Request & Question Thread

• Roflorz wrote on 2016-01-26 19:27

  Quote from Snowie Stormflower;1293223:

  Is it possible to get data on monsters by attacking them? Something like packets telling you how much your damage got reduced when you hit it?
  (Or any way to get data on monsters at all, I don't really know how to name these technical things, lol)

  
  
  Nope. Calculations are done and the resulting damage is sent back to you.
  If you want information on mobs, there's a database of all their stats, drops, etc in the leaked files.
• Snowie Stormflower wrote on 2016-01-26 19:30

  Quote from Roflorz;1293224:

  Nope. Calculations are done and the resulting damage is sent back to you.
  If you want information on mobs, there's a database of all their stats, drops, etc in the leaked files.

  
  
  Ah, so that explains where that one guy got those strangely accurate values from yesterday...
  
  I should look it up, thank you.
• Roflorz wrote on 2016-01-26 21:33

  Quote from Snowie Stormflower;1293226:

  Ah, so that explains where that one guy got those strangely accurate values from yesterday...
  
  I should look it up, thank you.

  
  
  Nah, he probably got it from packet logging or using the gm title client-sided.
• siraldric wrote on 2016-01-31 12:20
  Got another question regarding reforges that I couldn't find. I see that cylinders can get alchemy(earth,fire,wind,water) elemental reforges. Do these actually exist in game? I noticed it effects hydra damage so I thought it would be more known if you can get the reforge or not.
  
  For that matter is there any truth to it being easier to get a ele reforge on a ego then a normal one? Do the reforge tables change when a spirit weapon is involved?
  
  Any information on this would be amazing!
• Rydian wrote on 2016-02-01 03:31
  /data/db/itemmetalrealview.xml has the list of reforges that the client knows about. Ones with names like alchemy_earth do exist, but with just the info there I can't say what it may or may not go onto (chosen by the server) or the actual effects.
• Rynn wrote on 2016-02-01 03:49

  Quote from siraldric;1293486:

  For that matter is there any truth to it being easier to get a ele reforge on a ego then a normal one? Do the reforge tables change when a spirit weapon is involved?

  
  
  Whether it's easier or not no one can say, but the reforge pool does definitely change when a spirit weapon is involved.
  
  For example, ego 1h swords can get ele 2 at rank 3 while non-ego 1h swords can only get ele 1. Another example, spirit weapons cannot get "Max Attack Power" reforge while regular weapons can.
  
  And there's probably other changes in the table that go unnoticed. You can always ego a previously reforged normal weapon or spirit transfer a weapon that was reforged as an ego, so that's a loophole of sorts.
• LapisLazuli wrote on 2016-02-01 08:48
  Hi. Could I see the machine code for the client's rand() function? Thank you.
• X wrote on 2016-02-01 11:56

  Quote from LapisLazuli;1293507:

  Hi. Could I see the machine code for the client's rand() function? Thank you.

  
  
  No. The code would be very hard to obtain, very lengthy, and we probably couldn't post it here without getting yelled at.
  
  You could probably grab an old client that is not protected/virrualized (pre version 100) and disassemble that. Since you want to see the x86, I assume you know how to disassemble. Blade3575 should have old clients; you can find him on mabimods.
  
  However, before you go through all that, I can tell you that the client uses the mersenne twister for most of its rng needs, so the rand function is almost certainly a call to the twister. You can find implentations of the twister all over the internet, in a variety of languages. Maybe that's enough for you?
  
  Why do you want the code for rand anyway? I believe the only notable thing it affects is pet AI, which you have full control over anyway. "Important" numbers like damage and drops are calculated serverside, and thus unaffected by the client's rand function.
• LapisLazuli wrote on 2016-02-01 18:19

  Quote from X;1293508:

  No. The code would be very hard to obtain, very lengthy, and we probably couldn't post it here without getting yelled at.
  
  You could probably grab an old client that is not protected/virrualized (pre version 100) and disassemble that. Since you want to see the x86, I assume you know how to disassemble. Blade3575 should have old clients; you can find him on mabimods.
  
