Tools: I used Google Gemini Banana Pro to create the image restorations.

Example 1: New York 1920s

Before:

After:

Example 2: Disco 1980s

Before:

After:

Example 3: Puppy

Before:

After:

Example 4: Czech Immigrant

Before:

After:

The post Restoring Images with Gemini Banana Pro 🍌 (Before/After Examples) appeared first on Be on the Right Side of Change.

Problem Formulation: How to make a wide banner image with Midjourney, i.e., how to change the aspect ratio?

Here’s an example of the standard output for a given prompt:

banner of homer simpson drinking coffee on the beach --v 6.0

You can see that these are not usable as banner images.

To change the formatting to a wider aspect ratio, you can use the --ar X:Y argument where X and Y are width and height numbers, respectively. For example, --ar 4:1 is a good banner aspect ratio where the image has four times more width than height.

Here’s an example of the banner image prompt with the aspect ratio (--ar) argument:

banner of homer simpson drinking coffee on the beach --ar 4:1

Congrats, you’ve created a wide banner image with aspect ratio 4:1. Of course, feel free to make it wider if you need to!

Also check out my following article:

How to Use Midjourney to Create an Avatar by Uploading Your Selfie

The post How to Create a Banner Image with Midjourney (Aspect Ratio) appeared first on Be on the Right Side of Change.

Step 2: Click the + icon or pull the image file right in the input prompt. You want to have uploaded the image in Discord so you can get the image URL.

Step 3: Open the image you just sent and get the image URL via right-click > Copy Image Link.

Step 4: Type the anime/comic/manga avatar prompt adding the --cref suffix:

/imagine headshot image (avatar) in anime style (comic) --cref https://copied-image-link.com

Replace with your own prompt and image URL. The result might look like this:

Congratulations, you’ve created your own anime style, comic, or manga avatar using Midjourney!

Also check out this tutorial:

How to Use Midjourney to Create an Avatar from a Photo URL

The post How to Use Midjourney to Create an Avatar by Uploading Your Selfie appeared first on Be on the Right Side of Change.

Step 2: Type /imagine in the input prompt.

Step 3: Right-click on the original image and copy image link, type the avatar creation prompt, followed by the cref command with the copied image link like so:

Headshot image (avatar) in anyme style and futuristic 3d color --cref https://image-url.com

Replace with your own prompt and image link https://image-url.com

The output for this prompt looks like this:

If you want a more Anime/Manga/Comic style avatar, just change the prompt. For example:

/imagine Headshot image (avatar) in manga comic anime style --cref https://your-image-url.com

The post How to Use Midjourney to Create an Avatar from a Photo URL appeared first on Be on the Right Side of Change.

Image Prompt Example 1: Snapchat Train

 Midjourney version 6:
/imagine snapchat image of a person sitting in train

Here’s an image created by Midjourney:

 GPT-4 (ChatGPT)
create image: snapchat image of a person sitting in train

Here’s an image created by ChatGPT:

Current Score: Midjourney 1: ChatGPT 0

Image Prompt Example 2: Soccer Lego Lady

 Midjourney version 6:
/imagine female soccer player (lego)

Here’s an image created by Midjourney:

 GPT-4 (ChatGPT)
create image: female soccer player (lego)

Here’s an image created by ChatGPT:

Both are great images, so I’ll give a +1 for both models:

Current Score: Midjourney 2: ChatGPT 1

Image Prompt Example 3: Klimt Hacker

 Midjourney version 6:
/imagine painting of a programmer with hoodie in the style of Klimt

Here’s an image created by Midjourney:

 GPT-4 (ChatGPT)
create image: painting of a programmer with hoodie in the style of Klimt

Here’s an image created by ChatGPT:

Although the ChatGPT image has elements of Klimt, the programmer himself doesn’t use the “Klimt style” at all, so the overall image is flawed. Midjourney respected the request much better:

Current Score: Midjourney 3: ChatGPT 1

Image Prompt Example 4: T-Bone Steak Shuttle

 Midjourney version 6:
/imagine a tasty T-Bone steak flying through space with microbacteria living on the T-Bone space shuttle --v 6.0

Here’s an image created by Midjourney:

 GPT-4 (ChatGPT)
create image: a tasty T-Bone steak flying through space with microbacteria living on the T-Bone space shuttle

Here’s an image created by ChatGPT:

Although the Midjourney version looks slightly more spacy, ChatGPT respects my prompt better:

Current Score: Midjourney 3: ChatGPT 2

Image Prompt Example 5: Broccoli Train

 Midjourney version 6:
/imagine feature rich child image of a broccoli train driving through milk galaxy

Here’s an image created by Midjourney:

 GPT-4 (ChatGPT)
create image: feature rich child image of a broccoli train driving through milk galaxy

Here’s an image created by ChatGPT:

ChatGPT wins this one — getting even the green/white milk/broccoli mix right. Also the train itself partially consists of broccoli.

