xfr

I have the same problem, I used the memory viewer in VBA-M to record and save the Joybus stream but it cuts off pretty fast. It's pretty clear there are big sync/timing issues resulting is hangs/timeouts. The Dolphin source for GBA controller support is very messy clearly unstable (trying to change the synchronization always resulted in Dolphin all but freezing when a GBA is connected) but it's probably a safer bet than looking though the ISO. e: also if you looked at the junk ARM you can see a lot of checks related to the R/S save. From what I understand it is required to have a japanese Ruby or Sapphire save with the E4 beaten and National dex, and it must be saved inside a Pokemon Center. Possibly more restrications, also I doubt this works with FR/LG/E? or any non japanese save.

Correct, this does not prove the GBA celebi cannot be shiny. However despite a lot of effort I cannot have VBA-M recognized by Colosseum or the bonus dic using Joybus although it is supposed to be implemented both by VBA-M and Dolphin. It does react to the connection but eventually fails. So probably some more awful PPC and ARM ASM is required before we get a definitive answer.

I was wrong, the Ageto Pikachu is indeed a different method (not by much in fact) but the colo Celebi is actually common GC. I'll try to see if the Ageto Pikachu has the same shiny check but it is quite likely.

Ageto Celebi (written to Colosseum) cannot be shiny Although it is not yet possible to RNG abuse the Ageto disc on a retail console because of the unknown seeding parameters, it is however possible on Dolphin for the first Celebi that is written directly into the Colosseum save file. After selecting the Colosseum part of the disc the Colosseum RNG is seeded (memory location 0x80477098 as confirmed by Kaphotics/OmegaDonut/...). There's some advancement after the intro screen (about 100) then an non consistent number of frame advancements (more than a thousand, there are dialogs that wait for user input while the RNG keeps advancing which makes abusing much harder than the direct transfer to GBA). Then the function to create the Celebi is called at 0x8012d4f0 (it is the same for both methods of transfer). After two frames the IVs are generated (comon GC method). After a skipped frame the PID is generated by the function at 0x80121310. It has a recursive anti-shiny check as illustrated below: Since shiny Ageto is not possible here are two RNG abused HP Fire spreads: (Seed=7ddf12c8 PID=77e9f030, Nature=Bold) 31/14/31/30/31/30 http://www.pokecheck.org/?p=detail&uid=470139 (Seed=7c5f5fcc PID=ecbeb9d6, Nature=Modest) 31/2/31/30/31/30 http://www.pokecheck.org/?p=detail&uid=470141 Of course they should not be considered as RNG abused because they were obtained by altering the seed at the start of the disc. Note that because the VBA-M <> Dolphin trading does not work yet I had to write a program to extract and convert the Pokémon data from the Colosseum save to a 3gpkm file. If you have a GameCube and memory card you can download the Colosseum (US) save here: http://tmp.pokecheck.org/Colosseum_RNG_Celebi.zip and transfer the Celebi yourself.

They show up as hacked because they are hacked.

Fair enough, Taunt is "unfixed".

To be back on the topic of the RNG I would like to point out that it does not cover all the colo/XD Pokémon. For instance, all Poké Spots Pokémon on do not use this method. I tried looking at surrounding PID frames on the reverse RNG but did not find anything obvious (like a swapped/skipped frame). This probably deserves some investigation. Here are some examples: Also, to those who say the Ageto disc is figured out, they are forgetting that there are two kind of Celebi: the one written directly to the Colosseum save file and the 50 you can subsequently transfer to the game boy directly. Only the second uses the comon GC method. The Ageto Pikatchu also is not fully figured out yet.

Thanks, it's fixed

I tried to find our discussion about it but it seems I lost it (maybe I was on a laptop at that time). I'll try to ask you on IRC to fix that asap. There was a bug there, it should be fixed. Movesets for the Wormadam forms (other than plants) were wrong. It seems they are not in the veekun data for some reason. I hopefully fixed all (including TMs/tutors and checking there's no moves from other forms). Dragon Dance should be recognized, it's probably a problem with your Pokémon unless you post a link to confirm the problem. I fixed those. Not until I have a complete database of ALL past events. I discussed this with Sabresite, hopefully this will be done someday. It's probably pokesaved. Pursuit is an egg move and it doesn't have a egg location or date. The difference between Pokérus types is still unknown. I have some hypotheseses but never bothered checking if they were true. In other news support for Colosseum and Pokémon has been greatly improved. All XD/C moves should come up as valid now and thanks to the new research by OmegaDonut/Kaphotics/others (don't want to get into the ongoing argument about who did what) the PID/IVs are checked. The nature checks will be added whenever someone explains to me how they work (I'm rather tired of the Dolphin debugger that keeps crashing so I won't be looking into it myself).

