Creativity AI #33: Weekly update for AI art
News about AI art, Changes in Geeky Curiosity, Prompt Play, Inspiration
Creativity AI is a free weekly publication that brings together interesting articles and recent developments in AI art and creativity.
You can also read this newsletter on the website.
News about AI art
Midjourney office hours update: What's coming up next
Video features launching imminently
Start and end frame control for precise video creation
Video looping capabilities
720 pixel mode for higher resolution output quality
Potential turbo mode
Expected launch: today or tomorrow (yes, that soon!)
Example of a 720p video I generated yesterday.
This issue is also dedicated to Fido, which reminds me of my dog when I was still a vet student many years ago. He’s waiting for me at the Rainbow Bridge. (Fido at 720p)
Take a Midjourney survey and get FREE 2 fast hours
Here: https://www.midjourney.com/surveys?survey=demographics
Take a 5-minute survey about the "origins of beauty"
Get 2 fast GPU hours as a reward
Survey helps Midjourney understand aesthetic preferences globally
Currently seeking more participants from underrepresented countries like Japan
Comment: I'm not sure how 99 minutes is equivalent to 2 hours? But that's okay; I'm happy to take the survey even if Midjourney doesn't reward me anything.
Midjourney’s official YouTube channel
Midjourney finally launched their official YouTube page: https://www.youtube.com/@Midjourney
Worth checking out and following them for their updates
Comment: I also have a YouTube channel where I explore and create AI videos. If you are wondering. It's for me to experiment with different concepts and AI video making techniques. https://www.youtube.com/@AnimalsGeeky/videos
Midjourney’s medium to long-term projects
Style Explorer on the horizon
Will roll out after the video features launch
Version 8 development update
Midjourney team describes V8 as a "ridiculously good model"
Much better architecture than V7 for building additional features
V7 struggles with coherence and reference issues that V8 should solve
Possible V7.1 release as a stepping stone before V8
Timeline estimate: potentially by end of this year (not sooner)
API and enterprise considerations
Currently reviewing survey results for enterprise options
Considering smaller enterprise tiers
Possible personal version with fewer features still under consideration
Nothing confirmed yet
Future improvements in development
New upscaling system planned (David isn't satisfied with current options)
Built-in editor coming with V8
Video background sound generation eventually planned
David mentions liking Suno and 11 Labs, but no collaboration plans yet
Midjourney management philosophy & decision
Midjourney weighing closed system approach (like Apple) versus openness
Want to maintain high standards while considering ecosystem accessibility
David says he's "not a fan of Apple" but recognizes their approach has merit
Runway recently revealed the Act-Two feature, which allows users to create expressive character performances using motion capture. In addition to facial expressions, it now supports hand and upper-body tracking. It means you have to act, send the video to Runway, and it will create the lip and body sync with the video character for you. You can also upload a Midjourney video and have Act-Two transfer your performance to your character.
Hypernatural raised $9.2 million in funding for its AI video creation platform, which streamlines video production for producers and everyday people by generating 54-second videos from text prompts in under a minute. I tested it, and while it can make long form video, it now appears to be more of a slideshow with audio narration, and the image quality is terrible. Maybe they're still in the early stages.
Update from Geeky Curiosity
The Midjourney Style Resources #26 has been updated to V7.
This update is to showcase the latest improvement of Moodboard aesthetics.
Hedra Character-3 brings digital characters to life
Hedra Character-3 really caught my attention with its incredibly natural lip-sync and body movements. What's even cooler? It animates backgrounds too, making everything feel more realistic.
I was surprised it works brilliantly with non-human characters, blurry shots, and medium-distance scenes. Not just close-up talking heads like most tools.
So I decided to test it with our resident characters at Geeky Curiosity. Let's see what our Reptilian Sorcerer, Camille in Paris, and Elena at the Space Station think about this new tech!
(The audio clips were generated with ElevenLabs Voice Design V3.)
Reptilian Sorcerer
Hedra is ancient sorcery.. Use it with care.Camille in Paris
Merci, Hedra ! Now I can chat and stroll at this lovely Paris café, while my cat pretends to judge everyone walking by. He’s quite the critic, you know!Elena at the Space Station
Well, Hedra, you want me to do more but there’s no pay? Sorry, that’s above my unpaid grade!Prompt Play
I tested ChatGPT, Midjourney v7, and Ideogram with the same challenge: generate 16 different household objects in a top-down view. The results are quite interesting.