Current Score: Midjourney 3: ChatGPT 3

Surprisingly, there was no clear winner. I expected Midjourney to win this battle but subjectively, I cannot really see this confirmed with my little experiment.

Also check out this article:

Midjourney V5 vs V6: 10 Portrait Examples That Show the Differences (Before/After)

The post Who Wins in Image Generation? Midjourney vs ChatGPT: 5 Examples appeared first on Be on the Right Side of Change.

We use the Midjourney /describe prompting technique to create image prompts based on the actual house owned by Bill Gates (source).

Here’s the result for the first prompt:

The result of the second prompt:

The third prompt:

The fourth prompt:

I believe this is the best representation of Bill Gates’ mansion, as generated by Midjourney:

Let’s try a second image of Bill Gates’ House:

Click “Imagine all” and get the following pics:

I believe the most accurate one is this:

This was just a fun exercise, to learn about all Midjourney prompts, check out my blog here:

13 Best Midjourney Parameters with Examples (List)

The post Reverse Engineering Bill Gates’ $227 Million Mansion Using Midjourney appeared first on Be on the Right Side of Change.

Here is a list of some of the parameters:

1. Aspect Ratios (--aspect or --ar): This parameter is used to alter the aspect ratio of the generated image. It supports any ratio, but over 2:1 is experimental and may cause unpredictable results.
2. Chaos (--chaos or --c): This parameter changes how varied the results will be. Higher values produce more unusual and unexpected generations. The range is from 0 to 100, with the default being 0.

3. Stylization (--stylize or --s): This parameter influences the extent to which Midjourney’s default aesthetic style is applied to your images. It accepts values from 0 to 1000, where higher values mean that the image will have more stylization and abstraction, while lower values mean that the image will have less stylization and abstraction.
4. Detail (--quality or --q): This parameter influences the level of detail in the generated image. The default is 1, and it only accepts .25, .5, 1, and 2. Values greater than 2 are downgraded to 2.
5. Seed (--seed): This parameter allows you to use a specific seed number to create a field of like television static, as a starting point to generate the initial image.
6. Image Weight (--image weight or --iw): This parameter influences how much weight is given to the image in the generation process. Defaults and ranges differ by version.
7. No (--no): This parameter enables negative prompting. It allows you to remove an element from the image by specifying what you don’t want to see in the image.

8. Style (--style): This parameter allows for switching between different versions of the Midjourney model.
9. Version (--version or --v): This parameter allows you to use a different version of the Midjourney algorithm.
10. Stop (--stop): This parameter allows you to stop Midjourney from generating the image when it’s a certain percentage done. The earlier you stop the generation, the more the image looks like it has a Gaussian blur.
11. Niji (--niji): This parameter allows you to use an alternative model focused on anime style images.
12. Uplight (--uplight): This parameter allows you to use an alternative “light” upscaler when selecting the U buttons. The results are closer to the original grid image.
13. Weird (--weird or --w): This parameter allows you to add a weirdness factor to your image. Higher values produce unexpected generations.
14. Tile (--tile): This parameter allows you to create a repeating pattern (see below).

Remember, parameters are always added to the end of a prompt, and you can add multiple parameters to each prompt.

Recommended: Midjourney Tile Parameter — 5 Beautiful Examples (v6)

The post 13 Best Midjourney Parameters with Examples (List) appeared first on Be on the Right Side of Change.

Example 1: Watercolor Koi –tile

Prompt: watercolor koi --tile --v 6.0

Example 2: Mona Lisa –tile

Prompt: mona lisa --tile --v 6.0

Do you see the beautiful transitions between two aligned images? Note that the single output from the prompt was this image, and I arranged it as a tile:

Example 3: Soccer Girl –tile

Prompt: soccer girl --tile --v 6.0

Example 4: Fairy Tale –tile

Prompt: boy and girl fairy kiss each other on leaf in spring --tile --v 6.0

Example 5: Python Snakes –tile

Prompt: python snakes --tile --v 6.0

Bonus Example: Lord Voldemort –tile

Prompt: lord voldemort --tile --v 6.0

The --tile parameter is compatible with Midjourney Model Versions 1, 2, 3, test, testp, 5, 5.1, and 5.2. However, it does not work with the Niji model or Version 4. It’s important to note that the --tile parameter only generates a single tile. To visualize how the tile repeats and forms a pattern, you would need to use a pattern-making tool like a seamless pattern checker.

I used PowerPoint for this article (copy and paste)!

What Are Some Example Prompts That Can Be Used with the Midjourney Tile Parameter?

The Midjourney --tile parameter can be used with a variety of prompts to generate images that can be used as repeating tiles to create seamless patterns.

