Shiny Pokémon are not recognized as legal in PKHeX

[Jorylugui]

Hello. Shiny Pokémon are not recognized as legal in PKHeX. I used canonical mode in Pokémon Sapphire Rev2.

Here's the log:

=== Starting patch run (auto mode, 1/16) ===

--- Pokemon - Sapphire Version (USA, Europe) (Rev 2).gba ---
ROM detected:
Name: Pokemon Sapphire Version (USA, Europe) (Rev 2)
Game code: AXPE
Revision: 2
CRC32: 0x9CC4410E

Patch summary:
Requested odds: 1/16
Requested mode: auto
Applied mode: canonical
Applied threshold: 8 (base game is 8)
Effective shiny bits: 16
Effective odds: ~1/15.990

Canonical reroll attempts: 529 total PID roll(s)
Compatibility: PKHeX-canonical shiny logic preserved.
Output ROM: D:\pkmn expandidos\la new new pokedex de legalites direct\Nueva carpeta\roms\Pokemon - Sapphire Version (USA, Europe) (Rev 2).shiny_1in16.gba
Output CRC32: 0x23AA706E

Patched offsets:

• 0x03A7DE: replaced with BL 0x42D650 (canonical reroll hook)
• 0x42D650: wrote 100-byte canonical reroll hook (reroll_attempts=529, retry_target=0x03A7BE)
• 0x03AB34: replaced with BL 0x4150E4 (canonical fixed-personality wrapper A hook)
• 0x4150E4: wrote 84-byte canonical fixed-personality wrapper A hook (reroll_attempts=529, retry_target=0x03AAFC)
• 0x03AC32: replaced with BL 0x409A4C (canonical fixed-personality wrapper B hook)
• 0x409A4C: wrote 84-byte canonical fixed-personality wrapper B hook (reroll_attempts=529, retry_target=0x03AB52)

Finished. OK: 1 FAIL: 0

[eightmouse]

Are you playing the game with language set to Spanish by chance?
If that's so, it might be related to the language as different languages have tiny differences!
I'll look into this, thanks for the feedback ~

EDIT: It's not the language, I just checked the log, it shows CRC32 0x9CC4410E.

[eightmouse]

This looks more like a Sapphire starter-path legality issue on my side.

Can you send one more piece of information:

• the exact PKHeX legality message shown in the legality report window

If possible, also send:

• the full PID
• TID / SID
• IVs
• whether this was generated from a fresh boot or from a savestate

If you can attach the .pk3, even better.

From the screenshot alone, the main useful conclusion is:
I should investigate Sapphire starter generation specifically.

[Jorylugui]

0255 ★ - TORCHIC - 03C9208E8EC5.zip

[Jorylugui]

I'm sending you the compressed pk3 file because it won't let me send a .pk3 file.

[Jorylugui]

While I wait for your reply, I'm going to check the generation of shiny starters in the other GBA games.

[eightmouse]

I'm sending you the compressed pk3 file because it won't let me send a .pk3 file.

Thanks, this helps.
Problem is the Sapphire starter encounter correlation, not the shiny flag itself.
I’m tracing the Sapphire starter path specifically now, using the failing .pk3 as the target, I'll update once done!

[Jorylugui]

FR and LG.zip
I just realized a few things:
1: In FR and LG, the same thing happens as in Sapphire.
2: Shiny Pokémon don't appear in Emerald even though I decided to set a 1/2 probability in canonical.

[Jorylugui]

Note for point 1: it's rev1
Note for point 2: the lag that generates such a probability still occurs.

[eightmouse]

Note for point 1: it's rev1 Note for point 2: the lag that generates such a probability still occurs.

Working on Sapphire currently, found few issues.

For point two, I don't think I can fix it, I'm using 1/16 purely for testing, for normal play 1/256 is recommended, no lower than 1/128 tho or things get messy!

[eightmouse]

Thanks, the .pk3 files were useful.

I checked them directly, and they are genuinely shiny under the original Gen 3 shiny formula.

That means the remaining issue is most likely encounter / PID-method correlation, which is the stricter thing PKHeX checks for certain encounter types.

So:

• the patch is producing real shinies
• but some starter paths still appear to be failing PKHeX’s encounter-type expectations
  I’m tightening that now. Also, make sure you're using the latest desktop PKHeX.

PKHeX can get stricter later without Gen 3 itself changing, because its reconstruction heuristics can improve, which might still break 'legality' ~

[Jorylugui]

Okay. So then I just have to report that shinies don't appear in Pokémon Emerald even if I set the probability to 1/2.

[Jorylugui]

Two other things I have to report in FR and LG:
1: It seems that static encounters have a 100% chance of being shiny even if you chose a different probability (in my case, 1/128).
2: I transferred the shiny Bulbasaur from one save file to another, and now PKHeX recognizes it as legitimate.

0001 ★ - BULBASAUR - DB58A75FA226.zip

[eightmouse]

Two other things I have to report in FR and LG: 1: It seems that static encounters have a 100% chance of being shiny even if you chose a different probability (in my case, 1/128). 2: I transferred the shiny Bulbasaur from one save file to another, and now PKHeX recognizes it as legitimate.

0001 ★ - BULBASAUR - DB58A75FA226.zip

Thanks A LOT, these findings are very valuable.

The second point makes sense: transfer can normalize some fields, so a Pokémon that looked questionable before may look clean afterward. But that doesn’t prove the original generation path was fully correct, so I’m not treating transfer as the fix.

The first point is more important:
if FR/LG static encounters are effectively coming out shiny 100% of the time regardless of the selected odds, that’s likely a real bug on my side in the static/gift path, not intended behavior.

Right now I’m tightening the starter-generation matrix first across the supported Gen 3 ROMs, and after that the next audit is wild/static/gift paths exactly for reports like this.

[JonahBernal]

Hi, so does this confirm that the patch doesn't work for emerald? I've been playing on 1/256 for about 5 hours. The encounter lag is there, but no shinies so far (which is not impossible with chance of course)

Thanks for making this by the way!

[eightmouse]

Hi, so does this confirm that the patch doesn't work for emerald? I've been playing on 1/256 for about 5 hours. The encounter lag is there, but no shinies so far (which is not impossible with chance of course)

Thanks for making this by the way!

Not exactly~

I wouldn’t say this confirms Emerald is broken, but I also wouldn’t say the current public 2.0.0 build is fully validated..
At 1/256, 5 hours with no shiny is definitely rough, but still possible by chance alone, like you said ~
I’m actively tightening these paths for the next release!

I'm updating the readme daily, so if you're interested in the progress just keep an eye on there ~

Thank YOU for the feedback!