AMV Content Descriptors

After more than 20 years of looking like this, the site is planning a major rebuild! We need your feedback!!
Post Reply
User avatar
Phade
Site Admin
Joined: Fri Oct 20, 2000 10:49 pm
Location: Little cabin in the woods...
Org Profile

AMV Content Descriptors

Post by Phade » Sun Aug 03, 2025 4:13 pm

Hey Everyone!

The original plan for AMV viewer content communication was to have three variations of content notifications: a single-value indicator (like G, PG, PG13, etc.), a list of specific standardized content descriptors (blood, nudity, violence, etc.), and a specific warnings pool as needed (free-text words or phrases specifically meant as a warning). While the original proposal for the first single-value indicator was (putting it lightly) somewhat controversial (and still in flux), the second indicator of content descriptors seems to be more desired and anticipated.

This second system is meant to provide standardized content descriptors that most people would want to be informed about in the AMV, so that the viewer could include/exclude related AMVs from search results, or at a minimum inform the potential viewer that they may need to read the specific warnings or full AMV description carefully before proceeding.

One longstanding and fairly well-known content description standard is the Entertainment Software Ratings Board (ESRB) ratings guide. This guide has eight categories of thirty descriptors that are each weighed using frequency, intensity, and severity. Using the ESRB guide as a guide, we have come up with a list of five categories and ten total subcategory types that should cover the vast majority of our needs here.

The system will work by displaying a checkbox list of the ten types with the option of a “none, low, medium, high” severity selector for that type. If a type severity greater than “none” is selected for that content type, two additional qualifiers of “frequency” (how often the type occurs in the AMV) and “intensity” of that type become available with similar “low, medium, high” selectors.

Let’s say your video is fine except for “Alcohol” content, you would check “None” for all other types, click the corresponding “low, medium, high” severity indicator, select the qualifier value for “Frequency” (one or few, several, frequent) and “Intensity” (hinted or glancing, clearly depicted, intense / dramatic).

For example, the AMV “Trigun - Beer” by Hakura of Otaku Productions would absolutely have an “Alcohol” indication of “high severity, high frequency, and high intensity”. 😁

So without further ado, here are the ten proposed required standardized content descriptors based on the ESRB and our own specific needs.
  • Blood/Gore
  • Violence
  • Alcohol
  • Drugs
  • Humor
  • Language
  • Nudity
  • Sexual Behavior
  • Sexual Themes
  • Flashing
Most of these should be self-explanatory. However, “Humor” is an indicator of specifically crude humor, not whether something funny happens or not. “Flashing” is an indicator for photosensitive persons.

These roughly correlate to ESRB, except they have “gambling” while we have “flashing” as our less-obvious indicator.

NOTE: Words and phrases found in the “Severity Level Tooltip” are benignly intended; any word or phrase that may be considered ‘loaded’ is purely unintentional and due to my own personal dumbness and lack of education. Tooltip words and phrases are open to revision for better clarity, acceptance, and inclusion.

Image

Those few will be the required feedback; clicking through them should be quick and painless.

In addition to the required feedback, there will be many potential “other considerations” that can be attached to the AMV. This feedback will be at your discretion.
  • Audible Screeching (distorted or disturbing noises)
  • Audible Volume (jumpscare audio, sudden volume spikes)
  • Bodily fluids
  • Bullying
  • Clowns
  • Cohersion
  • Emotional abuse
  • Exploitation
  • Explosives
  • Gambling
  • Hate speech
  • Insects
  • Occult (Witchcraft)
  • Orientation
  • Religious
  • Reptiles
  • Supernatural
  • Tone
  • Torture, emotional
  • Torture, physical
  • Transformations (grotesque body deformation)
  • Undead
Any content descriptor (including required) can be added multiple times to the same AMV. For example, one descriptor can be “Nudity: medium (general nudity), several, clearly depicted” while a duplicate descriptor can be a “Nudity: high (focused genitals), one or few, intense / dramatic” entry as an addition within the same AMV.

We can easily add more to the list of “other considerations” at any time. However, any creator can add any word or phrase in the “specific warning” section of content communication (details coming next). If we see a specific word or phrase frequently appearing as a “specific warning”, we may add it here as a standardized content descriptor.

Much like in the previous thread, I greatly encourage your feedback, perspective, and guidance in this proposal.

Thanks again and have a great day!

Phade.

