Category talk:Item Icons

Item Icons

I guess all these 431 icons (as I write this comment) should not be appearing in the category itself but sub-categories? Moving one icon to Pocket Icons at least for now (I picked one randomly), so 430 to change eventually! :) --Goingbald 21:35, 19 August 2010 (EDT)

Auto-generating icons and icon tables

Currently, icons are named in the format "type NN (quality)-icon.png". There are good reasons for this, as many icons are generic and used for several different items. Some icons have more than the usual "qualities", for example if they are acquired from reputation vendors, so they don't really fit this format, but that's a minor problem. A bigger downside is that this makes using scripts to automate adding icons a lot harder: The assigned number according to this format is arbitrary and looking it up can't be automated easily, so manual work is almost always necessary when adding an icon. It's necessary to search the table icons that look the same, but have a differend background (and/or shadow/aura), which is prone to errors. The correct icon might be overlooked, or a similar, but different icon might be chosen.

Instead, I'd like to propose using the icon ID (as used by Lotro Companion, for example) as the icon file name. That way, every icon is uniquely identified: The order in which they appear on the icon table becomes irrelevant to their file name, and it's easy to programmatically check if a certain icon is available on the wiki. Uploading new icons by a script also becomes much easier. I don't believe there is a downside to changing the file name to the ID, except for it not looking as neat anymore: For users who don't know how the ID works, the file name might look like a random bunch of numbers. But finding a certain icon is done by looking through the icon table either way, and the file name that is used doesn't matter.

I'd also like to generate the icon tables automatically. The general format does not have to change for this, but ideally, every icon table would be replaced. This allows for easily adding new icons to the tables with little manual work, especially searching the tables for existing versions of the item becomes unnecessary. It's also much easier to be sure that the tables are not missing icons. For example: The table for necklace icons currently lists 250 icons. However, 290 necklace icons seem to be in use by the game. I uploaded a table containing all necklace icons on Discord. When someone (who is unfamiliar with icon IDs) is looking for a certain icon, the process to finding it remains the same as it is currently: Checking the table.

There are, however, some downsides to these proposals:

• Adding icons with the same name format becomes impossible for users who don't know how the IDs work. This sounds problematic at first, but I don't believe it is actually a problem. Uploading new icons and adding them to the table would be very easy to do with a script, so when new icons are added to the game, all of them can be added to the wiki by running the script. It should not be necessary for users to screenshot icons in the game, crop the images correctly, find the correct file name, upload it and add it to the corresponding table anymore at all.
  • When a game update with new icons is released, someone needs to run the script to upload the icons and update the tables. This is probably unproblematic. Short delays until someone finds the time may be possible, but if an icon is absolutely needed, it can still be uploaded manually the way it is now. On average, the time until new icons are added should greatly decrease, and less (or no) icons should be missed.
• For the icon IDs and the basic files needed to generate icons, we might become dependent on Lotro Companion and on it being maintained. I don't know if we have other ways to get this data. Should this happen, we could still continue using the tables generated up to that point and add new icons manually again. The only downside compared to the current situation would be inconsistent file names, but for the reasons explained above, I don't believe the file names actually matter a lot when not using scripts.
• Icon order in tables may change: Currently, similar, but different icons could potentially be listed close to each other. That might not be the case anymore when the ID is the only relevant factor. However, the certainty that a certain row contains all the icon variations it should have probably outweighs this by far.

The benefits:

• Icon Tables would be consistent and the required effort would be small. Icons with the same basic appearance could easily be assigned to a certain row, manually comparing them for little differences would be unnecessary. No more missing or duplicated icons.
• Icons (and items using them) could be uploaded very quickly after game updates. I'm mostly thinking about instance drops here, but having the items available on the wiki quickly is important to players running these instances. The Houses of Rest instance has been released around two months ago, Den of Pughlak and Assault on Dhurstrok around three months ago. Neither of these has all their items added yet, although the item stats and their drop information from new instances are among the most important information for players. Weeks or even months after release, this information is much less useful. This also applies to other new content, of course, but as the person who added most of the items from the new instances I can say that the prospect of having to search through the icon tables for every new icon was not a great incentive for me.

