This is my 100% AI Generated Music+Video.Here is a little more information on the process.The aim of this was to test free tier generative AI Video software.So I made this over a few days for fun.The track is an old track of my own.. that i totally remixed and added vocals with AI. And then masterted using AI software.Then I moved onto the video.The main AI’s I used were :- Hailau for a lot.. as it was free for 3 days.- Hedra for lip syncing.. (most lip syncing to music... Read more
This is my 100% AI Generated Music+Video.
Here is a little more information on the process.
The aim of this was to test free tier generative AI Video software.So I made this over a few days for fun.
The track is an old track of my own.. that i totally remixed and added vocals with AI. And then masterted using AI software.
Then I moved onto the video.
The main AI’s I used were :
- Hailau for a lot.. as it was free for 3 days.
- Hedra for lip syncing.. (most lip syncing to music doesn’t work well)..
- Runway… the free tier doesn’t offer too much.
- Kling was good.. but you’ll wait in a queue for 3 days for a 5 second clip right now as it’s too popular.. so basicly useless unless paying.
The trick I found most useful was to use chatGTP to write prompts for the video generators. This gives you a ton of detail in a format you might not consider. And generally gives more consistent results.
Using video as a source for things like Hedra often give glitchy messes when trying to lip sync. It takes a lot of trial and error. Not something you have many shots at on the free tier.
If someone wanted to drop a few thousand on subscriptions.. you’d have a lot of freedom to experiment more.
And that’s where it gets expensive.
But paying, does also cut render times down.
So.. something to think about.
This was all then edited together manually using Avid Media Composer.
Detailed edits is not something AI is very good at.. yet.
But give it a year. And I suspect it may be very diferent.
Here is a little more information on the process.
The aim of this was to test free tier generative AI Video software.So I made this over a few days for fun.
The track is an old track of my own.. that i totally remixed and added vocals with AI. And then masterted using AI software.
Then I moved onto the video.
The main AI’s I used were :
- Hailau for a lot.. as it was free for 3 days.
- Hedra for lip syncing.. (most lip syncing to music doesn’t work well)..
- Runway… the free tier doesn’t offer too much.
- Kling was good.. but you’ll wait in a queue for 3 days for a 5 second clip right now as it’s too popular.. so basicly useless unless paying.
The trick I found most useful was to use chatGTP to write prompts for the video generators. This gives you a ton of detail in a format you might not consider. And generally gives more consistent results.
Using video as a source for things like Hedra often give glitchy messes when trying to lip sync. It takes a lot of trial and error. Not something you have many shots at on the free tier.
If someone wanted to drop a few thousand on subscriptions.. you’d have a lot of freedom to experiment more.
And that’s where it gets expensive.
But paying, does also cut render times down.
So.. something to think about.
This was all then edited together manually using Avid Media Composer.
Detailed edits is not something AI is very good at.. yet.
But give it a year. And I suspect it may be very diferent.
We’re not there yet.
But AI is a tool. And we all need to get comfortable with the next wave of tools that are coming to our industry.
I remember coming into the industry just as non liner digital edit system were being introduced.
There were a lot of older guys that knew linear tape editing.. but found the jump to Avid systems a huge learning curve. As the younger people.. we felt like we had an advantage.
But I always knew that the day would come for us digital editors.
But the tools sped up workflows.. but do create other work.
At first we’ll be using AI to sort and find clips..then we’ll probably let it build montages if we’re in a hurry.. let it trim down a few interviews.. and things like that.. then we’ll find it’s better at matching action between shots than the average human.. so that will get handed over.
You can see how slowly we’re automating the edit.
But that’s the way it’s going .
We can no more change it than gravity.
So I think we need to god super comfortable with doing the tweaking and value we can add.
But it’s still a tool. As media creation is a deeply personal choice of what to include.
Even if one person created their ideal bit of media.. someone else would have notes/changes.
So an AI can produce things.. but it’s still users giving it the changes and notes.. which could be infinite.
So loosing control isn’t the issue in my opinion.
But it will speed up certain tasks for sure.
And if doing it the long way.. will probably have less time to be creative.