My latest video made using AI technology for generating images, animating them, and creating music.

Yeah, yeah, yeah, you hate AI and think anyone who uses it isn’t a real artist and is just a poser. To quote my favorite Chinglish T-shirt I saw when I lived in China, “YOU DON’T KNOW WHO I’M” Interesting grammar point there. Can’t put “I’m” at the end of a sentence or it sounds hilarious, especially when flexing. But, yeah, can a real artist whose been making art for over 40 years, and who has an MFA explore the possibilities of AI without getting an avalanche of shit dumped on his head? Uh, no. But, it ain’t gonna’ stop me. When the world gives you a shit wave, surf it.

Let’s just get out of the way that this isn’t easy to do. As soon as people see “AI” they think AI did all the work. Have a look at my video editing interface:

Timeline in video editing software.

In the timeline above you can see a section of my video with hundreds of rectangles. You see how they are stacked? That’s where there’s overlapping video, music, narration, sound effects, and other effects. An you might see little dots along a line. That’s me editing the volume of the soundtrack, and other audio so they blend well. The longest a video animation lasts is 4 seconds. This is a 7 minute video, and that means there are over 100 separate animations in there. Each animation took multiple attempts, and each one is based on an image that took multiple attempts.

Here’s the stats for the file on my computer for this video:

Not only is it over 5 gigs, it has over 1,187 files in it. That’s how many images, animations, and audio files I generated in order to assemble the video. I produced this in 10 days at breakneck speed, starting with my first sip of coffee, and ending when I shut off my computer late at night. Easily there are 50 or more hours of work in it.

If I make it look easy, that is to my credit. If you think AI just churned out the whole video, I’ll take that as a supreme compliment. Because in reality it is a hodge-podge of individual images each made through multiple attempts, and then made into animations, and then cobbled together. To the degree it is seamless, that’s all my video and audio editing skills.

Other people who are making AI videos don’t have sound effects. Mine include innovations in the new medium already, because I’m not a follow the template kind of an artist. I’m a break the mold and turn it up to eleven type. In fact, I haven’t even seen another AI video of this caliber, but I’m not really looking hard, either. So, with this new medium that is only really available in the last months or even weeks, I’m already hitting it hard. And that’s because I’m not locked into any approach or medium or ideological framework. I’m willing to jump ship and try something new, when I see how unbelievably promising it is. That ain’t a fault.

Sure, sure, using AI to make a single image and pretending you are an overnight Caravaggio is a level of clueless arrogance that staggers the mind. But this is a different cup of tea, and I think the ability for individuals to make videos, music videos, and short films on their laptops is extraordinarily liberating, and can lead to a maximalist artistic revolution, because it uses a whole range of artistic elements. Just trust me on that.

I’m too busy creating and trying to ride the crest of this wave before anything I do is lost in a great deluge of mediocrity by people who have enormous following or can pay to promote their garbage to explain everything.

I gotta’ get back to work. I’m working on another music video.

12 replies on ““War Of The Worlds, Reimagined””

  1. A friend took one of my images and fed it into Runway, then showed me the 4 second video it created. The last 3 seconds were crap, but that first second made me realise that I wanted to animate my characters too, just a bit. Which is a long way of saying that I found this post very interesting and have a ton of questions.

    First up, I’m assuming you created the ‘seed’ image? Next, how did you get the AI to create sequential images/video? Third, which video editor are you using? Fourth, which AI did you use?

    I use CorelDraw to create digital collage type images, and I’ve been teaching myself video editing over the last two years with VideoStudio Pro 2021. Until quite recently I’ve only been making videos related to gaming: player housing videos with music and voice-over. I would like to do a kind of graphic novel speaking book type video long term so all information gratefully received.

    Liked by 1 person

    Reply

    1. For image generation I use Midjourney, which IMHO is the best. But I also use Stablediffusion Cascade, which is free, and does a great job for a lot of stuff. For animating I use Pika, Haiper (free), and Runway. All have different problems.

      If you want to animate a single image you can get good results. Best if you don’t want something very specific. But the longer an animation or video you make, and the more specific it is, the difficulties multiply exponentially. But to play around, I recommend using Stablediffusion Cascade to generate images for free, and Haiper to animate them for free. But to make a video you need video editing software or some service.

      Liked by 1 person

      Reply

      1. Thanks, Eric. Much appreciate the headsup. I already use VideoStudio Pro for video editing/creation, and it does have a ‘motion capture’ function, but it’s pretty primitive.
        I’m currently creating single images for the stop motion function, just to see if what I end up with is worth the effort. Old fashioned animation is a pain, but 3D animation like Blender or Maya requires a huge learning curve.
        I might give Haiper a try, just to see what it does with my concepts.
        Cheers 🙂

        Liked by 1 person

      2. You can’t go wrong. Haiper is free and unlimited, while also being top quality. But you are basically rolling a die to see what you get, and need to try several times before it might give what you are looking for.

        Liked by 1 person

      3. I haven’t tried any of the LLMs so far so it’ll be interesting to see how well I can ‘communicate’ what I see in my head to a machine. 🙂

        Like

      4. In my experience it’s very hit and miss. I think there needs to be something like ChatGPT as an intermediary, because it understands written language very well.

        Liked by 1 person

      5. -giggles- isn’t ChatGPT the one that recommends eating rocks ‘for your health’?
        To be honest, I feel happier about LLMs now that we can see how limited they are [at least for now]. If they ever get context right, things will be different.
        I will approach it as a tool, nothing else.

        Like

      9. Mmm…no. Sorry, but I know how to do research and I try to go to primary sources where possible so getting an answer where I have no idea how accurate it is? No. There is literally nothing more valuable than accurate information.

        Like

Leave a comment