Community Ranking Opens for Native 2K High-Resolution Image Tuning

Midjourney has launched a new community ranking initiative focused on high-resolution image generation. By presenting users with pairs of raw 2K images, the developers aim to collect human preference data that will shape the aesthetic tuning of future model releases. This move signals a shift toward native high-definition output without relying solely on post-process upscaling.

What's new

The current ranking phase, titled 'Rank V8.1', focuses specifically on images generated at a native 2K resolution. Unlike previous iterations that often prioritized composition at lower resolutions before upscaling, this test evaluates how the model handles fine details, micro-textures, and complex lighting at a larger scale from the start.

Participants are presented with two images side-by-side and asked to select the one they find more visually appealing. These images are 'raw,' meaning they lack the heavy stylization or corrective filters often applied to finished generations. This feedback loop helps the AI understand which high-frequency details—such as skin pores, fabric weaves, or atmospheric haze—look natural versus which contain digital artifacts.

How it fits your workflow

For filmmakers and concept artists, this update addresses a common bottleneck in AI image generation: the 'upscale loss.' Often, a low-res preview looks perfect, but the detail changes or softens when blown up to 4K or 8K for a pitch deck or production asset. By training Midjourney to understand native 2K beauty, the developers are laying the groundwork for more reliable high-fidelity assets that require less manual cleanup in Photoshop.

Storyboard artists can expect more consistent results when generating wide-angle shots where small background elements previously became muddy. This focus on resolution puts Midjourney in closer competition with tools like Flux or Stable Diffusion XL, which have also prioritized higher base resolutions to meet the needs of professional print and digital media workflows. If you currently use Topaz Photo AI or Magnific to fix blurry generations, these upcoming native HD improvements may eventually reduce your reliance on third-party enhancers.

What it costs / how to try it

Ranking is available to any user with an active Midjourney subscription through the official website. Participating in ranking sessions often earns users free GPU hours, which are credited to their accounts as a reward for contributing to the model's development. You can access the ranking tool directly on the Midjourney website under the 'Tasks' or 'Rank Images' section.