3.0.0 compatibility

[ThexXTURBOXx]

To everyone wondering, here are my observations (will update as I get feedback).
I have extracted my weather seed from my ACNH save directly using NHSE, by the way. So, my weather seed is 100% correct.
In order to do this yourself, you would need a modded Switch! If your Switch is not modded, you need to brute-force it.

What works:

• Meteor showers still match exactly (even their seconds)

What does not work:

• MeteoNook does not show fog
• Some weather patterns seem to be outdated

Not yet tested:

• Auroras

Hence, if you want to find out your weather seed using MeteoNook, you are best off using meteor shower times until the weather patterns may be updated.
However: it is highly advised to not use the seconds during brute-forcing your seed.
This is due to the mentioned 4-second-leniency working only in one direction.
If you think you saw a meteor at exactly 21:13:39, but it was actually 21:13:40, the leniency will not help you and your seed will be directly excluded - even because of this single second error!
However, your seed will not be excluded if you tell the system that it was at 21:13:40, 21:13:41, 21:13:42, 21:13:43, or 21:13:44.
This also takes into account minute rollovers - so if you saw a meteor at 21:13:59, you can also tell the system that you think it was actually 21:14:00 or even 21:14:03. It will still work.
Hence, if you really want to use seconds, round them up and add 1 second for extra leniency.

My current guess is that half an hour of noting down the minutes, where there are meteors, and noting down those, where there aren't any, will be enough to always filter your list down to a handleable size (approx. 10 seeds).
Again: it is better not to use the seconds when brute-forcing. You can compare your measurements to those when you have found your probable seed(s) in the end. This is the recommended procedure.
Here is an example where I brute-forced the seed by @Rooni just using their meteor minutes (their seed is indeed the first one in the list):

[Rooni]

From my testing after 3.0.0, I get the feeling something they added doesn't make MeteoNook as reliable on certain days/meteor showers.

Possibly the seed finder is forcing itself to reverse the seed with the wrong cloud type, since the only option they give for the date I tested, 28/01/2026, are Fine00 and Billow clouds, which does not match what I see in the game (clear skies).

To clarify I've accurately timed the stars with a stopwatch doing laps for every star dropped for a total of 80 stars, from 21:30 to 21:50.
Given the 4 second leniency when filtering seeds, this should be completely valid data.
I can dump the stopwatch data here but I doubt it would help anyone.

I'm posting this to find out if anyone else is experiencing the same, because I'm seeing mixed responses online.

[ThexXTURBOXx]

Well, I will update my experience(s) above as soon as I get more data. I will try to keep it as up to date as possible, at least.

The observations I mentioned above are based on my experience from my own save file, which is on a modded Switch, as well as from my wife's save file, which is on a non-modded Switch 2.
I extracted the weather seed of my island directly using JKSV and NHSE.
I first tried brute-forcing my wife's weather seed, but I was using weather instead of meteor showers, which was a big mistake - hence my post to warn others to do it differently from the get-go and not waste too much time.

After two failed attempts, I transferred her island to my modded Switch, extracted the weather seed using the procedure above and transferred it back to her Switch 2. After doing this, I had my previous weather observations still in the "brute force tab" in MeteoNook and could rather easily tell what was correct and what not.
No joke - meteor showers are 100% correct - down to the second.
Weather works sometimes, but may also not be entirely correct some other times.

Regarding your observation: Fine00 has in fact a clear sky at 21.00, so I see no contradiction here.
The special cloud type is something you may experience on that day, but it also may not be there (most of the time it is not there - these special cloud types are rather rare as far as I can tell). So, also no contradiction here.
Why it's not finding seeds? Not sure. However, I can surely investigate when I have some free time and if you dump your stopwatch data

[Rooni]

Here's the raw data I've gathered. Do note that I started the stopwatch at exactly 21:30, so these times should be offset accordingly when filling them in.

First column is the lap number, second column the time since the last lap, and last the actual time during the lap.

  1      02:13.25      02:13.25
   2      00:26.69      02:39.94
   3      00:02.04      02:41.98
   4      00:10.13      02:52.11
   5      00:04.71      02:56.82
   6      00:11.16      03:07.98
   7      00:10.98      03:18.96
   8      00:26.95      03:45.91
   9      00:02.10      03:48.01
  10      00:01.84      03:49.85
  11      01:13.11      05:02.96
  12      00:05.95      05:08.91
  13      00:00.99      05:09.90
  14      00:03.85      05:13.75
  15      00:16.13      05:29.88
  16      00:04.97      05:34.85
  17      00:22.22      05:57.07
  18      00:01.89      05:58.96
  19      03:07.96      09:06.92
  20      00:02.81      09:09.73
  21      00:15.04      09:24.77
  22      00:01.96      09:26.73
  23      00:11.96      09:38.69
  24      00:05.16      09:43.85
  25      00:00.86      09:44.71
  26      00:11.62      09:56.33
  27      00:16.46      10:12.79
  28      00:05.04      10:17.83
  29      00:03.93      10:21.76
  30      00:04.92      10:26.68
  31      00:01.12      10:27.80
  32      00:09.93      10:37.73
  33      00:15.03      10:52.76
  34      00:01.91      10:54.67
  35      00:10.00      11:04.67
  36      00:11.12      11:15.79
  37      00:39.08      11:54.87
  38      00:01.98      11:56.85
  39      00:01.83      11:58.68
  40      00:06.22      12:04.90
  41      00:03.86      12:08.76
  42      00:06.09      12:14.85
  43      00:09.87      12:24.72
  44      00:06.02      12:30.74
  45      01:31.03      14:01.77
  46      00:08.04      14:09.81
  47      00:00.99      14:10.80
  48      00:03.02      14:13.82
  49      00:07.91      14:21.73
  50      00:18.17      14:39.90
  51      00:02.89      14:42.79
  52      00:07.00      14:49.79
  53      00:11.18      15:00.97
  54      00:18.12      15:19.09
  55      00:02.68      15:21.77
  56      00:27.97      15:49.74
  57      00:08.06      15:57.80
  58      00:02.03      15:59.83
  59      00:13.01      16:12.84
  60      00:01.89      16:14.73
  61      00:06.09      16:20.82
  62      00:06.00      16:26.82
  63      00:08.04      16:34.86
  64      00:01.85      16:36.71
  65      00:12.03      16:48.74
  67      00:22.07      17:11.02
  68      00:07.78      17:18.80
  69      00:29.00      17:47.80
  70      00:03.99      17:51.79
  71      00:01.06      17:52.85
  72      00:16.09      18:08.94
  73      00:11.89      18:20.83
  74      00:01.90      18:22.73
  75      00:24.33      18:47.06
  76      00:05.82      18:52.88
  77      01:11.00      20:03.88
  78      00:14.13      20:18.01
  79      00:00.79      20:18.80
  80      00:14.06      20:32.86
  81      00:02.03      20:34.89

