Can you clarify what is the difference between this and the current existing interface for entering multilingual descriptions during upload?

We're exploring several UI changes to support planned features for Structured Data on Commons:

• A "caption" (which is essentially what the Title field is now) that works similar to how labels work in Wikidata. Think of this as a way to get a multilingual filename. It's meant to be a very short field, unlike Description which can be much longer.
• In terms of the current UploadWizard UI, we will need a new functionality that allows entry of that caption in multiple languages, similar to how it works for Description now (but hopefully with a little improvement).
• Multilingual Descriptions aren't going away (but they will be stored in Wikibase instead of wikitext in the future, and their display on the file page will be improved). The caption is in addition to the description, which can be much longer and perhaps contain limited markup.

Thanks, that makes sense :)

Nice job with the caption field!
Captions totally make sense, as it is important to have a sort of multilingual label that described the file and can be edited.
It makes the uneditable and untranslatable filename/title field redundant, so I removed it in this wireframe:

https://commons.wikimedia.org/wiki/File:Multilingual_Captions_wireframe_caption%2Bdescription,_based_on_v3.png

As per the active discussion, I believe that we must get rid of filenames, and use identifiers such as "M84945" instead, because filenames contains human words whose meaning change with time, and sometimes become offensive. See https://www.w3.org/Provider/Style/URI on why words must not be used as URIs, and opaque identifiers must be used instead.

Cheers, keep up the great work!
Nicolas