Allen T. Profile picture
Aug 30, 2023 9 tweets 4 min read Read on X
Huge strides have been made for creating consistent characters with Midjourney.

Using an inpainting technique, we can go further.

With this workflow, you can bring a character into new scenes with different expressions.

This is big. Especially for video storytelling. Image
1) First, let's start with the amazing prompt from @chaseleantj and @juliewdesign_

Prompt: [Subject], split into 3 different images, shot from multiple angles --s 50

Important: You want three (3) panels for this workflow.
2) Once you have a generation that you like, upscale it.

Reminder: You want a three panel image. Preferably with a large panel and two small panels. Brunette woman, wide shot, kodak ultra f9, film grain, split into 3 different images, shot from multiple angles --s 50
3) With your upscaled image, click on the vary region button. Using the square selection tool, select only one panel. Make sure your selection slightly extends past the edges of the panel.

Important: Only select one, or at most, two panels. Do not select the entire image. Image
4) Edit only the subject of the prompt and change the stylization to low.

For example, if your subject is a brunette woman, you can change the subject to brunette woman at dinner. Do not change any of the remainder of the prompt.

Important: Add --s 50 to the end. Image
5) This is where the magic happens. It appears that by selecting one panel for inpainting, Midjourney will reference the subject in the other panels.

If you are having issues, try selecting a different panel and re-running the inpainting prompt. Also make sure to use --s 50.
Brunette woman at dinner, wide shot, kodak ultra f9, film grain, split into 3 different images, shot from multiple angles --s 50
Brunette woman at dinner, wide shot, kodak ultra f9, film grain, split into 3 different images, shot from multiple angles --s 50
6) To go a step further, upscale one of the inpainted versions and click the vary region button again. This time select every panel besides the one that you like.

This is less stable, but with some re-rolls you can create multiple images of your subject at the location.
Image
Image
7) If you would like to extend the height and width of your panels, I recommend using adobe's Gen Fill for this task.

This last step is optional.
Image
Image
Thank you for reading along! If you enjoyed these posts,

1) Follow me @Mr_AllenT for more AI content and tutorials.

2) Bookmark, like, and repost to share with your friends!

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Allen T.

Allen T. Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @Mr_AllenT

Jul 21, 2025
Sitcoms that don't exist

But I wish they were real.

1. Lord of the Rings: 🧵
2. Super Mario
3. Star Wars
Read 13 tweets
Jul 19, 2025
Retro games that don't exist

But I wish they did.

1. Lord of the Rings: 🧵
2. The Godfather
3. Game of Thrones
Read 12 tweets
Jun 14, 2025
One AI fashion model. Endless outfits...

Here's how:
1) Getting started

Head to @ai_affogato & select studio to get started

app.affogato.ai
2) Creating a character

For this example, I generated a fashion model with Flux on Affogato to use for the outfit swaps. However, you could use a real photograph as well.

Prompt: "A wide full body photograph of a beautiful fashion model that is asian with short bob cut orange hair. The background is solid white. Wearing business dress and heels"Image
Read 7 tweets
Jun 25, 2024
Toys "R" Us has released the first OpenAI SORA generated brand commercial

The use of generative AI in commercial work is here.
Full details:


@visiblemakers @ChristianF369 with the scoop!toysrus.com/pages/studios
Read 5 tweets
Jun 12, 2024
Luma AI dropped their Sora competitor & it's insane

The best part? They are actually letting people use it

10 awesome examples with prompts:
1) Prompt: Man with a television head walking through new york city times square

Note: I created each generation in this thread with Luma's text2image or image2image feature

#LumaDreamMachine
2) Img2Img + Prompt: Zoom in shot of a demon standing in front of a massive fire flapping it's wings
Read 16 tweets
May 7, 2024
Aging Morphing 🧒👴

Using the Img2Img feature on @getremixai & frame interpolation on @runwayml we can age characters in less than 5 minutes

I will teach you how in the posts below 👇

#RemixAmbassador
1) Create a character

Using the RealVisXL model I generated a starter image for this process

Prompt: A teenage man, close up shot, macro Image
2) Img2Img on Remix

On @getremixai go to the Img2Img feature and use your starter image as the base image and change the prompt to change the character age

Example:
Prompt: A 35 year old man, close up shot, macro

Repeat this process for each age you like to change your character to. Download all of the images for the next steps

See prompts in ALTSA young boy, close up shot, macro
A 35 year old man, close up shot, macro
A 60 year old man, close up shot, macro
A 80 year old man, close up shot, macro
Read 7 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

:(