what is obtainable on official hardware WILL be rejected.

Timing rules:

11. Timing starts upon the first frame the player starts moving, after getting past the beginning cutscene of the first Vision.

12. Timing ends upon the first frame of Nahatomb's last health bar breaking up into pieces, after decreasing to zero.

These are the specific disc speed settings to use for RetroArch and (standalone) DuckStation.

RetroArch: Use either the "Beetle PSX" or "Beetle PSX HW" cores. In the "Options" menu (which can be reached from the Quick Menu, while the game is running), set the "CD Loading Speed" to "4x"

DuckStation: Under "Settings"->"Console Settings", set "Read Speedup" to "2x (Quad Speed)"

I think I might still be leaning toward using the IGT for Balue's Tower as the "official" time, for two main reasons. The first is that, as far as I can tell, the IGT is deliberately designed to exclude sources of random variation, such as loading times, lag, etc. It only seems to count things that are directly under the player's control, and I think it would be a bit harder to do this with RTA timing. Basically, the IGT doesn't seem to be broken, so there's no need to fix it. The second reason is just that it's a lot more convenient, since with IGT it would never be necessary to do a bunch of frame hunting to re-time runs and things like that.

I decided to do some tests to see how a (potential) RTA timing method would compare to the IGT. I was particularly curious to see whether it seemed likely that there would be disagreement between the two methods on which run would be faster. I looked at six runs in total, two from Elsiz on the original version, two from Harutomo on the original version, and one from each of Elsiz and Harutomo on NamCollection. All six of the runs used the death warp (which does not save time in RTA, so would definitely lead to a discrepancy if some used it and some did not) and all were done on PS2, although not necessarily the same model. The RTA timing method I used was as follows: time starts on the first frame that the IGT changes from zero, and it ends on the first frame that the IGT displays the final time.

The results were actually not at all what I expected. The good news is that the variation from one run to the next by the same runner on the same setup seems to be extremely small. There doesn't seem to be a difference with NamCollection or the original version, either (which is actually a bit concerning, since going through doors has been shown to be faster on NamCollection in other situations, so why the difference would be missing here warrants some additional scrutiny).

The less good news is that the RTA minus IGT is actually larger on Elsiz' runs than Harutomo's. It's a pretty small difference, definitely less than a tenth of a second and probably not more than 60 ms, but it is still a statistically significant difference with well over 95% confidence. I can't explain where the difference is coming from at all (I'd have to say it's hardware differences of some kind, only because most other factors seem to be eliminated), but it does suggest that if runs for two different players were very close in IGT, using an RTA timing method could consistently favor one player over the other for reasons I can't currently explain, which seems not great?

This probably isn't a huge deal for now, since so far we've never had two different players post times within 100 ms of each other, but if anyone ever actually goes for a new record, it seems like this could start to become a problem.

Here's the raw data I used in case anyone wants to do their own analysis or notices that I've made a mistake of some kind.

(Elsiz, PS2, original) RTA 162.18 - 33.08 = 129.10 IGT 110.85 RTA-IGT = 18.25

(Elsiz, PS2, original) RTA 139.08 - 12.09 = 126.99 IGT 108.75 RTA-IGT = 18.24

(Elsiz, PS2, NamCollection) RTA 141.65 - 15.73 = 125.92 IGT 107.66 RTA-IGT = 18.26

(Harutomo, PS2, original) RTA 142.87 - 9.21 = 133.66 IGT 115.48 RTA-IGT = 18.18

(Harutomo, PS2, original) 140.00 - 7.01 = 132.99 IGT = 114.78 RTA-IGT = 18.21

(Harutomo, PS2, NamCollection) 135.06 - 5.29 = 129.77 IGT 111.58 RTA-IGT = 18.19

Also, Ganzofsorrow, don't feel too bad if Balneor manages to make quick work of your time. His record for the Any% version seems to be very near the threshold of what's humanly possible, and to the naked eye it's almost indistinguishable from the TAS.

Honestly, I wouldn't mind too much if the timing rules are different between Extra Vision and other (hypothetical) ILs. In fact, I think changing to RTA for Balue's Tower might cause some ambiguity as to what the "real" world record is, so I don't know that it's worth making a change like that just for the sake of consistency. It's unfortunate that Balue's Tower is the only example of IGT in the entire game or we wouldn't even have this problem.

I do think ILs for the other visions might be kind of neat as well. There are a lot of weird tricks that seem to be right at the edge of being RTA-viable for a full-game run, and ILs would probably give people an incentive to practice and explore those tricks. I imagine this could lead to some new strategies and setups that make them easier or more consistent, which would in turn allow them to be adopted in runs of the full game as well.

In fact, even my slightly silly attempts to beat the fastest boss times on RetroAchievements caused me to adopt a strategy for Ghadius in actual runs that I probably wouldn't have bothered with (or even tested) otherwise.

Maybe a filter/column for whether the death abuse is used would make more sense than a separate category, only because there are already relatively few runs on the leaderboard and splitting it further might make it seem even smaller.

I have no strong opinion so banning it is fine with me.

The only thing that might change my mind is if a really cool pause strat gets discovered that turns it into an entirely different run that wouldn't otherwise be possible (e.g. something at least as interesting as the death abuse, which also exploits the stopping of the IGT), but this seems extremely unlikely given that there aren't even any known TAS strats that aren't already used in the current RTA record, and no pause strats are known to exist in other categories either.

This is so minor that I almost didn't bother to post, but I wonder if the new wording will make people think that they need to wait all the way until the FMV actually appears, even though I know that's not what it actually says. Maybe it'd be even clearer just to say something like "video must continue until the first line of dialogue after the fight appears"?

To clarify a bit on what I tested and why I tested it in the first place, the idea is to get something closer to what you can get playing the original disc (not NamCollection) version on either the PS2 or the PSTV, both of which are already accepted platforms (although, unfortunately, both still slower than the NamCollection version on PS2 as well).

The particular quad speed setting in question ("4x" on Retroarch/Beetle, which should be equivalent to "2x (Quad Speed)" on DuckStation) was used in this run and compared to both this run (PS2, by agne567) and this run (PSTV, by Harutomo) and found to be slightly slower than both (which is most visible in boss transitions but applies throughout). The next speed up, 8x/4x depending on convention, on the other hand, appears to be slightly faster (although it's still much slower than on NamCollection). So in short, the 4x/2x speed is definitely NOT faster than what you'd get playing on either the PSTV, or whichever PS2 model was used for agne567's run, but it IS a meaningful improvement over playing on a plain PS1 (and consequently a plain PS1 emulator).

The reason I'm hesitant, for now, to consider speeds that bring loading times even closer to the NamCollection version itself (as nice as that would be) is that, unlike with playing the original on faster hardware, there are OTHER differences with that version beyond just loading times, so it's hard to say what would be "fair" quantitatively -- you'd arguably be playing a version of the game that doesn't exist on real hardware.

But for the 2x/4x speed, at least, there's no reason to think (nor any evidence thus far) that there would be any difference from playing the real original version on a real PS2 or PSTV, and in fact there's still a slight disadvantage over those platforms.

I like the background despite its brightness. And I find it pretty instantly recognizable as the clouds from vision 1 (and the pre-rendered cutscenes) but since that's not the case for everyone I'm not sure it makes a difference.