Nick’s Desperate Search for Intelligence
Let’s automate our pitch decks with AI !
I’m sure many of you are aware of the scary new tools for artificial intelligence generated images. Is it going to make artists obsolete? If so, what is the progression forward in job destroying progress? Photographers, Cinematographers, Composers… CREATIVE DIRECTORS?!
Let’s define two terms before we get started.
Traditional. I’ll use this to describe any image, sound or 3d model created by a human without the use of deep learning algorithms. The traditional way.. Like Blender, Photoshop, ProTools…. Or like, painting by hand? Crazy.
Ai. This is an artistic asset generated through an artificial intelligence using a multi-dimensional latent space. How That Works < link for fellow nerds.
I believe stock photography will not remain the most cost-effective or quickest way to do “Visual Research” for decks or treatments. Currently many AI tools have difficulty with generating both realistic human faces while posing their bodies in a useful way. Some are amazing at faces and others are great at background scenes. The uses are limited and unique.
Like with beginning human artists. The AI is still bad at hands
For now, Camlock has been combining organic and ai assets with great success. We’ll generate scenes or props with Dall-e and photoshop elements of those images into other stock assets.
Here is a workflow that as of this email, utilizes several tools to make incredibly specific images that communicate your vision far better than a traditional stock search.
Step One
Have a good idea - We want to show some Millennials on a peer having a great time and holding up our product, a chicken sandwich, with some jets flying by in the background.
*you got this in the bag*
Step Two
Find a stock photo that’s relatively close.
GREAT!
Step Three
Use Dall-e to erase part of the image that you’d like to enhance.
Then add your prompt. “White girl holds chicken sandwich” Woohoo.
Then do it again.
Wow, what a specific image.
Good enough? Maybe not. You can get these results with stock and photoshop, but how many frames and revisions can you create in an hour?
If you are communicating your ideas visually with a collaborator or client, you’re already beating out someone with an image that doesn’t match the script.
Two free tools for image generation.
Dall-E-2, Dream Studio and my new favorite MidJourney!
Promising applications for AI coming in the near future!
(that will destroy other career paths you might be interested in!)
-Video to Animation.
Anything that can be done in Photoshop will make it After Effects in 5 years. But AI is quicker than Adobe. So let's call it 6 months.
Stable diffusion ran on top of video frames has an incredible look.
Similar to the style of Richard Linklaters A Scanner Darkly, or Apollo 10&½. This could easily become a replacement for motion capture… if it’s not made redundant by text to 3d animation.
-Ai generative animations in 3d.
What is 2d will become 3d. There are great tools for creating low poly models based on 2d images. Which can be made from text now. So if you want a 3d printed dancing banana man, you can make that happen.
-Ai Music.
Yes, it will be the new creator of ukulele and handclap tracks for instructional videos years to come. It’s the furthest out right now. So, next week it’ll probably be working.
Please feel free to call Nick and chat endlessly about Ai tech and its potential applications for advertising. He needs someone else to talk to about this.
Happy Thursday!
-The CamLock Team
