Suggestion: Improved Glamour System

[Iscah]

He also said that about cross-server or PvP glamors, or being able to transmog a 1H wand/sceptor onto a two-hander staff/cane...

We still can't glamour a wand onto a cane.

And cross-server still has limitations on things that they said they wanted to do but couldn't, like making the marketboard fully server-wide.

Also, just because they worked out a solution to some things after initially thinking they couldn't, doesn't mean it's possible for them to solve every other thing they say can't be done. Especially when it's not just "we can't" but "we've really really tried and we're sorry we can't make it work".

[Shurrikhan]

just because they worked out a solution to some things after initially thinking they couldn't, doesn't mean it's possible for them to solve every other thing they say can't be done. Especially when it's not just "we can't" but "we've really really tried and we're sorry we can't make it work".

Sure, but it does run counter to the idea that "If Yoshi said so, it must be so." The "we really really tried, but it just doesn't seem feasible" applied to glamour in PvP, too, after all, and visiting other Data Centers.

[Visiting other data centers is] not the greatest example to support a point, though.

And yet something is being done for it. If it were to never arrive, sure, but it merely went from 'not possible' to taking longer than expected.

[Rongway]

visiting other Data Centers.

Not the greatest example to support a point, though. They scheduled that for 5.58 (this past July) and we probably won't have it until after February.

[DrWho2010]

housing data has nothing to do with attaching glamor log data to a character. the way character data is stored, transferred and checked on so much by the current system makes a glamor log impossible.

[Lihtleita]

housing data has nothing to do with attaching glamor log data to a character. the way character data is stored, transferred and checked on so much by the current system makes a glamor log impossible.

isnt that the same kind of check thats performed every time a house/retainer/saddlebag is used?

[Rongway]

isnt that the same kind of check thats performed every time a house/retainer/saddlebag is used?

The retainer database tables can only be accessed while you're at a bell. The saddlebag database tables can only be accessed when you're not in an instance. The internal housing database tables only when you're inside the house. They limit the places we can query the databases so that the servers aren't inundated with constant access requests, because access requests take server resources to look up the data and bandwidth to transmit the data. The server resources and the bandwidth are also things that the devs have an interest in minimizing, besides the storage space for the data.

[Lihtleita]

The retainer database tables can only be accessed while you're at a bell. The saddlebag database tables can only be accessed when you're not in an instance. The internal housing database tables only when you're inside the house. They limit the places we can query the databases so that the servers aren't inundated with constant access requests, because access requests take server resources to look up the data and bandwidth to transmit the data. The server resources and the bandwidth are also things that the devs have an interest in minimizing, besides the storage space for the data.

in this case wouldnt limiting acessing the glamour log to the glamour dresser be that smart move? thats already halfway what they did.

[Rongway]

in this case wouldnt limiting acessing the glamour log to the glamour dresser be that smart move? thats already halfway what they did.

I don't know exactly how the current glamour data is stored, but here are my best guesses as someone with a graduate degree in computer science.

There are almost 20,000 equipable items in the game. If you ignore belts, this number will go down; but let's assume the number of items will stay about the same or go up once we account for the items that are coming in Endwalker. That means we need log_2(20,000) --between 15 and 16-- bits to differentiate between all the different equipable items in the game, and another bit for HQ/NQ. So the glamour dresser needs 2 bytes per item to remember what the item is. The glamour dresser also remembers an item's dye. We have something like 120 dyes, and another byte can handle that. There is other information for items in general, but the dresser doesn't need to care about those things, and you're even warned that some data will be lost, such as condition and signature, when you put things in the dresser. So ignoring that information, we need 3 bytes per item in the dresser. A 400 item dresser then should require 1,200 bytes.

If instead we stored 1 bit for each equipable item we've ever obtained, the catalog would require 20,000/8 bytes. That comes out to 2,500 bytes, which is over twice the minimum data required to support a 400 item dresser. So if they're worried about storage space now, it could be much worse with a glamour catalog. There are of course data structure optimizations and compression algorithms that you could apply to reduce the storage required in either case, at the cost of increased processing requirements, but then we get back to how much processing would need to be done on many glamour catalog access requests over a period of time.

