An AI modeling agency that describes itself as a “photo studio” without cameras, real people, or a physical location has opened its virtual doors.
Deep Agency is “100% AI-generated” and uses the very latest image synthesis models allowing users to pick a model and choose their pose.
Virtual photographers can even choose which “camera” to use, the time of day, the shutter speed, and the lighting style. The virtual cameras available include the “Fujifilm XT3, Canon EOS Mark II, and a Sony a7.”
It’s not just generative AI models the site provides, for $29 a month users get full access that includes high-quality photos of themselves in different settings and styles, including nonrealistic.
“Hire virtual models and create a virtual twin with an avatar that looks just like you. Elevate your photo game and say goodbye to traditional photo shoots,” the site reads.
After MONTHS of work, it's finally here!
🚀 Deep Agency: AI photo studio & modelling agency
Full explanation in the next few tweets ↓ pic.twitter.com/aMOS76FFiL
— Danny Postma (@dannypostmaa) March 6, 2023
Danny Postma, a Dutch entrepreneur, is the man behind Deep Agency and is marketing it to social media influencers, marketers, and e-commerce product photography.
Postma has been boisterous on Twitter with his below Tweet receiving 22 million views and a generally negative reaction from the platform’s users.
Both of these models do not exist.
— Danny Postma (@dannypostmaa) March 5, 2023
‘Show us the Hands’
In the majority of the examples shown on Deep Agency’s website, the hands and teeth of the AI models are hidden.
“From now on if I don’t see hands or teeth in a photo, I’ll assume the person in said photo does not exist,” writes one Twitter user.
I'm personally not excited about this. From now on if I don't see hands or teeth in a photo, I'll assume the person in said photo does not exist. https://t.co/Ig6CSa4g8H
— star•gyal (@beequammie) March 9, 2023
Last week, PetaPixel explained why AI image generators struggle with hands so badly.
Part of the reason AI generates hands with multiple fingers is to do with the training data. Simply put, humans display their extremities far less frequently than their faces.
Image credits:Images courtesy of Deep Agency.