Excessive post of same character

Posted under General

If a post limit does become necessary, I think something in the 150-200 range seems fair. 100 posts are exceeded often enough that I think it would be too low. I looked at the most prolific posters using the DanbooruInsights script and not many users regularly crossed the 150 mark. As Talulah mentioned, having the limit too high would accomplish nothing, so I'm not sure it could be too much higher and still be effective. Also, a technical question, would "upload days" be separated by a common threshold like UTC, or would it be based on a user's time zone setting, or perhaps something like post age?

The problem is the spam of the same picture, not contributors uploading many different posts. We're trying to fix a migraine by shooting ourselves in the head here.

Adding a general hard limit makes no sense if the problem is well defined. 100 posts of the same character with just different expressions are gonna ruin people's experience of the site no matter at what time they happen, and a per-user limit does not fix the fact that it's not a single user dumping this stuff, but several of them.

We add a hard limit and then 8 people decide to dump 100 tachi-e each on the same day, and we're back at square one with the only meaningful result being that now users who upload a lot of different posts are limited for no reason.

Maybe instead of trying to shoot ourselves in the head, as you said, by applying a limit, perhaps we could instead start having modqueue strongly discourage too many variations of the same 1girl solo standing tachi-es, or even make it flaggable?

What I'm suggesting is to make these kinds of large variant sets against the rules, basically. Which probably has many cons I haven't thought of.

WRS said in forum #439783:

In my opinion, I think it's best to just leave today at only asking to interval your uploads if you're planning to mass dump hundreds or thousands of similar images like this, if you even somehow manage to find a case where that can happen besides very dynamic VNs.

That sounds even worse. At least when they're uploaded all at once I can scroll past them all at once in searches. If they're intervaled in upload it means there's actual posts I want to see hidden between them.

Uploaders feeling like they have to upload complete sets for archival purposes makes it worse as a site for browsing images.

First, let me say in general I don't have a problem with tachi-e's or large variant sets. I like browsing official art, so I like seeing them on the site. But this is a two-pronged problem: it has been a long time coming and it's one of those things that seem to be obvious to some but not obvious to others.

#1 basically can't be solved unless someone here invents a time machine. Someone should have directly confronted anyone who was doing this to farm approvals about it instead of making passive aggressive forum posts and gripes on Discord, ideally through a feedback or mod-sanctioned (or mod-sent?) dmail. Even back then evazion was saying "Uploaders realized it was an easy way to game the system to farm uploads." (forum #377788) so we (general we) were aware it was a problem but it wasn't seen as a serious problem until now. Lemmings follow lemmings and if Adam does it, Bob Curly and Daniel are going to say "But Adam did it!" when you confront them about it. Perhaps that's a lesson for us (general us) to learn for next time.

My suggestion for #2 is to add a line to help:upload rules or something about it. I don't want them to actually be against the rules unlike Knowledge Seeker because I actually like them. But I think it would be helpful for them to be categorized in the same way 3D, cosplay, or other borderline content that faces difficulty in modqueue would. That would prevent people from seeing them as a quick way to get Contributor, or something a good user who has already gained Contributor+ would do.

Not sure what can be done but it needs to be stopped or limited. As I mentioned it ruins my browsing experience to have to look at and scroll through two or more pages just to get to see a few images I might like. I have been browsing here for years and consider this my come to spot for its content. But lately just looking for new post has become tedious and causing me to drop my interest. Maybe I am being over sensitive about it, I just dont like it and see it as spam for the sole reason of numbers boosting. I cant imagine any other reason someone would have all these images and not only like them but save them. As I said pick one image of the series that the uploader likes and upload it. Ok I'll be nice maybe five images. But 50 plus images where the only difference is a small expression change just screams spam in my eyes. I would never even think about saving them all no matter how much I liked the game. This is akin to someone flipping an image and calling it a new image but done 50 times.

Banning a type of frequently uploaded post seems like it would encourage a great flagging war. Even if you grandfathered in old posts I imagine there would be much conflict.

As far as blacklisting goes there's no real way I know of to search for "only the first part of a LVS pool". You can sort of do it with "-large_variant_set -parent:any" if it's parented but for pools it seems like an issue.

Blacklisting also just shortens pages, if it extended the page instead and you could blacklist 'only second+ posts of LVS' I think this would become a total non-issue from a browsing perspective.

I have an idea. Instead of uploading each variant as an individual post, why don't we encourage stitching them together instead? For example, if each character has 9 variations, they could be arranged in a 3x3 grid and uploaded as a single post. This would let users upload more images without flooding tags and the front page as much. This should be easy to do if all the images are the same size, which would also make it easy for users to separate the images cleanly if they want.

This might not be enough for the extremely huge sets, but it can still help a little.

Tenshikuro said in forum #439813:

As I mentioned it ruins my browsing experience to have to look at and scroll through two or more pages just to get to see a few images I might like.

Try adding -tachi-e to your searches (not your blacklist). That should keep most of them from appearing at all and you'll still have 5 tags to work with.