*66 is removed since I accidentally double pressed the lap time there.

So, offsetting for 21:30 would result in the following data:

Northern hemisphere
21:30 nothing
21:31 nothing
21:32 (13, 39, 41, 52, 56) for 5 stars
21:33 (07, 18, 45, 48, 49) for 5 stars
21:34 nothing
21:35 (02, 08, 09, 13, 29, 34, 57, 58) for 8 stars
etc

And finally an export of the above from MeteoNook, which comes up with no seed:

{"days":{"2026-1-28":{"y":2026,"m":1,"d":28,"dayType":"2","showerType":"2","rainbowTime":10,"rainbowDouble":false,"auroraFine01":false,"auroraFine03":false,"auroraFine05":false,"types":[],"stars":[{"hour":21,"minute":32,"seconds":[13,39,41,52,56,99,99,99]},{"hour":21,"minute":33,"seconds":[7,18,45,48,49,99,99,99]},{"hour":21,"minute":35,"seconds":[2,8,9,13,29,34,57,59]}],"gaps":[]}},"hemisphere":0}

So far I've tried a few different combinations of the entire data, all end up narrowing down into 0 seeds the more I add.

If it helps any, I can have a capture card set up for the next occurence of a heavy shower.
I'll keep an eye out for it the following days.

[ThexXTURBOXx]

Your weather seed is 51229959 :)
I have not applied all your settings as there leaves too much room for only a single error to completely exclude your seed in the end.
After seeing that you have not set any "gaps", I tried setting them and this excluded more seeds than I thought.
Then, I just set the exact seconds for 21.32 and was able to filter down to just four possible seeds, which I then compared by hand:

However, this left an interesting question: why didn't your data work?
The answer: I think, the leniency does not work correctly...
As soon as I am off by only a single second in the 21.32 entry, I am not able to find your correct seed anymore.
I will try to take a look at it later. Maybe, it's an easy fix...

[ThexXTURBOXx]

Edit: removed wrong claims and updated with proper analysis

This is the code that does the per-second checks:

MeteoNook/src/lib.rs

Lines 913 to 948 in a7ea07a

	// do per-second checks
	if hg.second_mask[minute as usize] != 0 {
	// create u64 bitfield containing last 4 seconds of previous minute
	// (at the lowest position) followed by 60 seconds of current minute
	let mut actual_mask = (star_field << 4) | previous_second_mask;
	// these are the seconds that the user claimed they saw a star in
	let input_mask = hg.second_mask[minute as usize];
	let mut second_bit = 1u64;
	for _second in 0..60 {
	if (input_mask & second_bit) != 0 {
	// check 5 seconds worth for a match
	let mut check_bit = second_bit;
	let mut found = false;
	for _i in 0..5 {
	if (actual_mask & check_bit) != 0 {
	// found a match, remove it from consideration
	actual_mask ^= check_bit;
	found = true;
	break;
	}
	check_bit <<= 1;
	}
	if !found {
	// no match within 5 seconds distance, can't be true
	return false;
	}
	}
	second_bit <<= 1;
	}
	}
	// rollover the last 4 seconds
	previous_second_mask = star_field >> 56;
	}

The code only allows for leniancy in one direction.
It is always better to round up (or even add a whole second to the result) and put these in.
When trying to brute-force the seed with a meteor at 21:32:40, it will find your seed.
It will also find it if you use 21:32:41, 21:32:42, 21:32:43 or 21:32:44 instead.
However: it will not find your seed if you input 21:32:39 instead.

This is why your search yielded no result - the leniency sadly only works in a single direction.
I will update my first post about this.

[Rooni]

That makes a lot of sense! Can confirm that everything also seems to match with the seed you provided!
Thank you so much for spending the time and diving deeper into this!

As an extra, I also managed to find the seed on someone else's island with the same timing method, but rounding up and adding 2 seconds.

I believe somewhere on MeteoNook it was advised to only do it at your own island, but this also seems to work!

[ThexXTURBOXx]

You are more than welcome! I hope that your and my experiences with this will help others finding their seed(s) in the future!
Indeed, there is a hint on MeteoNook about other players' islands, but I actually don't 100% know why.
Haven't tested it myself, but I would guess that they should be synchronised between the two consoles.
Maybe, the hint is there because the consoles' times may not be synchronised or because there may be some delay through the internet connection between the consoles?
But yeah, if it worked for you, that's also great news! :)