Audiobus: Use your music apps together.
What is Audiobus? — Audiobus is an award-winning music app for iPhone and iPad which lets you use your other music apps together. Chain effects on your favourite synth, run the output of apps or Audio Units into an app like GarageBand or Loopy, or select a different audio interface output for each app. Route MIDI between apps — drive a synth from a MIDI sequencer, or add an arpeggiator to your MIDI keyboard — or sync with your external MIDI gear. And control your entire setup from a MIDI controller.
Download on the App StoreAudiobus is the app that makes the rest of your setup better.
Hekate Station : a short horror video with soundtrack by Svetlovska (Cthonicist)
On Svetlovska's post, " Mainly Moog, a little Luapsel: Black Site" @LinearLineman suggested the possibility of a collaboration.
I asked if I could use "Hekate Station" as the inspiration and soundtrack for an AI generated video and Svetlovska graciously said yes.
I had a wonderful time creating this and working with @Svetlovska.
If you aren't a fan of the horror genre you may find the video disturbing. For those that choose to, thanks for watching.
Comments
Just amazing video work, @MadeofWax ! Thank you for doing this, it’s great. I begin to understand why the Hollywood unions are so concerned. Give it a few years and you’ll be able to replace pretty much everyone in the film business with a laptop.
Thank you very much for sharing your excellent track with me. I appreciate it very much and hope we can work together again.
Any time!
I am effectively disturbed. Great work, @MadeofWax and @Svetlovska!
Haha! That was great. As I thought it would. It’s a special feeling to see one’s music “come alive”. Usually the music is designed to fit the visual. The opposite in this case (tho Fantasia comes to mind).
@Svetlovska ’s music got what it well deserved. And on a budget, too! Curious how long this took you @MadeofWax. I might have a track for you.
Thank you!
Thank you!
For the comment and for making the suggestion to collaborate.
The video process was actually very fast once I found the right concept for it. Svetlovska was incredibly helpful with suggestions and feedback throughout the process. I was also very lucky with the generations. Sometimes you can throw a prompt or image into a video generation service and get nothing but AI nightmares.
The apps are getting better but there is still a high degree of randomness. You have to be flexible and work with the output you get or be willing to change direction if something refuses to work.
Really great, you two! 👏 👏
@MadeofWax I'm curious how much it cost to make this roughly, in terms of using the credits or whatever in the AI video making app you're using.
Good times yall, thank you creators!
Thank you!
For a 3 minute video with upscaling it's very easy to go through 800 credits, basically a months worth in the $30 subscription plan. You never know what prompt will work out. Sometimes your generations are amazing, sometimes they are filled with errors. For the video I made for Paulieworld it took close to 1000. If I wasn't a part of their artist program I wouldn't be able to afford to make more than one video a month.
Thank you!
Thnx, yes that's very expensive really. Cheers!
And please remind me which service(s) you use.
I'm using Vidu. I've also used Luma and it is comparable. Both have a free trial where you can generate watermarked video if you want to try them. Both allow text to video so you can create with just a prompt or image to video if you prefer to use images. I generate images locally on my PC and use them as a storyboard. Some people use image services like Midjourney.
The most expensive and most advanced service would be Runway. I believe they had an unlimited plan for $95 a month last time I checked.
https://www.vidu.studio/
https://lumalabs.ai/dream-machine/creations
https://runwayml.com/
Thnx for the info!
Breathtaking! This could literally be a finished Hollywood product. A professional trailer for a big budget sci-fi horror film. The soundtrack and imagery are seamless and work perfectly together. I’m not a big fan of the genre but I recognize the high quality of the production.
I am now very interested in the way you set this up, and your workflow. Do you start with an image and add a storyline behind it? What are the “prompts” like? Short simple phrases, or detailed descriptions? How long does it take to render a scene? As I watch Moonshiners, I can’t help but wonder how you created those scenes that worked so perfectly together. Each scene is different, but they all worked together to tell the story.
This AI stuff is both incredible and terrifying. I often wonder if we have already reached and passed the singularity. Are these “machines” already self-aware, and if so, how would we know? Will they have a soul and a sense of right and wrong? I think about this stuff… then I pour a drink and play with my iPad.
