On the usage of "upscaled"/"downscaled".

Currently the downscaled and upscaled tags are being used in opposite ways at the same time.

Since examples are better than walls of text:

posts with "resized"+"downscaled" and smaller source (this happens when the artist makes the version at the source smaller, which happens more often than one would think):
post #2777476

posts with "resized"+"downscaled" and larger source:
post #2876323
post #2759496

posts with "resized"+"upscaled" and smaller source:
post #2916620
post #2877972
post #2934134

posts with "upscaled+waifu2x":
post #2832559
post #2812855

Shouldn't downscaled_revision (plus bad_revision where relevant) be used instead in cases where the artist edits the source afterwards? In cases where the image on danbooru is smaller I'm not sure how it could be handled, as it may happen that the artist makes a larger file available at a later date - you can't really be sure whether it's an upscale or a better version, so perhaps just using md5_mismatch with no resized variants would be best.
It doesn't make much sense to have tags that are used at the same time in two completely opposite and contrasting ways.

Pinging @BrokenEagle98 as this discussion started in the Danbooru Discord, after I noticed his bot adding upscaled to some posts with valid danbooru versions but sources that were downscaled by the artists afterward. He referred me to topic #13591 but it seems his questions regarding this in that thread were completely ignored, so perhaps a new thread will get some definitive answers on the matter.

Yeah, I'm willing to change how I tag things, but I do want to tag things so that they're easier to find.

For reference, the only thing I have to base the tagging decision on is the height and width on Danbooru and at the source. The following is the current tagging scheme for MD5 mismatch in the order they are processed:

• height_source == height_danbooru and width_source == width_danbooru
• md5_mismatch
• height_source >= height_danbooru and width_source >= width_danbooru
• md5_mismatch, resized, downscaled
• height_source <= height_danbooru and width_source <= width_danbooru
• md5_mismatch, resized, upscaled
• (height_source > height_danbooru and width_source < width_danbooru) or (height_source < height_danbooru and width_source > width_danbooru)
• md5_mismatch, resized

Bump. upscaled currenly has a lot of posts that are not upscaled, they're just posts with the wrong source (often uploaded by commissioners, self-uploaders or just people who used the wrong source).

@evazion should those posts really have the upscaled tag? They're not resized, and upscaled implies resized, but they're being autotagged via bot because the danbooru picture is bigger than the source.

source:*twitter* upscaled is an example. Basically no post in there is really resized.

I agree that upscaled shouldn't be used for source size mismatches. I suggest making a resolution mismatch tag (if we really want to go beyond just tagging it md5 mismatch) and reserving resized / upscaled / downscaled for third-party edits.

Maybe the tags should be larger source and smaller source?

Currently they are source larger and source smaller.

Danielx21 said:

Maybe the tags should be larger source and smaller source?

Currently they are source larger and source smaller.

No, those can be confused with the other kind of source_* tags, i.e. they tell what KIND of source it is, e.g. non-web source, second-party source, third-party source, etc. Plus *_source tags mirror other similar meta tags that describe off-site information, like PDF available and PSD available, where they use the noun adjective format.

BrokenEagle98 said:

No, those can be confused with the other kind of source_* tags, i.e. they tell what KIND of source it is, e.g. non-web source. Plus *_source tags mirror other similar meta tags, like PDF available and PSD available, i.e. noun adjetive.

OK. I don't mind, that was just an idea.

Well, I don't really see a big difference between larger source (or source larger) and non-web source. All these cases tell you something about the source. (It could even be source non-web if we are going to use noun adjetive, but I'm not suggesting that.)

Danielx21 said:

OK. I don't mind, that was just an idea.

Well, I don't really see a big difference between larger source (or source larger) and non-web source. All these cases tell you something about the source. (It could even be source non-web if we are going to use noun adjetive, but I'm not suggesting that.)

To me, larger_source is like, larger than what? I don't know, just that it's larger. Conversely, source_larger is clearer, as it's saying that the "source is larger" than the current image. That's one thing about the English language, in that different orderings of words like that have nuanced alterations to the meaning.

As a native English speaker, neither of these two replies make a lick of sense.

BrokenEagle98 said:

No, those can be confused with the other kind of source_* tags, i.e. they tell what KIND of source it is, e.g. non-web source, second-party source, third-party source, etc. Plus *_source tags mirror other similar meta tags that describe off-site information, like PDF available and PSD available, where they use the noun adjective format.

And how exactly would they be confused? What would be the confusion? Source larger is itself a meta tag describing off-site information, so I don't see how it looking similar to the other source descriptor tags has any sort of negative effect whatsoever, nor do I see what possible confusion would even arise from it being called larger source in regards to the other *_source. You also used source_* and *_source in the wrong order.

PDF available works that way because "available" is telling you that one exists, it's describing the PDF's "state" rather than the PDF itself, i.e. "A PDF is available." Reversing the order to "Available PDF" would be describing the already known to exist subject, i.e. "Please download the available PDF." The usage of PDF available is not arbitrary, it only works that specific way for the purpose it's intended for.

BrokenEagle98 said:

To me, larger_source is like, larger than what? I don't know, just that it's larger. Conversely, source_larger is clearer, as it's saying that the "source is larger" than the current image. That's one thing about the English language, in that different orderings of words like that have nuanced alterations to the meaning.

Um, larger than the image being viewed on Danbooru? Are you being serious? Literally, what else could larger source possibly be referring to except the source the image being hosted on danbooru came from? If a tag is added to an image, it is talking about that image, that's why they exist, to describe and provide additional information about the image they've been applied to.

You're trying to argue that larger source and source larger somehow mean two completely different things, when they don't. Nuances exist, word order is important and can change the meaning of a statement, but this is not one of those situations. Larger source is grammatically accurate, as it is saying the hosted image has a "larger source" than itself. Yes, "The source is larger" is a grammatically proper sentence, but adjectives that are describing the subject go before the subject when only the subject and adjective are present, i.e. "Big rock! Sharp stick! Hot fire!" Source larger is grammatically awkward and sounds like caveman speech. "Rock big! Stick sharp! Fire hot!"

blindVigil said:

As a native English speaker, neither of these two replies make a lick of sense.

Duly noted.