11 Comments

Have you checked out Musavir.ai ?

Expand full comment

I haven't actually. At first glance, looks like yet another UI for Stable Diffusion. Surprised at the cost, too, considering you can use Stable Diffusion for free or with generous credits on most other platforms.

Have you tried them, and is there anything that makes their platform stand out?

Expand full comment

I've seen some fantastic stuff on LinkedIn done on it. But no, I've not tried it myself as I couldn't see a free trial and wasn't prepared to sign up without more info. I got the feeling It might be a Middle Eastern/Muslim orientated and created site.

Expand full comment

Yeah it's quite an entry barrier to charge people upfront without really showcasing what's different about your platform. You have popular sites like Leonardo.ai and Nightcafe Studio that give plenty of free credits to start with. But if you do hear anything interesting about it, I'd be curious to know.

Expand full comment

RE bard: definitely limited on the videos as of right now, to those that already have a transcript.

Expand full comment

Yeah it does appear to rely on a captions file, which definitely takes away from the magic!

Expand full comment

How long, though, until all YouTube videos have a captions file? Even ones w/no words will inevitably have a description suited for an LLM, I would imagine.

Expand full comment

There's already auto-captioning on practically every video you upload. I know I uploaded a few with no captions of my own and had transcripts auto-generated. So yeah, plenty of food for LLMs.

Expand full comment

Does auto-captioning create a file every time, or is that something that needs to happen to reach all videos? I'm also thinking more about videos with no words, but with a description of what happens in the video. That's a quantum leap once we have that, I think.

Expand full comment

I think it's a toggle when you're uploading a video to YouTube. You can probably leave it off and no transcription will take place. And yes, being able to have AI actually understand the video without any accompanying text will be a massive upgrade to the current "chat with a video" tools.

Expand full comment

A really good challenge will be competitive sporting events with no commentary. I'm thinking specifically about Brazilian Jiu Jitsu matches, a very easy example for me to pluck out of my brain. There's a lot of expertise and analysis needed to describe what's happening on the screen. That would be an amazing challenge to overcome, and I'm sure other niche sports would provide something similar.

Expand full comment