It's not an Ageto Celebi because it has a type 1 PID. The GTS won't let you deposit any shiny Celebi because Nintendo doesn't consider it possible to exist legally (although it does, like shiny Manaphy or shiny Jirachi).

Although it might be useless/impossible to match the encounter slot with the PID (a shame, that would have been neat) the point is to at least ensure wild PIDs that are not (in-game) events (starters/fossils/gifts/in-game trades/legendaries) correspond to at least one plausible wild encounter. Please provide a link to the affected pokémon - water sport is on almost all latias from HGSS and it is properly recognized as far as I can tell.

This kind of problem won't be solve until encounter data is checked for previous gens, which is easier said than done since it requires an exhaustive database not only of all in-game events (starters, fossils, gifts, in-game trades... for all versions of the game since RSE) but also a database of all past events. Furthermore, this would be still prone to errors because the encounter location and level are both lost during transfer to gen 5 which is the main reason encounters are only checked for gen5. The Old Sea Map was never distributed outside Japan.

Thanks, I fixed the first two bytes. snorlax452 sent me a pokémon hatched on Korean Platinium, only D/P remains and we'll be done.

It works like this: - Start from the hatch trash byte string for your given version and language - Overwrite with default pokémon name - Overwrite with nickname For Japanese DPPt, the trash byte string is different for length 2, 3, and 4+, but it only depends on the length of the default name, not the nickname which is applied as usual

Thanks to helpful users (in this thread, snorlax452, on pokecheck) only spanish D/P and Italian DPHGSS remain, as well as all korean versions As for japanese DPPt it's possible to check the trash bytes depending on the length of the name (it's all FF for length=4,5 but the last bytes are seemingly random for length 2,3 - they are \0102\0000 for length 2)

Event exclusive moves show up as Hacked / Event only

Indeed, this PID is flagged as invalid because it is a gen 4 type 1 PID. You can argue it is possible for a pokémon to have an IV spread that matches a type 1 PID but the chance is 1 in 887,503,681 - this is a thousand times less likely that winning a state lottery. When Kaphotics says a very very very rare chance, it's really infinitesimal, you should be more afraid of being crushed by a meteorite than pokécheck being wrong.

It seems there are problems with many corean names. The names used in my database use unicode composition resulting in different unicode strings that represent the exact same name. It's very likely there is the same problem with other pokémon, just make a list of all the ones that show up wrong and I'll fix them. I know about the problem with some alternate form exclusive moves, that will be fixed soon. Edit: it is fixed now

Thank you for reporting this problem. It has been fixed.

You couldn't get the snarl zoroark at Japan Expo if you already had the one from GAME. It's the exact same wondercard.

I would very much like to see a list of the PID of the Mew on your save file, if possible in the order they appear in the boxes. This could help because - if a shiny seed is used on one of them but there's no shiny mews in the group (or it is skipped without an empty spot in the box) it could mean the generating program skipped the shiny mews. If there's a hole maybe the shiny mew was released or given away during the distribution. - having the PIDs in order of appearance can help understand how the RNG is reseeded with a restricted seed after each group. With this knowlege it is possible to theorize which restricted seeds can appear on a whole distribution card by knowing only 1 PID from that card (in perticular, if a shiny seed should have been used). Note that I have the impression Sabresite already knows how it is done but he hasn't been telling about it My impression given the scale of the event is that a few shiny Mew must have been generated for this event. I doubt they will resurface though, but we should be able to at least theorize their existence by proving at least one shiny seed must have been used on one of the distribution saves.

I ripped the french Snarl Zoroark because it was missing but it seems someone else submitted it first...

It has been brought to my attention that using an emulator (namely, Desmume) can yield different trash bytes than retail systems. Namely, I have been shown slot 1 pokémon from emerald and leafgreen that were pal parked on desmume (on a german platinium rom) that had trash bytes ending in \x6C\x1A\x0C\x02... instead of the usual \xA4\xA1\x0C\x02... on retail systems. Be wary of this when researching pal park trash bytes.

Here's a complete US Colosseum save. colosave_us.zip