ChatGPT-4o
ChatGPT nailed it completely. The objects appeared correctly, plus it threw in an extra key and eraser I didn't even ask for. The layout was clean, and each item was easily recognizable.
top-down view of a tidy grid layout of everyday objects, toothbrush, mug, scissors, key, spoon, candle, stapler, coaster, remote, comb, pen, tape, eraser, charger, magnet, ruler, arranged on a white surface with even spacing and soft shadows. Aspect ratio 16:9
Midjourney V7
Midjourney created beautiful images but struggled with prompt accuracy. It's caught in an interesting tension between following instructions and making things look good. Even using the --style raw parameters didn't help much, as several objects were missing entirely.
This highlights Midjourney's philosophy that it doesn't prioritize strict prompt following, as it believes doing so doesn't create the most visually appealing results.
top-down view of a tidy grid layout of everyday objects, toothbrush, mug, scissors, key, spoon, candle, stapler, coaster, remote, comb, pen, tape, eraser, charger, magnet, ruler, arranged on a white surface with even spacing and soft shadows --ar 16:9 --v 7 --p uw4q9o7 
Ideogram 3.0
The Ideogram showed many items, but they looked unusual. Spoons somehow merged with keys, creating confusing hybrid objects that didn't match the original list.
top-down view of a tidy grid layout of everyday objects, toothbrush, mug, scissors, key, spoon, candle, stapler, coaster, remote, comb, pen, tape, eraser, charger, magnet, ruler, arranged on a white surface with even spacing and soft shadows. Aspect ratio 16:9 Magic Prompt: Off
My takeaways
For multiple object generation, ChatGPT wins hands down. Midjourney and Ideogram both generate some items correctly but miss the mark on accuracy.
When you need precision over pretty, ChatGPT's your best bet.
Inspiration
"Overemployment" (OE) or polyworking is having a moment, but it's not exactly new.
I know several people secretly juggling multiple full-time jobs. Sure, extra income is nice, but that's not always the main driver. Sometimes it's about feeling undervalued at your current role or being fed up with corporate nonsense.
These folks want to explore opportunities where their talents get recognized without burning bridges at their day job.
Here’s my Midjourney exploration inspired by OE.
surreal split-screen image, one person living dual-career life, corporate vs creative, visualizing overemployment, hybrid environments, symbolic and moody --ar 16:9 --profile oaefodl --v 7
surreal split-screen image, one person living dual-career life, corporate vs creative, visualizing overemployment, hybrid environments, symbolic and moody --ar 16:9 --profile oaefodl --v 7
simple chair with glowing binary tree branching upward, digital minimalism, subtle metaphor for polyworking and overemployment, clean composition --ar 16:9 --profile oaefodl --v 7
abstract human silhouette, clean lines, fragmented mirrored reflections, representing overemployment and identity division, minimalist white space --ar 16:9 --profile oaefodl --v 7 
digital clone working at desk, surrounded by translucent ghost-copies, all performing tasks, neon lighting, abstract representation of overemployment --ar 16:9 --profile oaefodl --v 7 
Recent articles
Related articles
Free Geeky Animals' Sref code: 3165206288
Prompts are created using MJ Simple Prompt Generator (+ my brief idea)
A teenager walking a bicycle through knee-deep fog inside a neon-lit grocery store, shelves fully stocked, mist swirling with eerie hues, frozen moment of confusion --ar 16:9 --sref 3165206288 --profile oaefodl --sv 6 --v 7
A woman in a wedding dress waiting at a deserted bus stop under pouring rain, bouquet in hand, mascara running, stark streetlight casting dramatic shadows --ar 16:9 --sref 3165206288 --profile oaefodl --sv 6 --v 7
A businessman in a suit staring at a floating briefcase in a quiet laundromat, surreal lighting, still atmosphere, subtle reflections on the tiled floor --ar 16:9 --sref 3165206288 --profile oaefodl --sv 6 --v 7
Cover prompt: A golden retriever wearing sunglasses seated at a dinner table with untouched meals, candles lit, napkin tucked in, human expressions on the dog’s face --ar 16:9 --sref 3165206288 --profile oaefodl --sv 6 --v 7
I hope you like this article!
Thank you for reading and happy creating!