Skip to main content

Scenario 8: Human

Celebrity Photos​

This is a new V5 feature - Midjourney now recognizes many celebrities. You can generate photos of them by adding their name when describing the subject. Let's generate some classic movie shots:


The first is Keanu Reeves, the second is Vito Corleone (I didn't input the actor Marlon Brando). The prompts are very simple - try writing them yourself!

Answers at the end of the chapter.

Clothing Model​

Those in ecommerce may find this useful:

  1. Use AI to generate a virtual human
  2. Input a product photo (e.g. clothing) into the AI
  3. Have the AI blend the virtual human and product together

Frankly, I don't think Midjourney is quite there yet, but it is possible. It just requires the blend technique from Technique 9 - I won't elaborate, but basically blended a product and human image:


To verify consistency, I tried several images of Keanu:


It seems decent overall, but likely not robust enough for product shots, with noticeable differences from the original clothing (see buttons).

I think future Midjourney updates may improve this capability. And with custom model training, it seems achievable.

Old Photos​

This is an offshoot of celebrity photos that you may have seen - the prompts are simple, translate as needed:

What is the type?photography
What is the subject?a group of Chinese people gathered around Darth Vader
What is the background?on the street
What is the composition?nullNot too important, let AI decide
What lens?fuji filmAdded for vintage look
What is the style?style of 1990sAdded 1990s style

Similar prompts generated these images:


Tip 14: Add Styles - Time Periods​

The old photos used 1990s style - strictly speaking, decades fit better in subject or environment, but works in style too since we want both 90s subjects and a 90s style.

This is the last style word usage - time periods.

Modifying the illustration prompts, I didn't specify artists, just style of 1920s. The rose has the right vibe, the cat top right is hilarious.


Tip 15: How to Generate More Diverse People​

Diversity here refers to people's height, weight, body types, disabilities etc.

You may notice AI generated people tend to be very attractive. This reflects societal biases. Here are tips to get Midjourney to generate more diverse bodies.

Method 1: Slider

The weight method from last chapter also works for generating different body types. Here's an official example:

Plus-sized ginger female dressed in bohemian vest, maxi skirt, espadrilles, walking across autumn campus courtyard, morning atmosphere:: plus-sized::-0.3 --ar 2:3

Adjust the plus-sized parameter to make the person slimmer or larger. Note negative doesn't always mean thinner - if your word is thin, a negative slider value will make them larger:


Method 2: Direct Description

Midjourney recognizes some diversity terms like:

  • blind
  • deaf
  • cerebral palsy
  • accessibility


Answers of Celebrity Photo​

Keanu is simple - describe subject, iconic outfit, walking in rain:

Keanu Reeves, wearing a black long leather coat, walking down the street in the rain

And the Godfather one is even simpler:

Vito Corleone sitting on a chair