To summarize, I believe the ability to automatically upload icons greatly benefits the timely creation of new items, and less icons would be missed. Automatically generating icon tables allows them to be more reliable, as icons can be assigned to the same row based on objective properties (their ID).

I already uploaded several icons with the proposed naming, for example this one. If this proposal is rejected, they can be deleted without much effort. The name is comprised of three different IDs in this case, separated by a dash ("-"). Icons can have one to four of these IDs.

Making the proposed changes does not have to be done for all items at once, but automatically generating the tables requires the used icons to be uploaded with their ID as file name. Existing files would probably need to be renamed and pages using them would need to be changed accordingly. But as there are separate tables for each icon category, starting with a certain category (for example: Necklace Icons) would be possible to see how the whole process is working, and most of this can probably be automated.

I hope this is convincing, but feel free to voice your objections here or on Discord. Also feel free to voice your support, of course! --Fanchen (talk) 12:39, 15 February 2022 (UTC)

I will start my opinion from the bottom. I believe we should decide which system to use, and not mix the current one with the new one. Switching to the new system could be done with bot relatively fast.

However i think both, current and new system have their advantages. The new system makes it easy to maintain the icons with scripts but harder for common editors. If we have enough editors able to use these scripts it should guarantee that the icons table gets updated soon enough for them to use them.

The old names are descriptive, they tell you what is on the icon. In the old times it was a good practice to have images sensible named on web, in case they does not load. This is probably not an issue we need to worry about, but it might be still preferable for editors to have the icons named. In the current system it means chest and necklace icons will probably not get mixed up, and even if they does, it is clear from first sight.

I can also see some possible issues (not that they cant be solved) with autogenerating the tables. For example as you say there is 250 necklace icons on wiki and 290 in game - however some of the icons that are on wiki are not in game at the moment. The question would be if regenerating the table should keep these old icons that are no longer in use or not. Removing them might cause some items that were using it loose its icon. However the item is either using different icon, or is no longer in game itself. If it is no longer in game we might want to still keep it on wiki tho, as an obsolete item that once was in game - thus should probably have its icon preserved too.

Another issue is that there are icons that are same but have different id. Strictly speaking the wiki throws an error if you try to upload 2 same images, although it can be ignored and both images can exist. Question would be if we want these duplicate icons or merge the ids and have it only under one of them - maybe use file redirects.

About dependency on LotroCompanion. I personally wouldnt worry about that. Me and some other wiki editors have access to the code - which means we can generate the files, bigger issue would be changes by ssg, but i doubt that will happen.

Summary: I dont have a strong opinion about this. Uploading the icons under their id would make it a lot easier for me to make item pages etc with bot, but i understand if other editors have reasons against. In my opinion we should not mix these two systems. --Drono (talk) 13:55, 15 February 2022 (UTC)

I have not read every detail but understand the overall picture.

• I am in favour of automated updating and adding where it can be done
• I believe new editors to Lotro-Wiki must have a chance too

That said, if there is a way to automate handling of icons and that automation also generates tables or alike that is useful for new editors, or old editors such as myself but who do not edit many items, then my vote is cast already.

But if we are automating icons, why not also item stats? — Zimoon 14:02, 15 February 2022 (UTC)

That's a good point actually. The reason we still manually code stats rather than just calling CalcStat on items primarily is that we don't want an overreliance on a proprietary system maintained by a single person or small group of people. Manually entering stats on items is more work, but it's work that literally anyone can do, so it's highly flexible and resilient. I would say the same thing goes for icon/image titles. Overreliance on a coding system that requires a third-party tool to identify which ID belongs to which icon is brittle, versus just having our own naming scheme which is intuitive and easily human-readable. It's true that it cuts down on the amount of automation we can do but at the end of the day I think resiliency, transparency, and ease of understanding for (especially new) editors is of primary importance. --Gaerlin (talk) 16:11, 22 February 2022 (UTC)

I agree that we shouldn't become dependant on proprietary tools, but I'm not sure that icon naming is comparable to CalcStat. As far as I know, the IDs of the icons are used by the game as well, so even if Lotro Companion should become completely unavailable for some reason, they would still make sense. There is no tool that maps the ID to icons, but the icon is literally defined by the IDs.

