Have you ever been on a creative rut? Lacking the excitement you once had for your art? Or maybe you’ve had ideas that are so crazy it would take you years to learn the techniques to make them happen? Or maybe you've been dreaming of shooting in locations that are not accessible to you?
What if I told you that I haven’t been this excited in a while?
What if I told you that, with the help of your computer, you could create, in any style you want, almost exactly what you have in mind, and even more?
In this post, I will tell you about AI-assisted content generation and how I use AI to boost my creativity
AI-assisted content generation
For a couple of weeks I have seen a new kind of dreamy images on Instagram, all showing a hashtag saying #midjourney. I looked it up and learned this is an AI-assisted tool for artists. I had heard about DALL E, an AI system that can create realistic images from a description in natural language. There was a waitlist to use this tool, and I never thought that the results from an AI could be this breathtaking, so I didn't dig into it.
These AI-based art creators take sentences in natural language to create what you have in your mind. From a “prompt”, midjourney creates 4 images that are “low level detailed” that you can then make go through an iterative process to enhance details and resolution.
And this is what can be made! I copied some prompts, let’s see if you can identify which images they correspond to in the midjourney community showcase below :
small stream that passes through the middle of the forest, 8K, ultra realistic, photography, RTX, light mode, octane render
Fantasy Ruins of a castle on a cliff, landscape, sea, breaking waves, dramatic clouds and sunlight, magic, dramatic, epic (..)
cute pink cat with headphones, chunky cat collar with a big bell, long pink pretty hair, well composed, detailed, art by artgerm and Lois van baarle and Ilya kuvshinov (…)
So yes, all these images were created, from scratch by a computer, following what the use told it to imagine! [more on how to get started on midjourney with this cool video]
It's a small statment to say that this opens a door to endless possibilities. I immediately thought of this project as a way to create dreamy backgrounds or inspiration for future photoshoots.
Just imagine: no more need for stock images. You can create your fantasy compositions juste with the help of AI.
Although this might sound like an easy process, sometimes it takes some time to figure out which is the best prompt to use. Especially if you have something specific in mind, you probably won’t get what you wanted at the first try.
so, is this free?
It depends which platform you use. With Midjourney, for now, you get 25 free prompts, then you need to get a subscription. With Dall E, you get 50 free credits during your first month, and 15 free credits will refill every month after that. So I found another AI background generator called photoroom, which is free for now, and I played mostly with this one to get a sense of how to "talk to the AI" to get the best results.
Here are some examples of images I created using these tools:
With photoroom:
Some of the images I generated with Dall E:
And with midjourney (Idid not want to waste my credits just yet, so I only tested a couple of prompts so far). These are images that came out on the first run and were not enhanced.
If you know other tools available online, don’t hesitate to share in the comments below!
How I use AI to boost my creativity
I had my first underwater photoshoot with my friend Ele in Sicily a couple of weeks ago, and while the water is beautiful there, I wanted to give a fantastic vibe to these photos. Using an AI-generated background to enhance the dreaminess of these photos seemed like a nice challenge to start using this technology.
I chose an image I wanted to edit (see below on the left), and I started playing around with prompts with these key words: underwater baroque mirror, nude diver, fantasy world, turquoise and beige light rays, photography, (..), which created images like the one on the right:
Left: raw file from DJI osmo action camera; right: background generated with Photoroom.
Using photoshop, I merged the two photos into this composition.
Here is a quick video of the transformation:
A time saver?
In my opinion, this is probably not the best way to make my creative process faster. AI-assisted image creation is highly addictive and got me stuck on my computer for nights in a row, and as I previously said, sometimes it's tricky to find the right prompt to create the background you have in mind. Once I had the background, making this composition took me about an hour. I never spend so much time in photoshop anymore, but I also never created a "digital art" looking photo.
On another hand, it might speed up the process for photographers who use stock images and have to create the background they have in mind from multiple images. Since AI technologies understand lighting, there is no more need to manually adjust the light of a composite scene to provide coherence to the image. This is a great time saver.
This is also a time saver in a sense that I never had painted in photoshop before. I would have never thought to even try to paint such a background, but with the AI-generated background, I found myself drawn to experiment with all the brushes I had to make my photo match the style of the image I used. And honestly, this really made me want to dive into (pun intended) digital art techniques.
For the aficionados, Photoshop has now released a "neural filters" gallery that is supposed to help you unify the style of different images, but it hasn't worked too well for me so far, I preferred to manually paint over the image.
Here are some more images I created using backgrounds I exported from photoroom and photos I took under water:
Using these photos:
And this AI-generated background and jellyfish to add to the composition. I painted over them to blend them into the same style as the image I was composing in photoshop.
You can view these photos and more compositions in full size in this album.
What’s next?
As I already said, these tools are pretty addictive. After 4 nights, I feel like I only scratched the surface. This is the main reason why I resist to get a Midjourney subscription: I don’t want to get stuck on my computer forever! I have watched a couple of videos on youtube and all photographers say the same thing: within a couple of days they have generated thousands of images.
I personally see it as a boost of my creativity and an opening to so many new possibilities for photographers who won’t have to depend on stock images to create the compositions they have in mind, but I can’t help also thinking that maybe this process takes away from traditional artist who took so much time to learn their craft. This is why I don’t want to post these images as they come out of the AI, no matter how beautiful they are; I want to blend them with my photography, use them as a tool to get somewhere I could not have gone with just my camera and photoshop.
But just so you can see why I have been up at night, here goes a little sample of what I have been playing with. These images came out if the AI as is, with no modification.
All generated with Photoroom. Male characters were designed as a collaboration with my friend Jeremie.
Final thought: this is just the beginning
I am curious to know how you feel about AI art?
I am happy to hear your thoughts on this, the work I made with it, and I can’t wait to show you all the projects I have in my mind coming from this!
Cheers,
Artistically yours,
Adélaïde
ps: Meta-AI can now make AI-videos from prompts or from a couple of photos, and a new beta is coming for Midjourney. I told you it was only the beginning...
コメント