I know, 2.5 kB is not a lot. But with 24 million registered players, the total data required to store just glamour data would be about 60 GB.

And that's all before we start talking about how an increase in glamour data size would affect the data transfer volume (and cost), even with it restricted to dressers in inns.

[KhevanFarstryder]

There are almost 20,000 equipable items in the game. If you ignore belts, this number will go down; but let's assume the number of items will stay about the same or go up once we account for the items that are coming in Endwalker. That means we need log_2(20,000) --between 15 and 16-- bits to differentiate between all the different equipable items in the game, and another bit for HQ/NQ. So the glamour dresser needs 2 bytes per item to remember what the item is. The glamour dresser also remembers an item's dye. We have something like 120 dyes, and another byte can handle that. [...] So ignoring that information, we need 3 bytes per item in the dresser. A 400 item dresser then should require 1,200 bytes.

If instead we stored 1 byte for each equipable item we've ever obtained, the catalog would require 20,000/8 bytes. That comes out to 2,500 bytes, which is over twice the minimum data required to support a 400 item dresser. So if they're worried about storage space now, it could be much worse with a glamour catalog. There are of course data structure optimizations and compression algorithms that you could apply to reduce the storage required in either case, at the cost of increased processing requirements, but then we get back to how much processing would need to be done on many glamour catalog access requests over a period of time.

I realize that there are more items for which you would need to store information, but isn't there less data per item that would need to be stored? Since these are not items that could be restored, for example, you wouldn't need to keep track of quality (NQ/HQ), dyeable/undyeable (just replace the undyeable with the dyeable if you get it), maker's mark, etc. It would just be a boolean array for each character of 1 byte per item (2 bytes for items that have a dyeable/undyeable option - 1 byte for the true/false of having unlocked the item and 1 byte for the true/false of it being the dyeable version). Wouldn't need to keep track of pre-dye either, since I mentioned that a con to the original proposal would be that you couldn't pre-dye items (it would only be tracked on the glamour plates).

I know, 2.5 kB is not a lot. But with 24 million registered players, the total data required to store just glamour data would be about 60 GB.

Now if you took that and split it up per server (not all 24 million players are on one server and I mentioned in the original post that the database would be account wide so wouldn't need 1 database per character, unlike the current item dresser, just 1 per player), how much room would it take per server then? And if you limited access to it (couldn't access any of it, other than the glamour plates, if you do a cross server/datacenter visit or are outside an inn) would the server load be more feasible?

[Rongway]

I realize that there are more items for which you would need to store information, but isn't there less data per item that would need to be stored? Since these are not items that could be restored, for example, you wouldn't need to keep track of quality (NQ/HQ), dyeable/undyeable (just replace the undyeable with the dyeable if you get it), maker's mark, etc. It would just be a boolean array for each character of 1 byte per item (2 bytes for items that have a dyeable/undyeable option - 1 byte for the true/false of having unlocked the item and 1 byte for the true/false of it being the dyeable version). Wouldn't need to keep track of pre-dye either, since I mentioned that a con to the original proposal would be that you couldn't pre-dye items (it would only be tracked on the glamour plates).

One bit per item, and it wouldn't matter if it's dyable or not because a catalog system would only care if you've acquired the item or not. It would be up to the client to map the bits to items and tell you which ones are dyable.

At 1 bit per item, it comes out to twice as much data as currently required.

Now if you took that and split it up per server (not all 24 million players are on one server and I mentioned in the original post that the database would be account wide so wouldn't need 1 database per character, unlike the current item dresser, just 1 per player), how much room would it take per server then? And if you limited access to it (couldn't access any of it, other than the glamour plates, if you do a cross server/datacenter visit or are outside an inn) would the server load be more feasible?

I suspect it wouldn't make a difference if the people with multiple characters were allowed to consolidate their glamour data into a single catalog. There are probably far more people with single characters than with many alts, so account-wide glamour catalogs would only save so much space.

Besides, have you ever heard the XIV devs respond to requests for account-wide things? We may some day get a glamour catalog but I don't imagine for a second that they would make it account-wide, even if that could save them enough storage to allow a more accessible catalog.