V8 testers kept praising Midjourney's mood and visual voice while still flagging inaccurate limbs and changed SREF behavior. Run V7 and V8 side by side before moving a client style over.
--sref code produces different looks in V7 and V8, which matters for anyone carrying over a saved house style.The strongest early signal is tonal, not technical. In a detailed alpha thread, one tester says V8 feels “on a different planet” for mood even while calling out lingering limb mistakes. That lines up with Dustin Hollywood’s post, which shows motion-blurred portraiture, theatrical gowns, and makeup-heavy closeups that read more like fashion editorials than prompt-lab demos.
The examples themselves suggest V8 is currently strongest when the brief rewards atmosphere: blur, texture, dramatic lighting, and stylized beauty work. The images in
and
lean into color cast, fabric shape, and pose energy rather than perfect physical fidelity. A separate reposted example in another V8 image pushes the same direction with a luminous, smoke-wrapped portrait.
That explains the mixed reaction. V8 is getting credit for visual voice, but the anatomy caveat is not abstract; the tester explicitly says limb inaccuracies still stand out.
The most actionable finding so far is that style references are not behaving as a clean version upgrade. According to the SREF comparison post, the same style-reference code can produce materially different aesthetics in V7 versus V8. For creators with repeatable illustration recipes, that means a familiar --sref no longer guarantees the same comic, cover-art, or campaign look after switching models.
A second reaction reinforces that V8’s value may be less about replacing V7 immediately and more about discovering a new visual lane. In one creator’s response, they say they still love V7 and expect each model to open “new worlds,” but have not found V8’s yet. That is a useful read on the transition: V8 may reward exploration-first prompting instead of direct porting from mature V7 workflows.
For now, the safest interpretation is simple. V8 Alpha already looks capable of stronger mood-heavy renders, but anatomy reliability and shifted SREF behavior mean any established style pipeline should be tested side by side before it moves over.
Creator posts on March 24 pushed Midjourney V8 hardest on fashion, fine art and moody editorial imagery, while David Holz shared data showing longer prompts are climbing fast. It looks increasingly strong for art-direction-heavy work, but prompt complexity is becoming part of the workflow cost.
updateOpenAI has removed the Sora app as creators and Hacker News users debate whether novelty never turned into durable usage. Save projects now and plan to test ChatGPT-integrated or rival video tools next.
updateCapCut is expanding Dreamina Seedance 2.0 while Topview restored access within 24 hours, and creators are stress-testing it for vertical repurposing, long prompts and stylized start frames. Try it for fast video conversions, but budget cleanup passes for continuity and transitions.
promptCreators are turning Nano Banana 2 templates into reusable prompt systems for merch shots, sports ads, editorial portraits and modular scene builds. Keep the scaffold fixed and swap only brand, lens, action or environment variables to iterate fast.
workflowRiverside's Co-Creator reads transcripts automatically and turns chat-style requests into cuts, captions, thumbnails and social copy from one workspace. Use it when you need fast repurposing without timeline scrubbing, then polish the output by hand.
Midjourney v8 has been released. The same sref exhibits different styles in v7 and v8. Which one do you prefer? We will begin collecting the v8 version of the sref code. --sref 2489471477 1363686479 Left: V7, Right: V8
Been testing @midjourney v8 for a minute now on a wide range of stuff. Will have feedback in a write up on #NAKID Friday, started writing it today. 👌😎
