Creativity AI #15: Weekly dose of AI updates

Midjourney Office Hours (2025-04-09), news, prompt play, and inspiration

Apr 09, 2025

Available for iOS and Android

Creativity AI is a new publication that brings together interesting articles and recent developments in the AI world related to creativity and productivity.

Midjourney Office Hours (2025-04-09)

The new version is still in its early stages, and Midjourney is collecting feedback and bug reports. More image ranking parties are coming.

You will see a feedback box in the lightbox where you can provide feedback to Midjourney. Click on any of your generated images to see it.

You can tell Midjourney right away whether or not you like the generated image.

There will be a release/update every week for the next few months until the full version of V7 is ready and becomes the default.

Upcoming releases:

Today/soon: Add—tile, /remix, and a better Enhance Draft mode.
Next week: speed update.
Next two weeks: Upscaler and Omni-reference.
Next month (or more): Inpainting and Retexture.

Midjourney is migrating its servers (data centers) to speed up development and growth. The transition is taking place gradually as new servers are set up.

We can expect a 4x speed increase at no additional cost by the end of the year as a result of the server migration.

David affirms: "Our models are always worst in prompt understanding but best in aesthetics."

Update from Geeky Curiosity

The MJ Simple Prompt Generator has been updated to Midjourney Version 7. It now supports the latest Midjourney V7 Personalization codes. Members have lifetime access to this easy-to-use prompt generator, which I’ll continue to update - because I use it regularly to save time and mental energy! :-)
- For example, you can write a simple prompt: “a happy bunny celebrating Easter 16:9 vumoqva” (where 16:9 is your Aspect Ratio and vumoqva is the V7 Personalization code). No need to write the parameters in details. The bot will understand and generate 5 prompts in various artistic directions. Take this example:
- This is generated prompt:
```
a cheerful bunny jumping with a decorated Easter egg under a sunny spring sky, playful mood, rich textures, lush grass --ar 16:9 --profile vumoqva --v 7
```

The result of using Midjourney MJ Simple Prompt Generator in V7

The format of Creativity AI will be revised to make the content more concise and visual. Longer, more focused content will be published in separate articles.
I'd like to know what you think. I'll consider your feedback seriously when planning future content.
Loading...

News

Runway introduces Gen-4 Turbo, which generates a 10-second video in less than 30 seconds. This allows for rapid ideation and iteration during creative exploration.

ElevenLabs showcased on X (Twitter) how they created a voice agent and ordered a pizza with extra toppings. (oh, they also introduced their official MCP server).

Anthropic released an education report detailing how university students use Claude. Computer science students are early adopters of Claude. Students primarily use Claude for higher-order cognitive tasks such as creation (using information to learn something new) and analysis (breaking down the known and identifying relationships).

Perplexity offers a Startup Partners program for startups under 5 years old that have raised <$20 million in equity funding to receive $5000 in API credits and 6 months of Enterprise Pro. Apply here.

Google launched a free interactive Learning Light studio to teach you how to be creative with lighting design.

An example of Gen-4 Turbo video using Midjourney 7 image.

Prompt Play

I read that some people use ChatGPT-4o to create an image with 150 items. However, the outcome was not particularly good. Some items at the end of the list are very similar. Some are distorted beyond recognition.

So, I'm curious how many items in graphical forms ChatGPT can accurately represent in a generated image.

Here's my first prompt to generate 150 numbers.

Create a string of 150 numbers counting from 1 to 150.

Then, I ask ChatGPT-4o to generate an image with 150 numbers in it using this prompt:

Create an image in the style of knolling photography of items carefully laid out and photographed from directly overhead, where EACH and EVERY item listed is clearly seen ONLY ONCE: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150

Here is the result:

The first row has some immediate issues with two 5s and a distorted number 6. Numbers 11–77 are OK. Then 78 was replaced by 88, and there are numerous errors near the end. The bot appears to be having some issues at the beginning and end of the image generation process.

I try again, this time with fruit. Just fifty items.

Here's my prompt.

Create an image in the style of knolling photography of items carefully laid out and photographed from directly overhead, where EACH and EVERY item listed is clearly seen ONLY ONCE: Apple, Banana, Orange, Strawberry, Grape, Watermelon, Pineapple, Mango, Peach, Pear, Blueberry, Raspberry, Blackberry, Cherry, Kiwi, Lemon, Lime, Grapefruit, Plum, Apricot, Coconut, Avocado, Pomegranate, Fig, Papaya, Guava, Lychee, Dragonfruit, Passion Fruit, Cantaloupe, Honeydew, Starfruit, Durian, Jackfruit, Kumquat, Mulberry, Gooseberry, Elderberry, Date, Mangosteen, Kiwano, Salak, Feijoa, Soursop, Longan, Sapodilla, Cherimoya, Custard Apple, Jujube, Miracle Fruit, Acai Berry, Goji Berry, Cloudberry, Lingonberry, Ugli Fruit, Yuzu, Ackee, Calamansi, Jabuticaba, Santol, Salmonberry, Tayberry, Huckleberry, Bilberry, Barberry, Sea Buckthorn, Jostaberry, Crowberry, Hackberry, Gac, Cupuaçu, Buddha’s Hand

And this is the result:

Immediately, I noticed that an orange (third item on the list) was missing. Same for grapes and watermelons. Some of the items on the list, particularly the berries near the end, appeared to be very similar.

Based on this small experiment, I conclude that one cannot rely on ChatGPT-4o to generate accurate items in an image, especially when there are too many items. (I tried another 9-10 item image and it worked fine.)

Inspiration

I learned a new word today: "baksheesh," which means a tip, present, or gratuity. Do you recall the good old days when you could prompt one word in Midjourney and see what the bot created? How about trying Midjourney V7? I am intrigued.

baksheesh --ar 16:9 --profile vumoqva --v 7

a warrior who is ready to go to war, baksheesh --ar 16:9 --profile vumoqva --v 7

a woman warrior who is ready to go to war, baksheesh --ar 16:9 --profile vumoqva --v 7

My conclusion is that “baksheesh” is a Midjourney keyword for creating a character with headgear and a scene associated with Persia because the word originates from Persian.