Creativity AI #77: Midjourney’s busy summer, Runway lands in your AI assistant, and ElevenLabs…
Office hours roadmap, MCP-powered video generation, AI music with embedded sound effects, plus Kling AI’s Cannes debut.
Creativity AI #77: Midjourney’s busy summer, Runway lands in your AI assistant, and ElevenLabs ships Music v2
Office hours roadmap, MCP-powered video generation, AI music with embedded sound effects, plus Kling AI’s Cannes debut.
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 (May 27th, 2026): editing model, V8.2 delays, and a busy summer ahead
The latest office hours outlined a packed summer roadmap, with the editing model as the main research focus for now.
That editing model is the headline. Planned capabilities include inpainting, outpainting, workflows tied to Omni Reference (OREF), and next-generation transformations from input images. The team is debating whether to ship early with limited features or wait for something more complete, and the bottlenecks are still being worked out in testing.
Behind the scenes, a big infrastructure migration is slowing everything down. Midjourney is shutting down one cluster and bringing a new main cluster online, which means massive data and model transfers. The payoff is supposed to be faster research, support for new model types, and room for larger features going forward.
V8.2 was meant to land soon with aesthetic improvements. Then the “minor” issues turned out to be deeper system-level problems, so the team is fixing those first. They floated small “micro releases,” but worry these would disrupt workflows and create inconsistencies that people would start relying on. Bigger, longer-lasting releases are preferred.
On 8.1, it should soon become the default model on the main site, though OREF may continue to fall back to V7 for a while. Again, the migration is the holdup. Internally, the feeling is that users like 8.1.
A few more things are in the pipeline. Larger batch generation modes are being tested internally, and the core functionality reportedly works well already. The blocker is the interface, so the release waits until the UI is good enough. They’re also exploring fast style previews, fast settings previews, and real-time experimentation tools. Better systems for earning fast hours are planned too, possibly by letting users help with testing or research.
Website redesigns are in progress but slow, mostly because any released feature gets baked into people’s workflows fast, and changing it afterward gets messy. The team mentioned a possible experimental “alpha” site with faster-evolving interfaces, though they’re wary of changing tools too quickly. Expect larger but less frequent interface updates.
There’s also a teaser about hardware. Midjourney plans to reveal experimental hardware projects soon, not aimed at the general public, and frame them as ongoing work they want to be more transparent about.
For June, expect 8.2, the editing model, OREF improvements, and image preview features. Real-time generation, larger batch workflows, and more ambitious UI overhauls sit further out, with some bigger UI changes possibly pushed to July. Around 80–90% of the company stays focused on core image generation.
My take: Midjourney is taking its sweet time to make sure 8.1 is solid before moving to 8.2. While other tools chase the next feature, like Model Context Protocol (MCP) integration or better character reference, Midjourney stays happily settled in its own world, crafting the next b
Runway brings its models into your AI assistant with MCP
Runway just connected its image and video generation to the AI agents you already use, such as Claude, ChatGPT, Cursor, and Replit, via the Model Context Protocol (MCP).
The idea is simple. Instead of hopping over to Runway’s web app, you ask your AI agent in plain conversation to make images or videos, and Runway handles the generation in the background. You can see the full rundown at https://runwayml.com/mcp.
Link to video
[embed]
Setup is light. In a supported agent like Claude Desktop, you add a custom connector pointing to Runway’s MCP endpoint, sign in with your Runway account, and start prompting for visuals inside your chats. No API key juggling.
You get access to a solid roster of current models, including Gen-4.5, Seedance 2.0, GPT Image 2, Kling 3.0, Veo 3.1, Nano Banana Pro, Gen-4 Image, and Gen-4 Turbo. Runway highlights a few use cases, e.g., general image and video generation, marketing content from a product URL (point the agent at a product page and ask for a launch video), multi-shot dialogue-driven storytelling, and product imagery that can tie into building a shopping site.
The bigger shift is workflow consolidation. If Claude or ChatGPT is already your creative cockpit, you stop context-switching to a separate video tool and copying assets back and forth. You ideate, script, and generate in the same chat. It also opens up agentic workflows, in which the agent reads a product URL, drafts concepts, calls Runway for the hero video, and drops the hero video into a page build, all in a single flow.
My take: I like that Runway “suddenly” got more active with rollouts lately. It wasn’t like that before. Connecting to MCP is a big deal, since you can make videos just by chatting with tools like Claude and ChatGPT. On the downside, I still find it a bit odd that the creative parts, like exploring video concepts, picking the right image for generation, and browsing past assets, now get reduced to a few lines of code. Something feels missing, and I can’t quite name it. Maybe it’s the creative exploration. Or maybe I just need more time to adjust. What do you think?
ElevenLabs launches Music v2 with embedded sound effects
ElevenLabs rolled out Music v2, its new model for generating full songs with vocals, instrumentation, and structure from a text prompt. You can try it at https://elevenlabs.io/music.