But there's another, less abstract reason to automate icon creation: Manually adding the icons (and items) takes a lot of time. Take Assault_on_Dhúrstrok as an example: Most (or all, I'm not sure) of the equipment items was added by me recently, and that instance has existed for more than three months. I haven't added items with a non-existing icon, and no one else has either. And with more time passing since the release, this information becomes less useful anyway.

From my perspective, when manually searching for icons, it doesn't matter how they are named. The table is necessary either way. Automatic icon uploads should mean that manually uploading won't be necessary anymore, but even if there is a reason to do it, there's nothing preventing someone from uploading an icon and adding it to the table.--Fanchen (talk) 17:54, 22 February 2022 (UTC)

I was typing a very long comment but will make just TLDR;

• item and icon ids can be found trough lua api
• some templates/files are already restricted from editors
• it might sound difficult but it is actually easier to use and maintain

There will always be room for human edits, even if we botted everything available. --Drono (talk) 18:04, 22 February 2022 (UTC)

PS: Seeing Fanchen's example look at The_Court_of_Seregost thats a 4 year old instance with incomplete loot. --Drono (talk) 18:06, 22 February 2022 (UTC)

I would be against this proposed new system for icons, for many of the reasons already mentioned above by Drono and Gaerlin. And to that I would also like to add another scenario that hasn't already been considered I think:

As someone who did a lot of work with icons and things related to that (e.g. items, skills and traits) for the (then) upcoming additions of the Brawler class and the new LI system during their beta/previews – in order to have as much as possible of this information already in-place on this wiki, and available to users, when that content was launched – I can see that being a problem as well, with this proposed new system.

As I understand it; we would first have to wait until the content is actually released, then wait for someone (or a select few?) with the required capabilities to pull these IDs from the game data, then wait until the files have been actually uploaded with these new names (which, in my opinion, are very convoluted, unintuitive and much harder to work with — concerns that have been touched upon by others above), before we could actually start working on all the stuff — which might of course take a lot of time from then onwards. Whereas with the existing system, we can do all (or most of) that work before the content is actually released! And what's more; we can share that workload over a (much) greater number of people, since taking screenshots, cropping icons from them and uploading those with simple and intuitive names in the current naming scheme(s), is easy enough that pretty much anyone can do it! :D

As I see it, this proposed new system trades away ease of user editing to gain ease of scripting; making it much harder for human input and contribution, in order to gain the ability to run some automation on uploading icons for those select few who have the capabilities to do so. But the thing is: that would be a very rare (or at least really infrequent) occurrence compared with how often human editors would interact with those icons.

I don't think that this is the right direction for us to go, as a wiki. In my opinion, we should always strive to get MORE (new) editors in by making it easier for them to contribute, as much as possible! Not make things needlessly convoluted or otherwise overly complicated, thus raising the bar further on making contributions, by requiring extra 'advanced knowledge' that may be off-putting to people in general. Then we would only be hurting ourselves in the end, I think. Since I feel that this proposed new system would hurt far more than it would help, it gets a NO from me. I belive the existing system is a better way to go.

--Stargazer (talk) 00:07, 8 March 2022 (UTC)

First of all, thanks for your input! I'm not sure if you've already seen (I don't think you use the Discord chat?), but I've uploaded all necklace icons with the ID naming system a short while ago, and I've replaced the table in Category:Necklace_Icons accordingly. I deleted no files, so in the worst case, everything can be reverted to an earlier state. But this allows everyone to see how this would change things.

