Midjourney V8 is finally here, and I put it through hours of real testing to see what’s actually improved (and what’s still a weakness). In this video I walk you through early examples, where V8 shines for character design and cohesive “art directed” images, and where it still struggles—especially when you need accurate scene logic or truly natural skin texture. I also test the new text rendering, and it’s genuinely better: V8 can hit quoted lines character-for-character in a way Midjourney has historically lagged behind models like Nano Banana. I compare V8 to V7 and, honestly, a lot of changes are subtle—until you hit high-detail, specific scenarios where V8’s cinematic qualities start to pull ahead. Beyond image quality, the workflow upgrades matter: grid view, cleaner settings placement, faster rendering (it’s dramatically quicker), conversational prompting, voice-based iteration, and a new HD mode for native 2K renders (with the caveat that Relax isn’t supported yet, so you’ll be burning fast hours). Finally, I compare V8 against Nano Banana 2, Krea-style outputs, and other approaches—and I share my favorite hybrid workflow: I use Midjourney to establish the “taste” and visual identity, then I bring that into Nano Banana 2 (inside Google Flow) to refine, expand scenes with grid prompts, and build out whole worlds. I also cover censorship limitations and the specific words that tend to trigger moderation.