which ai image generator is the best?
The battle for best in the AI world is a hot topic. It really is a new brave world and as a storyteller I can tell you this new tech is game changing. So maybe you want to dip your toe in? I’ve always been the person who likes to swan dive, so here is what I’ve learned.
You start with a prompt just a few words and your ai pals open up. I have been working on a mini cooper spec and here is the prompt I used for my opening shot:
“a cinematic mini cooper car commercial an 18 year girl with flowing red hair, she is wearing a white graduation cap and gown. she is standing next to a blue mini cooper car, the car has a bow on the roof like it is a graduation gift --ar 16:9”
(16:9 is the aspect ratio- you can make a square still for social by using 1:1)
First Up! Old Faithful, my first and most often best choice for AI text to image is Midjourney
Midjourney has a cinematic quality and yes, sometimes it makes a monster woman with crazy hands, for the most part it’s solid and eventual with repompts and versions you can get there.
Next! This is runway I usually don’t make text to images in this program, as you can see below it looks like she is going to murder us with her very old haunted mini cooper from a country unknown.
That being said, runway is the best program I have found, so far, for adding motion to stills that are realistic. You have so much control with painting sections and camera movements.
Adobe’s has been in all the headlines for adding so many cool options with generative autofill in Photoshop and in Premier Pro the new tool that allows you to remove objects while editing is simple and amazing. Their text to image from firefly below gets a shout out for diversity, the other programs are not as good at that. I do like the image but it’s not clear that it is a mini cooper in the pic, it looks like generic blue car. in general though, adobe is doing amazing things.
Then there is DaVinci our girls is looking like a sexbot nurse who is stuck on a dirt road. Not the family vibe I was looking for with a mini cooper ad. But DaVinci is really great with creepy, cool, surreal images for horror and sci fi work.
And while we all liked, Mark Zuckerberg’s awesome chain for the meta ai announcement…
not all heroes wear capes: marky z’s meat ai moment
…the stills not so much. She’s lovely, but she is literally part of the car. And most of the meta ai still have the Polar Express effect. Shout out to Hanks!
Google Cloud’s vertex ai gave me a script, no matter what I did and it was confusing. Maybe it was user error (probably) but if after 10 minutes of me trying to figure it out I lost interest. It was not very user friendly. Nice Script though.
So for me, I am sticking Midjourney with for now. But every week there is a new app, an update or a new program either way I’ve loved learning about it all and go check out my new site to see what I’ve been up to.