Creativity AI #37: Weekly AI art update
News about AI art, LLM, Creativity, and 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 is taking a break from office hours this week while the team gets together in San Francisco. Perfect timing to catch up on other AI news and do other things (like tidy up the balcony garden or take a vacation?).
MIT just dropped a reality check that might sting a bit. The new NANDA initiative report shows 95% of company AI pilots are flopping. Not because the tech's bad, but because companies are doing it wrong.
You know those 19-year-old startup founders who went from zero to $20 million in a year? They picked one problem, solved it well, and partnered smart. Meanwhile, big companies are throwing half their budgets at sales and marketing tools when the real money's in automating boring back-office stuff.
If you're running AI projects at work, yeah, those 5% survival odds sound brutal. But knowing what works (focus on one pain point!) beats pretending everything's fine.
DeepSeek just casually dropped V3.1 on Hugging Face without any big announcement. This 685-billion parameter beast is scoring close to OpenAI and Anthropic on benchmarks, and it's completely open source. What's cool:
Massive context window upgrade (128K tokens)
Multiple tensor formats (BF16, F8_E4M3, F32)
Already hitting 71.6% on Aider coding tests
Free to download right now
Can't test it myself yet since Ollama doesn't have it, but researchers worldwide are already going wild with it.
Alibaba's Qwen team released something pretty neat yesterday. Qwen-Image-Edit basically gives you Photoshop-like editing capabilities through text prompts. Built on their 20-billion parameter model, it handles everything from tiny tweaks to major image overhauls. Just upload a photo, type what you want changed, and boom. I saw an example where they edited a conference photo, and the results were surprisingly clean.
Here’s my example:
Original image
After editing with Qwen-Image-Edit with this prompt:
change the outfit to red
ElevenLabs just dropped a video-to-music feature that's genuinely useful. Upload your video, get instant background music. No more hours lost searching for the perfect track. You can even layer sound effects and voiceovers to make your videos feel more cinematic. Easy-peasy.
The catch? Free users get watermarks. But oddly, I'm seeing watermarks on my paid account too. Since it's still in Alpha, I suspect they're ironing out the bugs.
What excites me most is the potential API. Imagine batch-processing music for dozens of AI videos automatically. That's when this tool becomes seriously powerful for content creators.
Video from Midjourney. Background music from ElevenLabs. But ElevanLabs claims it all :-)
Runway's making Act-Two more interesting by adding Voices. Now you can match your character's voice to their appearance without recording new audio. Pretty straightforward workflow, too. They've got a guide up at their academy site showing exactly how to swap voices around.
Update from Geeky Curiosity
I'm going on a short summer vacation. The publication will continue as usual, as I have planned and scheduled everything. I hope you had a wonderful time too!
Link to the music in case the player doesn’t work. Enjoy!
Refreshed Midjourney Style Resources
The Midjourney Style Resources #15 has been updated to V7 aesthetics.
Inspiration
Midjourney can feel addictive when you're churning out image after image. I got so caught up in the generation flow that I nearly forgot about exploring specific techniques I'd been wanting to try, like high contrast photography compositions. Time to slow down and experiment properly.
A lone tree illuminated under a dramatic spotlight, surrounded by dark storm clouds, photographic high contrast composition --ar 16:9 --profile tt3bo9d --v 7
White dove mid-flight captured in high contrast photography, deep black background, strong rim lighting on feathers --ar 16:9 --profile tt3bo9d --v 7 
Cyclist passing under a lone streetlamp on a foggy night, high contrast lighting, cinematic atmosphere --ar 16:9 --profile tt3bo9d --v 7 
Golden statue in a dimly lit stone chamber, strong directional lighting creating stark contrast, photographic realism --ar 16:9 --profile tt3bo9d --v 7 
Person in a bright red coat standing in a quiet, snowy forest at dusk, high contrast photographic style --ar 16:9 --profile tt3bo9d --v 7 
Handsome muscular man holding a vibrant red rose on deep black velvet, intense color contrast, studio portrait lighting --ar 16:9 --profile tt3bo9d --v 7 
Recent articles
Related articles
Free Geeky Animals' Sref code: 1682011180
Prompts are created using MJ Simple Prompt Generator (+ my brief idea)
In a high-stakes escape scene, a lone warrior clashes with grotesque extraterrestrials while civilians evacuate, intense lighting contrast, cinematic angles --ar 16:9 --sref 1682011180 --profile tt3bo9d --sv 4 --v 7
Epic scale combat: the heroine leaps from a wrecked vehicle, engaging giant aliens as her team retreats, adrenaline-charged moment, wide lens view --ar 16:9 --sref 1682011180 --profile tt3bo9d --sv 4 --v 7 
heroine on top of wreckage, dual-wielding plasma blades, fighting aliens as survivors escape into a portal, epic scale, glowing lights --ar 16:9 --sref 1682011180 --profile tt3bo9d --sv 4 --v 7 
fierce cinematic standoff between a lone female warrior and towering extraterrestrials amid fire and debris, intense expression, glowing weapons --ar 16:9 --sref 1682011180 --profile tt3bo9d --sv 4 --v 7 
Cover prompt: Amid smoke and debris, the heroine launches into battle with alien giants, buying time for a convoy to escape, explosive energy, slow motion feel --ar 16:9 --sref 1682011180 --profile tt3bo9d --sv 4 --v 7
I hope you like this article!
Thank you for reading and happy creating!