Creativity AI #34: Weekly update on AI art and creativity
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 new releases
The Explore "For You" feed got smarter. Instead of showing you random content on the Explore page, there's now a recommendation algorithm that actually learns what catches your eye. The page is particularly helpful when you want to select the images for your personalization profile.
Midjourney.TV launched today. There's a streaming site that plays user-generated Midjourney videos around the clock. It automatically fits your screen whether you're on your phone or computer, and you can hover over the creator's name to see the prompts that created them.
Watch it here: https://www.midjourney.tv/
Midjourney upcoming video features
Next week brings three new video modes that should make creators happy:
2.5-second mode costs half as much as longer formats and works great for quick animations and loops.
720p mode cranks up the resolution for professional work. Think TV commercials or music videos. It's starting as a higher-tier feature because the computing costs are brutal.
Turbo mode spits out videos in roughly 10 seconds, which is fantastic for rapid testing. You'll need a $60+ subscription since it's expensive to run.
There's also an "extend from any frame" feature coming that sounds pretty promising.
Midjourney medium and longer-term plan
Version 7.1 may arrive within the month. The highlights: better text understanding (finally!), improved coherence, stronger style reference capabilities, and an upgraded draft mode.
Version 8 has an architectural overhaul, which means both images and videos get major improvements.
Other plans
Midjourney developers are debating audio integration. Should they ship something quick or build a comprehensive sound with voice capabilities?
Style references may get an update.
Midjourney merchandise may be available soon.
Niji’s update
Niji now also supports video end frames. You can even loop the video, just like in Midjourney.
Runway Aleph
Runway just dropped something pretty wild with their new video model called Aleph. What's cool about this one is that it combines video generation and manipulation in the same model.
You can add objects, remove stuff, transform things, change camera angles, modify lighting and styles - basically edit while you generate. It's like having a Swiss Army knife for AI video creation.
The exciting part? You might be able to use it to modify videos generated from other platforms like Midjourney.
You can watch the video here
And the official announcement page: https://runwayml.com/research/introducing-runway-aleph
Here’s what I created using Runway Aleph
Runway Aleph prompt:
The lady is pointing at a flying meteor at night, clear sky with starsOriginal video before the editing with Aleph
Ideogram Character
Ideogram just launched its Character feature, and it's pretty impressive for character consistency. You feed it a single reference image, and it'll generate your character in different scenes while keeping all those tiny facial details intact. They've got Magic Fill for dropping your character into any scene or meme, plus Remix for style transfers while keeping your character's unique features.
Quick heads up: if you watch their YouTube demos, you might think Ideogram does video generation. Nope! It's just creating consistent character images that you can then animate using other video generators. The facial detail capture is solid, though, especially those small nuances that other tools often miss.
Here’s the result of my experiment:
Kling AI Elements
Kling AI rolled out its Elements feature for image-to-video generation, letting you use 1 to 4 reference images to guide your video creation. But here's the frustrating bit - it only works with Kling 1.6, not their latest 2.1 model. I tested it out, and while the quality is okay, it's not quite matching what you get from, say, using Midjourney for reference images, then creating scenes and animating with Runway Gen-4. Classic Kling move, though. They always make you think new features work with the latest model, then boom, you log in and realize it's limited to older versions.
Hedra Live Avatars
Hedra's calling their new Live Avatars "the most advanced streaming avatar in the world." Think of it like Google's NotebookLM podcast hosts, but with actual faces that react and express while talking to you.
The avatars respond in real-time using ElevenLabs voices, creating a more engaging conversation experience. The setup isn't overly complicated. You need Hedra and OpenAI paid API access. But watch out for costs: it's about 16 credits per minute, which adds up fast if you're chatty.
The resolution isn't super high yet, but the real-time facial expressions and conversational flow make it pretty engaging. It's definitely pushing the boundaries of interactive AI characters.
Update from Geeky Curiosity
I’ve updated the Midjourney Style Resources #12 to V7 with updated aesthetics.