About your objections: I don't see why this would make it necessary to wait for anything before editing. This does not mean that pages shouldn't be modified manually anymore or anything like that, it would only concern using icons. Assuming a new item with a new icon is on Bullroarer and someone wants to create a wiki page for it: Just create it in exactly the way it was always done. The icon ID can be found by a plugin, if that is wanted, or from the game data. Alternatives are not uploading the icon immediately, or uploading it the way it is done currently, but with a preliminary name. I'd expect that to be less effort than searching for the correct icon in the current tables (in case the icon might not be new). I don't know of any possible work that would be blocked by waiting for an icon upload.

You say that the current naming scheme is simple and intuitive, and that many people can work on it together. I disagree on the first part; while it may be simple, it's not intuitive at all. Rather, it's completely arbitrary. For the icon type, you're right, but the number is needed, and the number depends on when the icon was added to the wiki. If you don't know that, you have to scan the complete icon table. This is also where mistakes happen: I'm sure there are more, but I noticed that Shield icons use a single table for all shield types, although Warden's Shields are sorted into the same table -- but at the end of it, with a different naming scheme. Heavy Shields are just named Shields. And at least one icon only used by Warden's Shields is sorted into the usual Shields category. In some cases, Light Armour uses an icon that is sorted into the Medium Armour table, etc. It's not surprising that this happens from time to time, and it's certainly not the reason to abandon the current system, but I feel fine saying that this system is not simple and intuitive. I'd rather say: It kinda works.

Many people can work on it together: Well, yes, but actually many people have to work on it together. Taking screenshots, cropping them, finding the correct name and setting it as the icon for the relevant items takes a lot of time. Drono mentioned the Court of Seregost having incomplete loot, and the new 3-man instances I mentioned aren't complete yet either. So, even if it may be true that the current system might allow more users to add icons, they don't. That might be different for other pages where icons are needed, but for instance gear, manual uploading is just too slow. Players who don't run instances don't care about it anyway, and for players who do, the wiki doesn't provide useful information.

It's true that uploading icons would become a rare/infrequent event, as that would only be necessary after a game update. Ideally, no-one else would need to upload icons. I don't see a problem with that, it seems like a good thing to me. Using icons should work the same way as before, it should be neither more nor less complicated.

I agree that hurdles for new editors should be low. I don't agree that screenshotting, cropping, naming and uploading icons is a useful activity for new (or experienced) users. I also don't agree that having many users doing manual edits is what we should strive for: Rather, we should try to have as much accurate, up-to-date information as possible. There are many pages where it's good to have more players willing to add information to the wiki, like the Walk-through section of the new raid, which has seen quite a bit of activity in my opinion. Not providing readers with information about item rewards, because they get the chance to add it themselves if we don't, is not what I would call helpful.

Hopefully, this eases some of your concerns -- please let me know what you think.--Fanchen (talk) 01:20, 8 March 2022 (UTC)

Ad mixed tables - there might be some errors, i think there were also some historic cases when icon was switched from one type to others, there are some cases where all armour types use same leggins icon etc. Unless the new tables have all types in one big table - there will probably still be stuff like that. Btw. seeing the tables i actually dont see much value in them. Either you know the icon ID in which case you dont need the table, or you need to visually find the icon in the table, which you dont need the table for, as it doesn't provide much ease for finding - its actually quite sparse compared to the list of files at bottom - which is ordered by names - while the item part of the icon is first, meaning 'same' icons are next to each other in the list.

Ad Court of Serogost - at that time there was less active editors than now, later it is probably not priority for many editors which means they focus on something else - so it wasn't added to this day.