Obviously, this won't be as helpful for blue users since that only leaves them with 1 tag.

I think the best way to go is to warn offending users whenever this happens. From my understanding the last time this took place was months ago and unless this is a common occurrence every month it's not something that is worth spending time trying to fix. Ideas such as limiting the amount of daily posts are completely unwarranted in my opinion and blacklisting as a way of dealing with it is the way to go instead of complaining about it.

Blank_User said in forum #439848:

I have an idea. Instead of uploading each variant as an individual post, why don't we encourage stitching them together instead?

Would break hash checking and cause the images to be encoded twice (we want to avoid because every lossy encode is a quality loss). Due to the order of the images changing between uploads the duplicate system will have a hard time finding duplicates.

Tenshikuro said in forum #439813:

But lately just looking for new post has become tedious and causing me to drop my interest. Maybe I am being over sensitive about it, I just dont like it and see it as spam for the sole reason of numbers boosting. I cant imagine any other reason someone would have all these images and not only like them but save them.

Please don't take this the wrong way but In my opinion you are indeed you are being a bit over sensitive. Uploads like this are not just for boosting numbers but also for archival purposes. I believe it is wrong to assume bad faith when there are better explanations out there. I, as an example, often self-datamine VNs and games myself and save them on my computer but I'd always like to keep them on a remote location to save myself space and have peace of mind. A small script that takes a local Hydrus and uploads them to Danbooru would effectively achieve the same thing.

If there is to be done, in my opinion there could only be a small warning banner for users who achieve massive uploads in short times and some motivation to space out uploads of these kinds of images. But once again, I think it is best to deal with this yourself and shoot the person a dmail about their uploads whenever it happens.

tamuraakemi said in forum #439814:

Banning a type of frequently uploaded post seems like it would encourage a great flagging war. Even if you grandfathered in old posts I imagine there would be much conflict.

Blacklisting also just shortens pages, if it extended the page instead and you could blacklist 'only second+ posts of LVS' I think this would become a total non-issue from a browsing perspective.

I agree with these though the latter sounds like a hell to implement. It would be quite nice to have though.

ggraphy said in forum #439878:

Please don't take this the wrong way but In my opinion you are indeed you are being a bit over sensitive. Uploads like this are not just for boosting numbers but also for archival purposes. I believe it is wrong to assume bad faith when there are better explanations out there. I, as an example, often self-datamine VNs and games myself and save them on my computer but I'd always like to keep them on a remote location to save myself space and have peace of mind. A small script that takes a local Hydrus and uploads them to Danbooru would effectively achieve the same thing.

This may be true, but also… Danbooru is just really not designed to be an archive for large variant sets. There's a reason I didn't upload all 81 variants of post #9102928 as posts. I have all of them as assets on here (you can see the list in comment #2527898) but they really do not need to be individual posts. In cases like this I'd say simply archive them somewhere else.

Ylimegirl said in forum #439880:

This may be true, but also… Danbooru is just really not designed to be an archive for large variant sets. There's a reason I didn't upload all 81 variants of post #9102928 as posts. I have all of them as assets on here (you can see the list in comment #2527898) but they really do not need to be individual posts. In cases like this I'd say simply archive them somewhere else.

The way you have kept and listed them as assets under makes a lot of sense! I sometimes forget that Danbooru deletions are soft deletions and if variants are still accessible in this manner I think it's a very good way of handling it. It keeps the archive intact without making it look like it's to boost upload numbers. This can be the norm for handling similar cases IMO

Would it be feasible to code a new third type of pool for variant sets, that makes only one post from the pool appear in searches? Say you're searching smile, the only post from the pool you see on the search results page could be determined by:

  • Pick the smile post with the highest id
  • Pick the smile post closest to the beginning of the pool
  • Pick a random smile post from the pool
  • Maybe some other way I didn't think of

And then have some visual indication that the post represents a pool, like a "+50 more" over the thumbnail. Since a bunch of facial expression variations don't need to be put in any particular order, they don't really need to continue to be "series" pools. But I think that'd necessitate breaking down a pool like pool #29931 for example into smaller pools for each base pose/outfit. Also you'd probably need to create a metatag or setting for enabling the display of all posts from these variant set pools in searches.

Well I've stated my case. Seems I am being over sensitive so I will bow out. Was not aware being able to datamine these images from a game or VN you already own did not count as being archived. Furthermore I was not aware that this site was for archiving. Pretty sure a cloud based alternative would serve better than taking up space and bandwidth here for personal archives.

fuzzbawlz said in forum #439882:

Would it be feasible to code a new third type of pool for variant sets, that makes only one post from the pool appear in searches? Say you're searching smile, the only post from the pool you see on the search results page could be determined by:

  • Pick the smile post with the highest id
  • Pick the smile post closest to the beginning of the pool
  • Pick a random smile post from the pool
  • Maybe some other way I didn't think of

And then have some visual indication that the post represents a pool, like a "+50 more" over the thumbnail. Since a bunch of facial expression variations don't need to be put in any particular order, they don't really need to continue to be "series" pools. But I think that'd necessitate breaking down a pool like pool #29931 for example into smaller pools for each base pose/outfit. Also you'd probably need to create a metatag or setting for enabling the display of all posts from these variant set pools in searches.