Prompt Play
Today I'm testing which AI video generator handles complex emotions best in just 5 seconds. My prompt is deliberately long and detailed - maybe too vague for the models to interpret properly. But that's half the fun of experimenting.
The video prompt is the for all models, except Midjourney has additional parameters (motion high and raw) to guide the model.
A woman stands beneath a tree, reading a letter intently as the wind stirs her hair and rustles the paper in her hands. Tears stream down her face as she absorbs the words, her emotions raw and visible. Suddenly, her expression shifts—she breaks into a joyful smile, radiating relief and a sense of closure. Behind her, the tree’s leaves dance in the breeze, mirroring the movement of her hair and the letter, capturing the atmosphere of revelation and hope. The camera slowly dollies in from a wide shot to a close-up, emphasizing her emotional journey and the subtle motion of her surroundings --motion high --rawMidjourney V7
Midjourney's video captured the facial expression quite well, though it skipped the tears entirely. The camera zoom worked exactly as intended. Of course, Midjourney had a slight edge here, as it creates four variations at once.
Kling AI 2.1 Master
Kling AI really let me down on this one. It completely abandoned the original scene midway and switched to something totally different. I suspect it couldn't recognize the pine tree in the background and "panicked," creating an entirely new tree scene instead. Pretty frustrating.
Runway Gen-4
Runway Gen-4 delivered dramatic emotional intensity in the character's face, but it missed those subtle details that show the inner emotional journey. Plus, it zoomed out when I specifically wanted it to zoom into her face.
My takeaway
When I compare all three, Midjourney comes out on top. It captures the details beautifully, offers multiple variations to choose from, and gives you the best value for your money.
Inspiration
Adding faces to everyday objects instantly makes people curious. There's something weird and wonderful about seeing a coffee mug with eyes or a book that’s sleepy. Check out what happened when I gave these ordinary items some personality.
a toothbrush with a heroic human face, standing tall in a bathroom cup, dramatic under-lighting like a stage performance, whimsical and bold photographic style --ar 16:9 --profile uw4q9o7 --v 7 
a pair of expressive boots with startled faces, splashing in a puddle, reflective wet ground, dark stormy twilight sky, surreal and whimsical photographic tone --ar 16:9 --profile uw4q9o7 --v 7 
a steaming ceramic coffee mug with a sleepy human face, eyes half-closed, sitting on a misty windowsill bathed in golden sunrise light, photographic depth of field, storybook dreamlike atmosphere --ar 16:9 --profile uw4q9o7 --v 7 
a slightly melting wall clock with a bored human face, distorted like surreal art, lit by warm afternoon sunlight, soft shadows and whimsical tension, photographic realism --ar 16:9 --profile uw4q9o7 --v 7
a stack of old books with sleepy yawning faces, leaning on each other under the soft glow of a desk lamp, cozy and dreamlike photographic style --ar 16:9 --profile uw4q9o7 --v 7 
a whimsical porcelain teapot mid-pour with a wide-eyed face full of wonder, surrounded by soft glowing kitchen ambiance at dusk, magical lighting and rich texture --ar 16:9 --profile uw4q9o7 --v 7 
Recent articles
Related articles
Free Geeky Animals' Sref code: 3138060801
Prompts are created using MJ Simple Prompt Generator (+ my brief idea)
. --sref 3138060801 --profile uw4q9o7 --sv 6 --v 7 
. --sref 3138060801 --profile uw4q9o7 --sv 6 --v 7 --ar 16:9
. --sref 3138060801 --profile uw4q9o7 --sv 6 --v 7 --ar 16:9
. --sref 3138060801 --profile uw4q9o7 --sv 6 --v 7 --ar 16:9
Cover prompt: a cat making friend with a dog --ar 16:9 --sref 3138060801 --profile uw4q9o7 --v 7
I hope you like this article!
Thank you for reading and happy creating!