User avatar
Phade
Site Admin
Joined: Fri Oct 20, 2000 10:49 pm
Location: Little cabin in the woods...
Org Profile

Re: AMV Content Descriptors

Post by Phade » Mon Aug 04, 2025 6:23 pm

I've gone ahead and filled out the rest of the optional categories, types, and level guidelines. Again, I may need some help with phrasing, so feel free to make suggestions. Are there any other categories or types that you think should be included in the initial deployment?

Image

User avatar
Mango
Joined: Tue Sep 22, 2015 9:38 pm
Org Profile

Re: AMV Content Descriptors

Post by Mango » Mon Aug 04, 2025 8:57 pm

I like the idea of tags but by adding severity levels for a single tag, instead of opting for a different tag entirely, you end up adding subjectivity, confusion, and a monotony to a task that should be arguably simple, which loops back to the same issue of the first attempt with number scores.

Just take nudity for example "more skin than usual" what is "usual?"

I'm not trying be pedantic but take for example I'm Muslim, under this descriptor I can't think of a single anime that won't get the nudity tag from me.

Heck do I give a nudity to Beastars? With how it features shirtless anthropomorphic characters that are covered in fur?

I applaud you for trying to add an attempt of clear meaning to these terms / tags but by bogging it down with severity levels (especially levels that are based on subjectibity bias), you end up losing the point of these tags entirely.

But why not start small with a general tag system such as
- For Kids
- For General Audience
- For Adults

Even just to see how well on this site would adopt such a system?
-----
I mean this respectfully as a casual user of the org. I often use AMVNews for video filtering whenever I design panels for Anime NYC, and I fear by having such a large list of tags, folks would get bogged down that they end up nixing to include them entirely, or miss including some, thereby making such a tagging system useless.
AMV Contest Coordinator @Anime NYC

Submissions Open! Deadline June 21st 2025!

User avatar
Phade
Site Admin
Joined: Fri Oct 20, 2000 10:49 pm
Location: Little cabin in the woods...
Org Profile

Re: AMV Content Descriptors

Post by Phade » Tue Aug 05, 2025 9:38 am

Mango wrote:
Mon Aug 04, 2025 8:57 pm
I like the idea of tags but by adding severity levels for a single tag, instead of opting for a different tag entirely, you end up adding subjectivity, confusion...

Since there are four static options (none, low, medium, high) for something that is essentially an amorphous judgment call, there will always be interpretations that must be made. I knew that if I just left the options as “low, medium, high” (or even just “Yes/No”), the next immediate question would be “What do you mean by low, medium, high?” The tooltip guideline is exactly that: a guideline. Captain Barbosa said it best: “The code is more of what you’d call guidelines than actual rules.” And this sentiment is absolutely true here.

Even the counterexample of “kids, general, adult” is the same “low, medium, high” scale with exactly the same judgmental nuances and the same subjective biases. What do you mean by “kids”? Is Little Witch Academia a kids show? There’s lots of witchcraft and supernatural things going on there. What about Dragonball Z? I know kids watch it, but there’s lots of physical violence that maybe kids really shouldn’t be exposed to in their lives. Evangelion clearly has specific “adult” themes, scenes, and violence in the anime, but a large number of Evangelion AMVs would (to me) be fine in a “general” showing unless those specific “adult” scenes are in the AMV. The list goes on and on.

And you are correct, every culture has different interpretations of acceptability. For the example of Nudity you highlighted, there may be a difference between what you as a person and general people would believe is “none” vs “low” vs “medium” when considering viewing the human body. I know the USA is kinda infamous for “violence galore is fine but showing slightly too much skin is OMG!!” while Japanese culture is obviously much more permissive when it comes to showing the human body by comparison.

If you are worried about, “What if other people disagree and get mad?” All I can say is the same thing I constantly tell myself: feel free to worry less about it. I know I overthink things when providing answers to people (/me points broadly at my chronic existential dilemma of “am I being too permissive or too purdish for including/excluding this AMV from my exhibition show”) and I know that people in the AMV community can often have the same challenge.

If someone replies to you, “Hey! You put there was a nudity level of ‘medium’, but I didn’t see any general nudity, and I would barely maybe put the nudity level at ‘low’ for your AMV. I demand a refund!!” Just roll your eyes and virtually pat them on the head. 🙂

So, what do we do? Yes, there will (eventually) be a simplified kids, general, PG13, adult type indicator as soon as we settle on some guidelines of what even that means. Tagging is important, and the “content descriptors” section here is designed to be a universal set of options to click through (rather than type) that hopefully highlight considerations important for most people; only a small number of category types will require responses. Interpreting the “none, low, medium, high” severity as “kids, general, older teen, definitely adult” is another perfectly valid way to view the levels and the system will still work exactly as designed; the guidelines provided are just guidelines, not rules. General free-text tagging will also be available for any additional non-standardized communication the AMV creator desires.