The headline features cover mid-track genre transitions, fast rap and dense lyrical delivery, improved multilingual generation, and long-form composition support. The one that stands out is non-musical sound effects embedded directly inside a track, something Suno can’t do. Improved inpainting is here too, so you can select any section, e.g., a bridge or a single verse, and regenerate just that part while keeping the rest intact.
Link to video:
[embed]
Music v2 powers three platforms with one model: ElevenMusic for listening, remixing, and creating; ElevenCreative for downloadable licensed music for ads and video; and ElevenAPI (coming soon) for embedding generation into your own product. ElevenLabs says pricing on ElevenCreative Music drops by up to 50%, with the same cut coming to ElevenAPI Music. The model is claimed to be trained on licensed data and cleared for commercial use.
Worth keeping in mind that this sits alongside the rest of the ElevenLabs V2 lineup, e.g., the Multilingual V2 text-to-speech model and the V2 speech-to-speech voice changer. Stack those together and a creator can narrate a video, dub it into other languages with a cloned voice, score it with original music, and add character dialogue, all from text or a rough voice take.
My take: The embedded non-musical sound effects are the part I like most, since you can’t do that inside a track with Suno. Pair that with custom voice generation, and you’ve basically got an unlimited pool of singers queuing up to perform your song.
Kling AI debuts an AI film at Cannes
Link to video
[embed]
Kling AI made its official debut at the 79th Cannes International Film Festival with a panel at the Marché du Film on May 18th, 2026, titled “From Creative Possibility to Production Reality: Kling AI in Cinematic Workflows.” It featured three films built with the platform.
The standout is Born of the Tide, described as the first AI-generated cinematic epic about China’s Tanka community, the “sea nomads” known for their floating lifestyle. It was directed by Li Wei, who co-directed the 2020 box office hit Jiang Ziya: Legend of Deification. At Cannes, he said Kling AI cut roughly a third off both production time and budget compared to his earlier animated work. The official release points to 4K scenes like dragon boat races, fish market explosions, and mountain confrontations, and quotes him on how the platform held his intended color palette while rendering waves, heavy rain, and firelight reflecting off wet ship decks.
My take: Ever wonder what “good” looks like when experts make an AI film? This is one answer. Watching what a seasoned director does with these tools is a different thing entirely from the quick clips most of us generate, and the production time and budget savings are the part that should make studios sit up and take notice.
Inspiration
Xanti Schawinsky was a Bauhaus-trained Swiss artist who studied under Klee and Kandinsky, then fled to the US in 1936. His work resists easy categorization: painting, photography, stage design, performance art.
The visual language ranges from bold geometric man-machine imagery to surrealist compositions with thin lines and strong color. He literally drove a sports car over paint-covered canvases. For AI art creation, Schawinsky is useful when you want Bauhaus structure colliding with surrealist unpredictability.
Moodboard code: — p nu7n1nl
(To use the code, add it to the end of your prompt)
. --ar 16:9 --profile nu7n1nl --v 7
. --ar 16:9 --profile nu7n1nl --v 8.1
. --ar 16:9 --profile nu7n1nl --v 7
illustration of technology ascending like a symbol of human ambition, beams of light, disciplined geometry, mechanical forms, and visionary balance creating a dramatic narrative of progress, in the style of Xanti Schawinsky --ar 16:9 --profile nu7n1nl --v 7
. --ar 16:9 --profile nu7n1nl --v 7
Woman cradling a cup of coffee, deep in thought, seated by a quiet café window, soft morning light touching her face, distant gaze, intimate cinematic portrait, subtle emotional realism, contemplative atmosphere --ar 2:3 --profile nu7n1nl --v 8.1
Woman cradling a cup of coffee, deep in thought, seated by a quiet café window, soft morning light touching her face, distant gaze, intimate cinematic portrait, subtle emotional realism, contemplative atmosphere --ar 2:3 --profile nu7n1nl --v 7
. --ar 16:9 --profile nu7n1nl --v 7
Recent articles
Cover prompt: . — ar 16:9 — profile nu7n1nl — v 7
I hope you like this article!
Thank you for reading and happy creating!
메타데이터
- post_id
- eeb4e5fd608d
- slug
- creativity-ai-77-midjourneys-busy-summer-runway-lands-in-your-ai-assistant-and-elevenlabs-eeb4e5fd608d
- url
- https://medium.com/creativity-ai/creativity-ai-77-midjourneys-busy-summer-runway-lands-in-your-ai-assistant-and-elevenlabs-eeb4e5fd608d
- canonical_url
- https://medium.com/creativity-ai/creativity-ai-77-midjourneys-busy-summer-runway-lands-in-your-ai-assistant-and-elevenlabs-eeb4e5fd608d
- author_url
- https://medium.com/@geekyanimals
- status
- ok
- fetched_at
- 2026-06-12 07:40:50