I actually suggested something similar in Discord as we have related issues with comics and duplicates/"1ups" which will still be an issue even if nobody ever uploaded a large variant set again. Ideally we could just have a solution that addresses all three at once, so

ggraphy said in forum #439878:

[...] unless this is a common occurrence every month it's not something that is worth spending time trying to fix.

just seems like a bit of a cop-out. The problem is, of course, being able to implement it in a way that doesn't destroy the site's performance.

Tenshikuro said in forum #439902:

Well I've stated my case. Seems I am being over sensitive so I will bow out. Was not aware being able to datamine these images from a game or VN you already own did not count as being archived. Furthermore I was not aware that this site was for archiving. Pretty sure a cloud based alternative would serve better than taking up space and bandwidth here for personal archives.

It's more just that the average builder doesn't actually browse the site so things like this that legitimately make the experience worse for those who do don't even register as issues to them. Given this is not even the first thread on this topic, it's not like you're the only one who finds it annoying.

ggraphy said in forum #439878:

Would break hash checking and cause the images to be encoded twice (we want to avoid because every lossy encode is a quality loss). Due to the order of the images changing between uploads the duplicate system will have a hard time finding duplicates.

There won't be any loss in quality if the correct method is used. All you need to do is copy the pixel data to a new image with a lossless format.

As for the hash checking, that is certainly a valid issue with Internet sources, but what if we only limited the stitching method to self-datamined game assets? For a lot of these, there's a good chance they will be made from compositing layers anyway (topic #35073, topic #36052).
Those images would already be altered from the original, so taking the extra step of stitching them won't really affect much (though admittedly, that doesn't solve the issue with duplicates).

That said, I support fuzzbawlz's pool idea. If it's feasible, we shouldn't need to stitch anything in the first place.

Tenshikuro said in forum #439902:

Furthermore I was not aware that this site was for archiving.

This is a reposting site. Why would you think it's not for archiving?

I'm not saying people should be using Danbooru as their personal storage, but I can see why users may think it's worth sharing these, especially when access to some of these games isn't guaranteed in the future (gacha, etc.).

Tenshikuro said in forum #439902:

Was not aware being able to datamine these images from a game or VN you already own did not count as being archived. Furthermore I was not aware that this site was for archiving. Pretty sure a cloud based alternative would serve better than taking up space and bandwidth here for personal archives.

I understand your frustration but at the same time consider the amount of works that gets deleted that is still stored in Danbooru, the goal here is not to just archive but to archive the "good" content so people can access them in the future even if the artist has disappeared out of thin air, or else we could just keep the links of works and only store those links to avoid the storage costs completely, embedding the image from their sources, why not? You also have to consider that Danbooru doesn't even delete content, and keeps the content even if they are deleted available for view after disabling some filters. Tell me if this is archiving or something else?

Talulah said in forum #439914:

just seems like a bit of a cop-out. The problem is, of course, being able to implement it in a way that doesn't destroy the site's performance.

If you would refer to forum #439881 and the very bottom of forum #439878 you would see that I'm not completely against this but at the same time a site-wide change for something like this is not a small undertaking. If something is not worth spending time and effort over or provides minimal improvement, I believe it's not worth the additional burden. Once again, I am all for a good solution, simply to put, most of the current suggestions are not good, that's all.

We've mostly been talking about the really large sets, but when exactly would uploading a set reach the point of spamming? How many expression variations would be too much? How many expression/costume combinations? And when would it get to the point that not even uploading a few at a time would help?

LightSolas said in forum #439758:

Default blacklisting large variant set is not a solution. 100 tachi-es extracted from a game is not the same as a 10-post CG set with a plot.

Alright, make very large variant set then, though also, I personally think that the 6 images required for large variant set is too small. I use the largest thumbnail size and I'm not bothered by seeing 6 images in a row but I sure am bothered by seeing 50. Or plotless varient set though I like that idea less. Seems like the default blacklist idea is the best one because it doesn't punish people for archiving these images while also not punishing average users. Also, tangential but I agree that uploading 1000 of the same image probably shouldn't be upgrade worthy.

Blank_User said in forum #439946:

We've mostly been talking about the really large sets, but when exactly would uploading a set reach the point of spamming? How many expression variations would be too much? How many expression/costume combinations? And when would it get to the point that not even uploading a few at a time would help?

"I shall not today attempt further to define the kinds of material I understand to be embraced within that shorthand description, and perhaps I could never succeed in intelligibly doing so. But I know it when I see it."

Not every fuzzy concept is clearly definable, but most people exercising common sense know when something is a problem. Just like uploading "good art" is subjective, but most people here have a consensus on what meets minimum standards.

1 2 3 4 5