For now, my advice is this: the answer you provide for the content descriptors should likely be a cross between your personal perspective and what you think your intended audience would think. You are free to weigh both of those perceived values to determine the best answer of “none, low, medium, high” (“kids, general, older teen, definitely adult”) on your own. Ultimately, the answer you give will be the correct answer for you and the AMV. If you need more guidance about a particular edge case, a forum post or Discord conversation about it would likely help you decide.

I hope this helps clarify the intent and use of the “content descriptors” section.

(You have been very kind in bringing this up, and I hope kindness and understanding also comes across in my reply. I greatly appreciate your input!)

Phade.

User avatar
Phade
Site Admin
Joined: Fri Oct 20, 2000 10:49 pm
Location: Little cabin in the woods...
Org Profile

Re: AMV Content Descriptors

Post by Phade » Tue Aug 05, 2025 9:53 am

...

Actually, now that I think about it....

A good number of content descriptor tags could conceivably be accurately prefilled in many cases. Let’s say you upload a “My Neighbor Totoro” AMV. The site could prefill all Body, Brutality, and Substances category types as “none”. The optional “Supernatural” could be prefilled as “low”. The server could automatically scan the video for flashing and prefill a value there. The Expressions category with Humor and Language mainly depend on the song. If we had a more detailed music database (and several exist that we could utilize), we could potentially prefill those values as well. Even most optional category types could be accurately prefilled when we know the anime and music used in the AMV. And, yes, a highly skilled editor could make an AMV that should have a higher level than prefilled for any category type, but it would be up to those very few editors to make the adjustment for the betterment of the audience.

Prefilling category type level values for AMVs with multiple sources could default to the most severe general value associated with the anime used as a whole. The AMV creator would then adjust down only the category type levels that don't apply to their AMV because they skipped over the known objectionable scenes in their AMV.

Hmmmmmm..... 🤔

User avatar
SQ
Doesn't have a title
Joined: Fri Nov 08, 2002 8:11 pm
Status: youtube.com/SQ
Location: Upstate NY
Contact:
Org Profile

Re: AMV Content Descriptors

Post by SQ » Tue Aug 05, 2025 12:56 pm

I do want to see this sort of accessibility promoted in fan vids, but I also recognize a lot of people simple don't care. If the org's goal is still archival of all AMVs, then forcing even just the 10 content descriptors will prevent many people from submitting.
The people who care about accessibility will add it on their own videos or (if people can label other's videos freely) it will be supplied to videos by other community members.
Even AO3, which has a VERY accessibility-minded audience, allows people to decline giving their version of content descriptors. Keep in mind too, that allowing submissions to deliberately go undescribed can also be a form of accessibility.

People responding to you on the forum and possibly elsewhere have a vested interest these topics, but we don't form all (or even most) of the people who are using or who will use the site.

I'm not completely object to the idea of None/low/med/high severity, but I think simplified "present" vs "high severity" might be easier to parse and label. Additionally, some of the optional types and guidelines seem a little out there to me, but maybe you are privately asking people for their contributions so who am I to judge someone's sensitivity.

If you want a quick list of people's sensitivities, it might be a good idea to release an actual poll that can be passed around social media, listing the categories and subcategories you've already described here with checkmarks plus an additional text area for additions.
This way you can see how many people want to be warned for the things you've already defined and look at what type of written responses you're getting that may have been outside your awareness.

Furthermore, if your intent with these updates is to draw more feedback at every step of the process, I would heavily suggest utilizing your bsky, twitter, and/or discord server to advertise these wants. Most people are not looking at the forum unless prompted to do so.

I have set up a discord webhook to advise me of all new forum topics, but due to the need for a VPN I haven't bothered replying to anything until now because it has not been worth the effort. I can't imagine I'm the only one in this predicament.
Again, I may need some help with phrasing, so feel free to make suggestions.
I think there's been a lot of thought put into these and for the most part they are easy to understand.
I would advise rethinking:

Humor
It's unclear and you've already noticed it's unclear. Try something else like "Crude humor" ? "insensitive humor" ?
I'm not the best at wording things either, but "humor" by itself should definitely not be the final wording.

Language
Has the same issues as humor, but to a lesser degree.