Here are some examples of prompts that can be used with the --tile parameter:

1. "scribble of moss on rocks --tile"
2. "watercolor koi --tile"
3. "beautiful fantasy map in the style of Lord of the Rings --tile --c 1"
4. "3d tetris, white background --tile"
5. "modern minimalist pattern --tile --v 5 --s 250"
6. "elephant retro seamless pattern, nursery decor, simple, 16k resolution --tile --s 500"
7. "monstera leaf seamless pattern --tile --q 2 --s 500"
8. "100 x 100 tiled game map with large grasslands and river, top down view, styles of SNES --tile"
9. "game map with varied terrain, style of Advance Wars for Game Boy Advance --c 1 --tile"
10. "frog pattern, breathtaking illustrated, simple, featured on pixiv, muted colors indigo dark with minimalism, irina nordsol kuzmina, a hazy memory --ar 16:9 --tile"
11. "peonies and buds, Baroque Jean-Honoré Fragonard style, seamless wallpaper, pastel colors on a teal blue --tile"
12. "a geometric pattern --tile"
13. "a floral design --tile"
14. "a starry night --tile"
15. "Ancient magical forest, super mario level, pixel art, by Dan Mumford –ar 2:1 –tile"
16. "seamless pattern that features small cute pastel colored forest animals --stylize 250 --tile"

Remember, the --tile parameter only generates a single tile. To visualize how the tile repeats and forms a pattern, you would need to use a pattern-making tool like a seamless pattern checker.

Feel free to check out the following Midjourney article as well:

Recommended: Midjourney V5 vs V6: 10 Portrait Examples That Show the Differences (Before/After)

The post Midjourney Tile Parameter — 5 Beautiful Examples (v6) appeared first on Be on the Right Side of Change.

In this article, we’ll look at something more boring but 10x more useful: prompting Midjourney to generate BORING images from everyday people or situations.

Images that 99.9% of people wouldn’t suspect are created by an AI.

I WILL GIVE YOU THE PROMPTING TRICK AT THE END OF THE ARTICLE!

Let’s start with a series of 10 Midjourney, 100% AI-generated images that are boring as F$%K and nobody would assume these are AI generated.

Here’s the prompt that accomplishes these images — replace X as you need:

Prompt: phone photo of [X] posted to reddit in 2019, --style raw --s 0 --ar 9:16

Every word in this meta prompt is designed to make the images look as normal and, yes, boring and everyday as possible. I found this prompt here.

The post The Boring Midjourney Prompt: Create Random Boring Photos of Everyday Situations appeared first on Be on the Right Side of Change.

Altman’s Mansion

Original image.

First, I passed this image into ChatGPT and asked it to create an image prompt.

ChatGPT suggested this prompt:

 ChatGPT Image Prompt

Create an image of a luxurious multi-level house perched on a hillside during twilight. The house should have a traditional upper section with dark wooden siding and a modern lower section with expansive glass walls and concrete details. The interior lighting should be visible, casting a warm glow against the dusk sky. Lush greenery surrounds the house, including well-maintained bushes and mature trees. Include a set of sleek, concrete stairs with glass railings leading up to the entrance. The backdrop should feature a gradient of the evening sky transitioning from blue to hints of orange at the horizon, and the foreground should have a slight urban feel with the presence of distant city buildings.

Feeding back the image prompt into it, ChatGPT creates the following image:

Looks similar already.

Now, passing the image into Midjourney, it comes up with these image prompts:

These are the 16 images (4×4), Midjourney generates when passing these image prompts back into Midjourney:

While all images are pretty beautiful, I don’t think they capture the essence of the real villa in the Lombard Street that’s more classy and less modern.

Passing the ChatGPT prompt into Midjourney also doesn’t help. It creates these images that are still pretty different:

Let’s try for the living area of Sam’s $27 million mansion next!

Altman’s Living Area

Original Image

Passing this prompt into GPT-4V generates the following image:

Not so bad!

Next, I use the /describe command in Midjourney to create a prompt based on the given image. It creates the following prompt:

Wow, Midjourney figured out that San Franciso is the city in the background! This is amazing – DALL-E couldn’t do that.

Hitting “ Imagine all” creates the following 4×4 image suggestions:

The images are beautiful but very different from the original. Let’s try the sauna next!

Altman’s Sauna

Again, I’ll start with ChatGPT (GPT-4 Turbo):

Original Image

ChatGPT recreates this image:

Even prettier than the original!

Let’s try Midjourney on Discord:

This gives us the following four prompts:

Hitting Imagine all again:

Great inspirations! Remind me again: What do we need architects for when generative AI can create stunning designs like these? Just kidding.

Check out the following learning path to stay on the right side of change and participate as a tech disruptor!

Recommended: 10 Steps Learning Path: Becoming a Python Prompt Engineer on the Finxter Academy

The post Midjourney vs ChatGPT: Reverse Engineering Sam Altman’s $27-Million Mansion in San Francisco appeared first on Be on the Right Side of Change.