The Movie Database Support
Support → Website Support
Default language of uploaded images
posted by bullmoose20STAFFMOD on February 20, 2024 at 7:11 AM
I propose that images that are uploaded have a default status of unclassified. Today, if someone is looking for a textless poster, they would look for "No Language" and there they would see, in many cases, a textless poster, and usually a bunch of other photos that contain text in a language. The issue is that the default when uploading a poster is that tmdb sets it to "No language".
This is a problem when you query via api as many times a poster with text is returned.
If the default language was "Unclassified" when you upload a poster, that would make it very clear that the file was uploaded but did not have it classified yet and thus needs a second step.
This would improve the metadata on tmdb and eliminate/reduce the "poisoning" of the "No language" posters which should literally have 0 text on it but as of today, continuously gets poisoned because all uploads end up in this language category.
The second benefit is when someone misclassifies a poster. For example, someone says a poster is in Russian. Then a Russian person comes along and says...that's not Russian. I don't know what that is, so let me set it to another language or to "No Language" thus poisoning the data again. Giving the Russian person the ability to set it as "Unclassified" is the best option.
So the change request enhancement is simply: 1- new language category called "Unclassified" 2- set the default language on upload to "Unclassified"
Over time, the data will clean itself.
Reply by Travis Bell
on February 20, 2024 at 11:04 AM
Hi @bullmoose20, we have an open ticket for this feature request here. You can watch/vote it.
Reply by bullmoose20
on February 20, 2024 at 12:42 PM
Wow..... so the trello is from 2019 and the original request dates back to 2013. So it's been there for 11 years.
I thought my proposal was a little more unique and less invasive and does not change the current workflow. I am talking about created a new category called Unclassified and dumping images to that instead of into the No language category.
No extra steps for users.
Another option? Create a category called Textless. This would be far more user impacting IMO as it's been established that if you want textless images from tmdb, you use "No language" as a filter.
Imagine if your current flow had been to dump images into the English category. This would mean that the English category is constantly poisoned. Pretty frustrating, right? The same now applies for No Language. Constantly going to clean up. And what if there is a language that I don't know what it is? What would I classify it? It's definitely a language so it should not be in No language but where?
Reply by bullmoose20
on February 20, 2024 at 9:11 PM
Here is a good example: The Color Purple for No Language has 4 images. 1 of which actually has English text. So there is approximately a 25% chance that the api call will return the one with text on it because its not classified properly in terms of language on the poster. It also assumes that all votes are equal but I have also seen images with text in the textless filter that have text and are the highest voted. This fix for The Color Purple is pretty easy... assign the misclassified one to English. Image of The Color Purple for No Language
Reply by bullmoose20
on February 26, 2024 at 8:59 AM
What do I do with this one? Planet Earth? I don't know this language. Definitely has text on the poster so its not "textless" or "No Language" so what should user do now?
Reply by FS.Corrupt
on March 9, 2024 at 2:13 AM
I am of the same opinion as @bullmoose20 this would simplify everything and prevent grabbing things via API that are incorrect. A new category where everything defaults to unclassified and is called 'Unclassified' without any workflow behind sounds good. This way, the user can simply select the correct one after uploading, or even during uploading? This ensures that no artwork ends up in 'Textless' due to laziness or forgetfulness."
Reply by yozoraxcii
on March 23, 2024 at 7:30 AM
Fully support changing the default language to "unclsassified", as @travisbell stated in this original thread:
https://www.themoviedb.org/talk/5117785f19c295079f0d78f1
"I believe that no data is better than wrong data."
The default behaviour of "No language" is wrong data in a lot of cases, so changing it to "Unclassified" or "Not Specified" would be more accurate until someone does specify a language.