@Paulieworld ,
I start with the basic idea and look for models that I think will give me images to support the story. On my computer I'll start rendering images for a rough outline, adjusting the prompts depending on the output.
For Moonshiners I usually just used fairly basic descriptions like, "a sheriff smoking in a dimly lit bar in a Louisiana bayou town" and would see what I got. If I need specific details I'll add them a few at a time. Some concepts work great. Some things don't work at all.
If a prompt gives me good output I will make several more images using the same prompt with different random seeds so I will have multiple images to choose from when I get to the video stage. I try to think about continuity when it comes to day or night settings but it helps to do images of both in case I need to make changes.
Once I have a good amount of images for a storyboard I start loading into them into Vidu one at a time with a basic prompt matching the image and some description of the camera movements.
There's an "enhance prompt" checkbox that I tend to leave checked and an option for 4 or 8 second videos. The longer videos tend to have more usable parts but it's almost like using a slot machine. You never know what is going to work.If I can't get a usable video from an image after 3 tries I usually move on. Sometimes changing the prompt helps. Sometimes unchecking the "enhance prompt" checkbox helps.
As I'm generating videos, sometimes I realize that an idea just isn't going to work. Sometimes I'll get a video that looks amazing and I'll try to adapt the story to incorporate it. It's almost like improvisation at times, reacting to the output and trying to adapt as you go along.
After I have a good bit of footage I'll start assembling it in my video editor. I'll see where things fit together and where they don't. Sometimes I'll render more images to fill in gaps in the story. Sometimes I'll realize that some good footage isn't going to fit. I'll tweak it, watch it, and tweak it some more.
Eventually I get to the point where I feel like it's done. I'll export it and watch it again. 99% of the time I will see something else I forgot to fix and I'll end up exporting it again. After that I'll share it with the person I'm collaborating with for final approval. Then upload it and hope someone else enjoys it.
Agree this is incredible work, how far things have come in a short period of time.
While I can't commit $100 a month this really is ridiculously cheap for access to the tech to do this, although @MadeofWax is clearly very talented at using it too!
And, @Paulieworld, you don't need to worry about sentient AI yet, its getting very good at faking it and telling folks what they want to hear (based on statistical weightings and probabilities) though.
Thank you @belldu ,
I do think the unlimited plan for Runway is ultimately a bargain. Considering what it would cost to make movies with traditional tools it is absolutely worth it.
When I got the email from Vidu saying that I would be part of their artist program it was a huge help. Their tools are getting better all the time, they’re very responsive to feedback and I’m able to be creative with them without going into debt.
I’ve been very fortunate with the way AI has developed. If my computer was any weaker I never would have tried to install the early text to image apps. If it wasn’t for the encouragement of people here on the forum I would have given up making videos a long time ago.
Great piece of work (both of you) 🙏
Also fascinating reading how you do it. The end result must take you hours but you definitely have a gift for it 👌
Thank you @GeoTony
It's very similar to the process I use to make music, oddly enough. I've gotten comfortable trying to piece together randomness into something that hopefully is cohesive and entertaining. It helps to have great sound to work with. I'm grateful to have such talented collaborators.
I give it about a year and you won't be able to tell these videos are generated unless you have visuals that give it away deliberately.
And now you can properly add "Sound Designer" or "Soundtrack by..." to your resume.
Fantastic work.
At this point the biggest hurdle is consistency. The unpredictable quirkyness is can be a fun part of the process but I imagine it probably wouldn't be appreciated in a production pipeline. When you have to make certain shots happen on a deadline it isn't quite ready for prime time yet. But considering I made my first 512x512 image using AI just 2 short years ago it truly is amazing what can be done now. I wouldn't be surprised if you're right about the timeline.
Thank you.
Great stuff; great pairing of talents.
Also thanks for sharing your process.
Those spaceship interiors: someone's going to have to clean up that mess!
Thank you @MrStochastic
I'm definitely not volunteering for cleaning duty.😀