  However, before you go through all that, I can tell you that the client uses the mersenne twister for most of its rng needs, so the rand function is almost certainly a call to the twister. You can find implentations of the twister all over the internet, in a variety of languages. Maybe that's enough for you?
  
  Why do you want the code for rand anyway? I believe the only notable thing it affects is pet AI, which you have full control over anyway. "Important" numbers like damage and drops are calculated serverside, and thus unaffected by the client's rand function.

  
  
  Uh, couple reasons. Mostly I reasoned that for the blacksmithing minigame to be both entirely client side and have a random pixel offset, it would probably call rand() to decide what it was. If that was true, then theoretically manipulating it could result in better blacksmithing grades.
  
  I was also curious because I seem to fail at 99% alot more frequently than a 1% failure rate would suggest, so I figured if there was significant jitter in their rand() implementation then that might be the cause, and I figured the server might use a similar implementation since it was the same developers.
  
  There is also, of course, idle curiosity. Thanks for the reference to Blade, I'll get in touch.
• X wrote on 2016-02-01 19:35
  I don't mean to sound rude or anything, but I'd hate to see you waste your time wading through the client for stuff like that. (Ive wasted a lot of my time on similiar quests) Any numbers of importance are generated on the server. This includes both examples you gave.
  
  Rydian has investigated blacksmithing, and Aura has it implemented if you're curious how it works. The same goes with repairs. Ry has done statistical analysis on repair failure rates, showing some interesting trends that make it seem like you're failing way too much when you actually aren't. I wish I could find his posts...
  
  With regard to changing stuff, packet editing would be an easier way of influencing the outcome of minigames like cooking or blacksmithing, as opposed to changing the implementation of the RNG. However, packet editing has been (ab)used so much in the past that a lot of the game is hardened against it now. For example, ry has found evidence that even with a perfect cooking minigame result, the server injects its own randomness so you'll randomly get 4 stsr instead of 5.
  
  Mabinogi used to trust the client RNG way too much. Then some idiots decided to kick over everyone's sandcastles by packet editing a whole lot more nefarious things than minigames. Suffice it to say, the sever no longer trusts the client if it can at all be helped. So changing the RNG of the client won't do anything for you.
  
  If you're curious about the actual implementation, I'd suggest finding some C source for the mersenne twister and looking at that. It'll be easier to come by and easier to read.
• Rydian wrote on 2016-02-01 19:43
  Hell, in some cases nowadays (like dyes) the server plays a copy of the minigame with you and then ignores your result to use the one it got instead.
• kuronogami wrote on 2016-02-01 20:09

  Quote from X;1293515:

  I don't mean to sound rude or anything, but I'd hate to see you waste your time wading through the client for stuff like that. (Ive wasted a lot of my time on similiar quests) Any numbers of importance are generated on the server. This includes both examples you gave.
  
  Rydian has investigated blacksmithing, and Aura has it implemented if you're curious how it works. The same goes with repairs. Ry has done statistical analysis on repair failure rates, showing some interesting trends that make it seem like you're failing way too much when you actually aren't. I wish I could find his posts...
  
  With regard to changing stuff, packet editing would be an easier way of influencing the outcome of minigames like cooking or blacksmithing, as opposed to changing the implementation of the RNG. However, packet editing has been (ab)used so much in the past that a lot of the game is hardened against it now. For example, ry has found evidence that even with a perfect cooking minigame result, the server injects its own randomness so you'll randomly get 4 stsr instead of 5.
  
  Mabinogi used to trust the client RNG way too much. Then some idiots decided to kick over everyone's sandcastles by packet editing a whole lot more nefarious things than minigames. Suffice it to say, the sever no longer trusts the client if it can at all be helped. So changing the RNG of the client won't do anything for you.
  
  If you're curious about the actual implementation, I'd suggest finding some C source for the mersenne twister and looking at that. It'll be easier to come by and easier to read.

  
  
  I remember that repair post pretty well. I repair a big part of is was repairing per point vs repairing all at once and it concluded that repairing per point was best since repairing all at once is basically a repair per point except automatic so you would have a worst success rate should you fail.
  
