What Changed in V7
Midjourney V7 introduced three major improvements that genuinely matter: Character References (consistent characters across images), Style References (extract and apply any artistic style), and significantly improved prompt adherence for complex compositions. After 2,000 generations testing V7 against V6, here's the real breakdown.
Character References: Finally Solving the Consistency Problem
The biggest limitation of AI image generation has always been character consistency — generating the same person across multiple images. V7's Character Reference system solves this remarkably well. Using --cref with a reference image, characters maintain their distinctive features across different scenes, lighting, and compositions. Not perfect, but a massive improvement.
Prompt Adherence: Noticeably Better
V7 follows complex prompt instructions significantly better than V6. "A woman in a red dress standing in front of a blue door holding a yellow umbrella" — V7 gets all four elements right roughly 80% of the time vs 55% for V6. For commercial use cases requiring specific compositions, this matters enormously.
Still Frustrating: Text Rendering
Midjourney still cannot reliably render legible text in images. This is a fundamental limitation compared to FLUX.1 which handles text beautifully. For any image requiring readable text — signs, labels, titles — use FLUX instead.
The Price Question
Midjourney's Basic plan ($10/month) includes 200 fast image generations. For casual exploration, this is enough. For professional use, the Standard plan ($30/month) with unlimited relaxed generations is the right choice. The Pro plan ($60/month) makes sense for agencies with high volume needs.