Nudity
As already pointed out, this one is especially subjective and at times politically-motivated. Might I suggest adding an adjective like "Sexual nudity" ?

Sexual Behavior, Sexual themes
These are easily conflated and in most cases could be combined (including combining it with sexual nudity).

Your severity tooltips for "Sexual themes" are covered by other things -
"low" is covered by genre ("romance", assuming genre is a thing that will exist in the rebuild),
"medium" is covered by "drama," or potentially user tags such as "dubcon" (dubious consent), and
"high" is covered by sexual violence-related tags.

"Sexual behavior" seems unnecessarily puritan. Is "low" holding hands or is that "medium" ? Is "high" kissing or sex? The current wording is vague and the descriptor as a whole seems like it would be covered by other areas.

For the additional, non-mandatory descriptors, "self-view" is too encompassing. Self-doubt should not be included in the same descriptor as self-harm. Maybe "mental illness" or "mental health" would be a better descriptor name (though I realize this wording could also be problematic). With severities:
low = depression, delusion, hallucinations, etc.
medium = harm to self or others?
high = suicide ideation or attempts

However even this seems problematic; with the way mental illnesses work, all these tags really should be separated into their own thing, especially when it comes to any type of self-harm.

"Manipulation" I'd like to applaud - this feels like a good way to describe a wide variety of things people may be uncomfortable with, however in the same vein it also feels too versatile. I'm personally inexperienced with this type of abuse so I will leave it to someone else to possibly confirm or say otherwise. However, the high severity being emotional torture feels like it'd just fall under the regular torture tag.

"Tone" is a good idea but potentially needless if "genre" still exists. However, I would say I do like the concept of a tone indicator so maybe you could combine it or streamline it some way with genre.

"Transformations" you may want to include "Body Horror" in one of the severity examples?
Phade wrote:
Tue Aug 05, 2025 9:53 am
A good number of content descriptor tags could conceivably be accurately prefilled in many cases.
I think this would be helpful, but I'd hope there would be a confirmation screen to alter the automatic tagging. Context can greatly change perception. With increase in technology and changing of editing styles, we can also add things that weren't originally present in the source and/or take things away.
Multi-source AMVs with competing auto-loaded descriptors will also be a concern.
(I realize you mentioned this at the bottom of your post, but I figured it bears repeating.)
The server could automatically scan the video for flashing and prefill a value there.
This would be great but it also might be a technological marvel. Currently, there's only a handful of tools that do that and the (supposedly) good ones are paywalled. The current standard is The Harding Test which broadcasted content seems to use. The free tool everyone loves to reference is PEAT, which requires an uncompressed RGB avi, which makes it incredibly difficult to use.

Thankfully, EA came out with IRIS recently, which seems promising. I have no personal experience with it, though. (I did try to use PEAT briefly a few years back.)
WCAG has guidelines on what flashing to test for here, which IRIS appears to abide by.

There used to be a coding bounty out for PEAT alternatives; maybe IRIS got it, because I can't find the page with the bounty info now.

This is not to demotivate you. I have a personal interest in making fan vids more accessible to photosensitive people, so seeing any push for more inclusion is lauded. However, flashing is the absolute minimum of inclusion. IRIS also includes some patterns, for example.

Obviously I am not going to let perfect become the enemy of good, so I will just say to please consider a more open framework for additional safety concerns in the future, especially when it relates to photosensitives. In this case, separating the "physical triggers" (for lack of a better term) from the other labels may be beneficial.
It appears you have somewhat thought about it already with the inclusion of the audio categories. Which is good!

For another personal interest, might I suggest adding some type of support or label for if subtitles/captions are included (meaning that the AMV itself has captions for the song lyrics or what have you)?

----

Overall I am very hopeful & optimistic of the implementation of this or something like it. A lot of thought & consideration was put into the two tables you've already created and I don't have many complaints.
Discord: @standardquip (Vars)
BentoVid.com

User avatar
SQ
Doesn't have a title
Joined: Fri Nov 08, 2002 8:11 pm
Status: youtube.com/SQ
Location: Upstate NY
Contact:
Org Profile

Re: AMV Content Descriptors

Post by SQ » Tue Aug 05, 2025 1:38 pm

Sorry for the double-post.

The "orientation" tag seems fine at first, but after further reflection, I think it should be removed or at least separated from your low/med/high rating system.

The current rendition of the "orientation" descriptor implies that all relationships that are not heteosexual are deserving of a content warning, and that less-traditional romantic relationships are more wore harmful than others.