  You would repair 4 points and it would look something like this. Half the points would be repaired at 98% instead of 99% since you failed the repair and lost the blessing.
  
  99 -> (repaired) -> 99% (failed) -> 98% (repaired) -> 98% (failed)
• X wrote on 2016-02-01 20:13
  Ahh, found it:
  
  http://mabination.com/threads/73694-Probability-And-You
• LapisLazuli wrote on 2016-02-01 20:33

  Quote from X;1293515:

  I don't mean to sound rude or anything, but I'd hate to see you waste your time wading through the client for stuff like that. (Ive wasted a lot of my time on similiar quests) Any numbers of importance are generated on the server. This includes both examples you gave.
  
  Rydian has investigated blacksmithing, and Aura has it implemented if you're curious how it works. The same goes with repairs. Ry has done statistical analysis on repair failure rates, showing some interesting trends that make it seem like you're failing way too much when you actually aren't. I wish I could find his posts...
  
  With regard to changing stuff, packet editing would be an easier way of influencing the outcome of minigames like cooking or blacksmithing, as opposed to changing the implementation of the RNG. However, packet editing has been (ab)used so much in the past that a lot of the game is hardened against it now. For example, ry has found evidence that even with a perfect cooking minigame result, the server injects its own randomness so you'll randomly get 4 stsr instead of 5.
  
  Mabinogi used to trust the client RNG way too much. Then some idiots decided to kick over everyone's sandcastles by packet editing a whole lot more nefarious things than minigames. Suffice it to say, the sever no longer trusts the client if it can at all be helped. So changing the RNG of the client won't do anything for you.
  
  If you're curious about the actual implementation, I'd suggest finding some C source for the mersenne twister and looking at that. It'll be easier to come by and easier to read.

  
  
  Oh, really, so it probably doesn't work that way? That's a shame, I was hoping I could fix something up for my blacksmith friend. Oh well, I guess I'll have to tell them no. :(
  
  

  Quote from Rydian;1293516:

  Hell, in some cases nowadays (like dyes) the server plays a copy of the minigame with you and then ignores your result to use the one it got instead.

  
  
  Really? My god that's ridiculous. Wow, people really did ruin everything.
  
  Regarding the probability stuff, yeah, I know how probabilities compound. I was more concerned with things like crafting at 99% on Sky lanterns failing 4 times out of forty crafts for me every other day, or how sometimes I get like three thames set drops in 10 minutes hunting ancients. Mabi luck always seemed very 'streaky' to me and that always struck me as odd.
• X wrote on 2016-02-01 21:07

  Quote from LapisLazuli;1293527:

  Oh, really, so it probably doesn't work that way? That's a shame, I was hoping I could fix something up for my blacksmith friend. Oh well, I guess I'll have to tell them no. :(
  
  
  
  Really? My god that's ridiculous. Wow, people really did ruin everything.
  
  Regarding the probability stuff, yeah, I know how probabilities compound. I was more concerned with things like crafting at 99% on Sky lanterns failing 4 times out of forty crafts for me every other day, or how sometimes I get like three thames set drops in 10 minutes hunting ancients. Mabi luck always seemed very 'streaky' to me and that always struck me as odd.

  
  
  Here's Aura's implementation of smithing, NA should be roughly analagous: https://github.com/aura-project/aura/blob/c7e0f89550a23d465ae16180ad84ad4441e3fa13/src/ChannelServer/Skills/Life/Blacksmithing.cs#L181
  
  A little clarification on what ry means by "play with you": For dyes, the server knows what palette the item uses, and what distortion parameters it gave you. So when you dye an item, it distorts the palette itself, then picks the colors at the coordinates you chose. This is in opposition to the client sending the color codes directly to the server, and protects against manipulation of the palettes. Otherwise, with a little bit of data-mod trickery, you could get any color you wanted, even flashy, with a regular dye amp with 100% success. Boy, you could make a lot of money doing that...
  
  The fossil minigame is a weird one. It's possible to craft a pixel-perfect restore, but sending that to the server STILL fails about half the time. My suspicion is the ones they can't play WITH you, they added an RNG to play AGAINST you.