Heavy Midjourney tester says v8.2 is better at tiny-subject wide shots

ciguleva · x · 2026-07-26

A heavy Midjourney user says v8.2 is noticeably better at wide shots and realism

A creator who generates hundreds or even thousands of images says their evaluation of new Midjourney models comes from mixing moodboards, personalization codes, SREF codes, image prompts, and exhaustive parameter testing.

Their take on v8.2:

The post is less a benchmark than a practitioner’s field report: the model’s vibe only becomes obvious after pushing many prompt and blend combinations through it.

Original post →

More from Multimodal

Multimodal channel →