I would suggest changing the severity label text for this one tag, so instead of choosing no/low/med/high, it shows the specific orientations instead, such as hetero/homo/both/more (or something similar) OR removing this label entirely and making it user-specified only.
Discord: @standardquip (Vars)
BentoVid.com

User avatar
SQ
Doesn't have a title
Joined: Fri Nov 08, 2002 8:11 pm
Status: youtube.com/SQ
Location: Upstate NY
Contact:
Org Profile

Re: AMV Content Descriptors

Post by SQ » Tue Aug 05, 2025 1:47 pm

Would you consider allowing us to edit our posts in this forum? XD

I noticed that you did include opposite sex in your guidelines, but it's under what the "high" severity would be? Unless I've been looking at your tables all wrong. So you definitely don't imply the things I said it implied but now it's even more confusing for entirely different reasons.

If the purpose of the orientation descriptor is just the presence of an orientation, then that is covered by the other descriptors and possibly also genre.
Discord: @standardquip (Vars)
BentoVid.com

User avatar
Phade
Site Admin
Joined: Fri Oct 20, 2000 10:49 pm
Location: Little cabin in the woods...
Org Profile

Re: AMV Content Descriptors

Post by Phade » Tue Aug 05, 2025 5:15 pm

SQ wrote:
Tue Aug 05, 2025 12:56 pm
I do want to see this sort of accessibility promoted in fan vids, but I also recognize a lot of people simple don't care. If the org's goal is still archival of all AMVs, then forcing even just the 10 content descriptors will prevent many people from submitting.
Yes, the site’s core mission is: catalog, archive, and educate. If we can get autodetect and autofill for the options to be highly reliable, it should be very easy for editors to enter. Here is what I had in mind for the UI (though I am not a UI expert):

Code: Select all

Category: Level     None - low - mid - high

Brutality: Blood     O      O     O     O
Brutality: Violence  O      O     O     O
Substance: Alcohol   O      O     O     O
Substance: Drugs     O      O     O     O
Expression: Humor    O      O     O     O 
Expression: Lang     O      O     O     O
Body: Nudity         O      O     O     O
Body: Behavior       O      O     O     O 
Body: Theme          O      O     O     O
Visuals: Flashing    O      O     O     O

See more options: O
Hovering over the radio button would show the level tooltip guide. Again, hopefully these will be prefilled based on the indicated anime used.
SQ wrote:
Tue Aug 05, 2025 12:56 pm
The people who care about accessibility will add it on their own videos or (if people can label other's videos freely) it will be supplied to videos by other community members.
Even AO3, which has a VERY accessibility-minded audience, allows people to decline giving their version of content descriptors. Keep in mind too, that allowing submissions to deliberately go undescribed can also be a form of accessibility.
In the current proposal, “no answer” is a valid response for all optional category types. I only made these required because ESRB uses them for their content rating determinations.
SQ wrote:
Tue Aug 05, 2025 12:56 pm
People responding to you on the forum and possibly elsewhere have a vested interest these topics, but we don't form all (or even most) of the people who are using or who will use the site.

I'm not completely object to the idea of None/low/med/high severity, but I think simplified "present" vs "high severity" might be easier to parse and label. Additionally, some of the optional types and guidelines seem a little out there to me, but maybe you are privately asking people for their contributions so who am I to judge someone's sensitivity.
The optional ones are category types that I commonly use when selecting videos for a public audience exhibition.
SQ wrote:
Tue Aug 05, 2025 12:56 pm
If you want a quick list of people's sensitivities, it might be a good idea to release an actual poll that can be passed around social media, listing the categories and subcategories you've already described here with checkmarks plus an additional text area for additions.
This way you can see how many people want to be warned for the things you've already defined and look at what type of written responses you're getting that may have been outside your awareness.
Polling the users will definitely occur. Also if a text-based warning is frequently added, we may consider moving it to this standardized section so that more people are aware of the need for that content descriptor. However, the purpose of this post is to show that something could be built and here are the mechanics of the system.
SQ wrote:
Tue Aug 05, 2025 12:56 pm
Furthermore, if your intent with these updates is to draw more feedback at every step of the process, I would heavily suggest utilizing your bsky, twitter, and/or discord server to advertise these wants. Most people are not looking at the forum unless prompted to do so.
Initial new posts here in this forum are automatically posted in Discord.
SQ wrote:
Tue Aug 05, 2025 12:56 pm
I have set up a discord webhook to advise me of all new forum topics, but due to the need for a VPN I haven't bothered replying to anything until now because it has not been worth the effort. I can't imagine I'm the only one in this predicament.
I’ve tested out my own phone and it is currently blocked by the bot blocker. We’re still working on a better solution.
SQ wrote:
Tue Aug 05, 2025 12:56 pm
I think there's been a lot of thought put into these and for the most part they are easy to understand.
I would advise rethinking:

Humor
It's unclear and you've already noticed it's unclear. Try something else like "Crude humor" ? "insensitive humor" ?
I'm not the best at wording things either, but "humor" by itself should definitely not be the final wording.
This is essentially a copy/paste from ESRB. I figured if they thought it was a good phrasing, we might as well use it, too. We can certainly clarify it for our community's needs.
SQ wrote:
Tue Aug 05, 2025 12:56 pm
Language
Has the same issues as humor, but to a lesser degree.
Also copy/pasted from ESRB.
SQ wrote:
Tue Aug 05, 2025 12:56 pm
Nudity
As already pointed out, this one is especially subjective and at times politically-motivated. Might I suggest adding an adjective like "Sexual nudity" ?

Sexual Behavior, Sexual themes
These are easily conflated and in most cases could be combined (including combining it with sexual nudity).

Your severity tooltips for "Sexual themes" are covered by other things -
"low" is covered by genre ("romance", assuming genre is a thing that will exist in the rebuild),
"medium" is covered by "drama," or potentially user tags such as "dubcon" (dubious consent), and
"high" is covered by sexual violence-related tags.

"Sexual behavior" seems unnecessarily puritan. Is "low" holding hands or is that "medium" ? Is "high" kissing or sex? The current wording is vague and the descriptor as a whole seems like it would be covered by other areas.
My thought with the three required “Body” category types comes down to three distinct aspects: How much of the body can you see? What is the interaction with other bodies? What is the intent of the interaction? These three questions correlate to the nudity, behavior, and theme categories.

The combination of the three category types should quickly let you know what is going on. An AMV label “consensual, clear physical contact, more skin than usual” should be readily distinguishable from the combination of “non-consensual, passionate contact, focused genitals” as to what you are about to see. Relying on the external genre indication or knowledge of the source material to infer intent requires even more interpretation rather than looking in one place for the answer.
SQ wrote:
Tue Aug 05, 2025 12:56 pm
For the additional, non-mandatory descriptors, "self-view" is too encompassing. Self-doubt should not be included in the same descriptor as self-harm. Maybe "mental illness" or "mental health" would be a better descriptor name (though I realize this wording could also be problematic). With severities:
low = depression, delusion, hallucinations, etc.
medium = harm to self or others?
high = suicide ideation or attempts
I specifically wanted to include “self-harm” as a content descriptor, but it needed to be part of a group of three to fit into this system's mechanics. So, I kinda lumped it in with “I’m bummed, I’m depressed, I’m suicidal” to round out the need for three. Again, phrasing can be improved.
SQ wrote:
Tue Aug 05, 2025 12:56 pm
However even this seems problematic; with the way mental illnesses work, all these tags really should be separated into their own thing, especially when it comes to any type of self-harm.
Yes, nuances exist. But accounting for all nuances within a system has the tradeoff of simplicity vs overwhelming choices. I wanted the initial proposal to have at least an acknowledgement of this category type of content, so that you at least know I’ve considered it.
SQ wrote:
Tue Aug 05, 2025 12:56 pm
"Manipulation" I'd like to applaud - this feels like a good way to describe a wide variety of things people may be uncomfortable with, however in the same vein it also feels too versatile. I'm personally inexperienced with this type of abuse so I will leave it to someone else to possibly confirm or say otherwise. However, the high severity being emotional torture feels like it'd just fall under the regular torture tag.
Visually representing physical torture is easier to display than psychological torture. It was easier for me to come up with three distinctions of bodily torture than emotional torture. If you can come up with three levels and believe the optional tag will be used frequently enough, we can certainly add it.
SQ wrote:
Tue Aug 05, 2025 12:56 pm
"Tone" is a good idea but potentially needless if "genre" still exists. However, I would say I do like the concept of a tone indicator so maybe you could combine it or streamline it some way with genre.
Again, I was trying to be inclusive within this system and put it there as a ballpark indicator of what the viewer is in for rather than a tight indicator of content.
SQ wrote:
Tue Aug 05, 2025 12:56 pm
"Transformations" you may want to include "Body Horror" in one of the severity examples?
Yeah, we can swap “bone breaking” with “body horror”. (again, was trying to think of something at the time)
SQ wrote:
Tue Aug 05, 2025 12:56 pm
Phade wrote:
Tue Aug 05, 2025 9:53 am
A good number of content descriptor tags could conceivably be accurately prefilled in many cases.
I think this would be helpful, but I'd hope there would be a confirmation screen to alter the automatic tagging. Context can greatly change perception. With increase in technology and changing of editing styles, we can also add things that weren't originally present in the source and/or take things away.
Multi-source AMVs with competing auto-loaded descriptors will also be a concern.
(I realize you mentioned this at the bottom of your post, but I figured it bears repeating.)
Yeah, this was something I realized immediately after the previous post, so it’s a fresh idea that isn’t fully developed. The user interface might still likely resemble the above example, but our UI team is much better at it than I am. 🙂
SQ wrote:
Tue Aug 05, 2025 12:56 pm
The server could automatically scan the video for flashing and prefill a value there.
This would be great but it also might be a technological marvel. Currently, there's only a handful of tools that do that and the (supposedly) good ones are paywalled. The current standard is The Harding Test which broadcasted content seems to use. The free tool everyone loves to reference is PEAT, which requires an uncompressed RGB avi, which makes it incredibly difficult to use.