Editors vs Botting - as someone that can write scripts, its a bit painful to see the wiki is lacking information that could be with some changes easily botted. I agree with Fanchen that we should not strive for having many editors doing manual work. In contrary, the point of wiki is to provide accurate information, the number of editors shouldn't matter. The reader when reading an article is not asking and not even thinking about how many editors contributed to the article, they are interested only in the information and they (or at least me, both as reader/editor) get frustrated when its missing. That being said, botting does not remove the need of human editors, huge part of the wiki is still the human contributions. If people were interested solely in the data they would be using LotroCompanion, as it does even more and it does stuff wiki will never do. For example with items, even if they were botted with pretty much all information, there is room for human editors to extend the item page with images. And items are one of the pages that are mostly factual only, there is even more room for human edits on others - eg walktrough for quests/deeds/instances.
I understand that changing the names might raise the bar for newer editors and i don't like that aspect of the change either. However the bigger picture goal is that editors wouldn't need to do this stuff manually. I also put my scripts when finished on github, so other editors could use them. --Drono (talk) 08:12, 8 March 2022 (UTC)

I do a lot of bulk icon editing and uploading. It's really not very time consuming for me, I just don't always have the time to do it. I understand the frustration from others when the icons aren't already on the wiki or uploaded where they'd expect. Icon naming can be a hindrance so this proposal is intriguing but I've been skeptical about its overall utility. However, I'm not sure I comprehend exactly what's being proposed.

As I understand it, icon tables on category pages would be automatically generated and sorted, more or less, by when the icons were added to the game. When item pages are created, the editor will still need to look up the icon in the table. If this is the case, I don't see much benefit over the current naming scheme. We're still looking for an icon based on visual appearance. The real game changer in terms of saving time and standardization comes when we can automatically generate item pages. Is page generation part of this proposal or does that come later? I don't think we should be changing the icon naming scheme until we're ready for page generation.

Other concerns:

1. What happens when an icon gets an updated appearance? What happens when an item gets a different icon? Are there checks for these or would they still have to be detected manually?
2. Does this proposal extend to all item icons (e.g., decorations, potions, quest items) or just gear? For items with a unique icon (most decorations) the icon name matches the page name. We wouldn't be able to copy and paste the item tooltip template as easily if we change the naming scheme.
3. Does this proposal extend to skill icons? Mount icons often have names that correspond to their skill icons and war-steed appearances often share the same icons for item and skill.

Dadwhereismom (talk) 10:54, 9 March 2022 (UTC)

The motivation behind is for item generation, in my opinion there are two big challenges at the moment - icon names, and duplicate item names. However the icon challenge concerns all item creation, so with it at least unique named items would be able to be generated. Duplicate names can be also somewhat or semi automatically solved.

1. When icon appearance is updated, new icon is uploaded over the old one so all items will use the new icon. It is possible to make script for checking if item has changed, than just run it after game update.
2. In my opinion this would be only for gear icons. The item template is not changing, only thing that changes is the value you enter into the icon field. It does not matter to the template if it is bunch of numbers or human readable string.
3. See 2), but if desired it would be possible to extend it in future to other/all icons.

--Drono (talk) 11:21, 9 March 2022 (UTC)

With what I have understood from all discussions that have been going on, both here and on Discord, I think this looks good. There are concerns, yes, but as long as we know the limits and work together I am positive, FWIW. — Zimoon 13:22, 9 March 2022 (UTC)

Late to the party here, but I have mixed feelings on this. Mostly I also dislike the dependence on a third-party database and script. Also I was going to upload an icon and I really can't process the steps I need to take. It is past my bedtime, so maybe tomorrow I'll figure it out, or maybe I'll just wait for someone else to upload it. That seems like a problem to me, though. *shrug* If it's harder for uploaders but saves a lot more time for item creation/editing, then I suppose it's worth it. Also, screenshotting and cropping icons is really not that time-consuming. I made a couple animated gifs of ambient decorations the other day, and /that/ is time-consuming. --Ruby (talk)

I would propose removing the Style column. It makes the diffs harder to read and does not serve any purpose, especially if it changes between versions. --Drono (talk) 14:17, 3 April 2022 (UTC)

I agree. I don't see a reason to keep the column, it's basically meaningless now. If there are no objections, I'll remove it with the next table update.--Fanchen (talk) 18:04, 3 April 2022 (UTC)