We could easily do the most basic of scan: collect the average brightness of each frame of the AMV and set a threshold of brightness change over time. This should give us very few false positives. However, it is only one aspect of photosensitive triggers. But given the typical nature of AMV editing styles, it should positively identify a large portion of problematic videos.
SQ wrote:
Tue Aug 05, 2025 12:56 pm
Thankfully, EA came out with IRIS recently, which seems promising. I have no personal experience with it, though. (I did try to use PEAT briefly a few years back.)
WCAG has guidelines on what flashing to test for here, which IRIS appears to abide by.

There used to be a coding bounty out for PEAT alternatives; maybe IRIS got it, because I can't find the page with the bounty info now.This is not to demotivate you. I have a personal interest in making fan vids more accessible to photosensitive people, so seeing any push for more inclusion is lauded. However, flashing is the absolute minimum of inclusion. IRIS also includes some patterns, for example.
Yeah, those systems aim for a clinical detection with few false negatives. We can implement whatever we want if we can get the algorithm.
SQ wrote:
Tue Aug 05, 2025 12:56 pm
Obviously I am not going to let perfect become the enemy of good, so I will just say to please consider a more open framework for additional safety concerns in the future, especially when it relates to photosensitives. In this case, separating the "physical triggers" (for lack of a better term) from the other labels may be beneficial.
It appears you have somewhat thought about it already with the inclusion of the audio categories. Which is good!
Yeah, this post is a “Here’s the mechanics of a system with example content. What do you think of the mechanics? We can tweak the content at any time.”
SQ wrote:
Tue Aug 05, 2025 12:56 pm
For another personal interest, might I suggest adding some type of support or label for if subtitles/captions are included (meaning that the AMV itself has captions for the song lyrics or what have you)?
Funny you should mention that. I already have “autodetect burned-in subtitles and words” as a “Phase 2” feature.
SQ wrote:
Tue Aug 05, 2025 12:56 pm
Overall I am very hopeful & optimistic of the implementation of this or something like it. A lot of thought & consideration was put into the two tables you've already created and I don't have many complaints.
Thank you! After the butt-stomping I got in the previous thread, I was a little apprehensive about this one. Again, the focus here is on the mechanics of the system. Once built, we can easily tweak the content.
SQ wrote:
Tue Aug 05, 2025 12:56 pm
The "orientation" tag seems fine at first, but after further reflection, I think it should be removed or at least separated from your low/med/high rating system.
Yeah, that was a last-minute addition when I was trying to think of obvious things I could have forgotten.
SQ wrote:
Tue Aug 05, 2025 12:56 pm
The current rendition of the "orientation" descriptor implies that all relationships that are not heteosexual are deserving of a content warning, and that less-traditional romantic relationships are more wore harmful than others.
Again, it was a last-minute add to improve inclusivity. But oddly enough, the heterosexual orientation is placed in the “worst” category level in the example guideline.

The orientation type does not necessarily imply romance within the video, just the orientation of the person(s) portrayed. A heterosexual or same-sex person can easily be portrayed without engaging in any romantic interests or activities. It would only be in combination with a non-none Body category type of Behavior or Theme that romance could be inferred.

Example: “same-sex, consensual, clear physical contact, non-nudity” is highly likely to be “romance”, but the content descriptors give you a much clearer picture of what to expect than just a “romance” genre indication. Even knowing the subject of the originating anime may not be enough; some industrious editors have made two heterosexual persons from two separate anime appear to be romantically involved through clever editing.

We can certainly separate orientation into its own non-weighted section of the form.
SQ wrote:
Tue Aug 05, 2025 12:56 pm
I noticed that you did include opposite sex in your guidelines, but it's under what the "high" severity would be? Unless I've been looking at your tables all wrong. So you definitely don't imply the things I said it implied but now it's even more confusing for entirely different reasons.
Yeah, the “low, medium, high” weight was not in mind when I made the level guideline, just that I wanted some kind of inclusion and these three encompass the highest frequency of AMV subjects.
SQ wrote:
Tue Aug 05, 2025 12:56 pm
If the purpose of the orientation descriptor is just the presence of an orientation, then that is covered by the other descriptors and possibly also genre.
Again, I wanted to be inclusive within this section of the system so that inferences could be made with only the information provided in the “content descriptors” without relying on genre to complete the picture.

(I'm just glad this conversation is more 🤔 than 😠)

🙂

.

User avatar
Mango
Joined: Tue Sep 22, 2015 9:38 pm
Org Profile

Re: AMV Content Descriptors

Post by Mango » Tue Aug 05, 2025 9:50 pm

Thank you for considering my words! ☺️

I do want to add on however, because I'm not sure my initial point was clear.
Phade wrote:
Tue Aug 05, 2025 9:38 am
If you are worried about, “What if other people disagree and get mad?”
It's not at all that I worry about folks disagreeing with what tags may be on a video. It's that I worry the org would become convoluted and otherwise confusing for newcomers to join or otherwise browse the great amvs hosted on the platform.

I wish to re-express, and re-emphasize this initial point
Mango wrote:
Mon Aug 04, 2025 8:57 pm
adding severity levels for a single tag, instead of opting for a different tag entirely, you end up adding subjectivity, confusion, and a monotony to a task that should be arguably simple,
What I am suggesting is not necessairly to get rid of the "levels" but instead transform them into much more digestible, and arguably more intuitive/clearer tags.

Take, for example, Blood/Gore "low" and how it encapsulates videos that feature just mere scrapes. I would not intuitively connect the term "blood/gore" (even when associated with the word "low"). In fact, I would be outright uncomfortable attaching such a tag onto my video because I would be afraid of turning away those who would otherwise enjoy my video but avoided it for wanting to filter against blood/gore in it's entirety.

Now, what if instead of Blood/Gore "low" we have a general fighting tag? Now I am comfortable in adding "fighting" to my video, as there is no visible blood or gore, but for sure there is fighting (its how they got the scrapes afterall!).

Now let's I do use a source that showcases blood, well instead of blood medium, I can just use the word blood, and if its a show like naruto, I can also incorporate the previously mentioned fighting tag.

Finally, for "high" lets say this video breaches into including gore, well now they editor can (potentially) use all three tags, fighting, blood, gore.

By finding general, but clear, terms instead of arbitrary levels, you can intutively convey, and discern your want of a "low, medium, and high" for your tags.

Also now that we've separated these tags into clearer terms, it's much more intuitive if we were to mix and match tags with tags under a different "category"

Subjectivity is definitely unavoidable, even under my proposed system. But it's much easier to discern whether a video features blood, as opposed to determining where the blood is on a scale of low, medium, or high.

I'm coming at this as someone who organizes AMV panels. Just last year I hosted an 18+ panel that featured fanservice, horror, gore, mature topics, and/or raunchy jokes. Those 5 descriptors that I just used are much clearer than, "humor high, language high, sexual themes(medium? high(?)) sexual behavior(medium(?)), alcohol high violence high[...]" and so on.

If the system is too subjective that searching / filtering videos is not straightforward, I (and I assume others) will pursue other means to find the videos I need / want (usually AMVNews, word of mouth, and so on) which leads into my last point.

I want the org redesign to breathe life into the community where we can maybe point those starting off in the hobby (either editors or just casual viewers) to check it out, maybe find a video that they like because god knows finding videos under the current YouTube algorithm is very difficult. Tags aren't at all useful if it will cause people to turn away from either videos they may enjoy, or the website due to confusion or annoyance.
AMV Contest Coordinator @Anime NYC

Submissions Open! Deadline June 21st 2025!

Post Reply

Return to “Org Site Rebuild”