Midjourney V8.1: HD Now Default — 3x Faster & Cheaper

Midjourney just shipped V8.1 — and quietly reset what HD generation costs across the AI image generation platform. High-resolution mode (HD), which previously carried a 4× credit premium over standard jobs, is now the default for all V8.1 users. That pricing collapse, combined with a 3× speed jump, signals Midjourney is competing aggressively on both quality and cost — without making a public announcement about it.

V8.1 also restores image prompts (the ability to upload a reference photo and use it to steer a generation toward a specific visual direction — absent from V8's initial launch). Add an automatic Prompt Shortener and faster standard-resolution outputs, and V8.1 reads less like a minor point release and more like a full rethink of the V8 product.

Midjourney V8 and V8.1: AI Image Generation Speed Numbers

V8's generational leap over V7 was already dramatic: the model generates images roughly 5× faster than its predecessor at full quality. V8.1 pushes that momentum further in two distinct directions:

• HD mode (new default): 3× faster and 3× cheaper than V8 standard resolution, rendering at native 2K resolution with additional detail passes that sharpen texture, fine structure, and edge fidelity.
• Standard mode: 50% faster and 25% cheaper than V8's standard quality tier — a meaningful cost reduction for high-volume workflows.
• Speed parity milestone: V8.1 standard at full quality now matches V7's draft mode speed — the quick-preview setting V7 users relied on for rapid iteration before committing to a full-quality render.

There's also an alternative SREF variant (SREF stands for Style Reference — a feature that locks in a specific visual aesthetic across multiple generations using a reference image or numeric code) running at 4× faster and 4× cheaper than standard. The tradeoff: reduced style accuracy versus the original SREF implementation. Best for rapid style exploration; less reliable for precision brand-consistent work.

HD: From 4× Upcharge to Default Experience

The structural change in V8.1 is the HD pricing reset. Previously, HD mode (which applies additional detail-rendering passes on top of base generation, producing outputs at native 2K resolution) cost 4× more credits per job. That made it a deliberate — and expensive — opt-in for subscribers who needed higher-fidelity output for print, portfolio, or client deliverables.

V8.1 eliminates that friction entirely. Midjourney's framing from the update post was unambiguous: "It's so cheap we're making it default for V8.1." The pricing drop arrived as a default setting change — not a press release. Every subscriber is now generating at a quality tier that previously cost 4× the baseline rate.

For comparison: Adobe Firefly's high-resolution enhancement pass and OpenAI's DALL-E 3 both treat quality upscaling as an additional step or higher-tier feature. Midjourney is baking that ceiling into baseline access — a move that increases pressure on competitors to match the value proposition without raising subscription prices.

What Still Costs More: The --q4 Exception

One caveat: the --q 4 flag (a maximum-quality rendering parameter that runs 4× slower and uses 4× more credits than standard — useful for final-delivery hero images) still carries the original cost multiplier. V8.1 significantly improved HD economics, but Q4 remains a premium option. Note also that --hd and --q 4 cannot be combined simultaneously in Relax mode (the unlimited generation tier available to Standard, Pro, and Mega subscribers, which queues jobs instead of consuming credits per image).

Image Prompts Are Back — With a Prompt Shortener Added

V8's initial release shipped without image prompt support — a notable gap for production workflows. Image prompts (using a reference image alongside text to maintain visual consistency across a project, match existing brand imagery, or iterate on a specific composition) are central to how many professional Midjourney users operate.

V8.1 restores full image prompt functionality including image weights (--iw), a parameter ranging from 0 to 3 that controls how strongly the reference image influences the output versus the text description. A value of 0.5 gives the text prompt more creative latitude; at 2.5, the reference image dominates the aesthetic direction and the text becomes secondary framing.

Also new in V8.1: the Prompt Shortener, which auto-activates when a text prompt exceeds the platform's length limit. Rather than throwing an error or silently truncating the input, it compresses the prompt while preserving core intent — particularly useful for complex multi-element scene descriptions with detailed style specifications, nested character layers, or lengthy negative prompt chains.

Community Rating Parties Are Tuning the Model in Real Time

One of the most structurally unusual aspects of V8 development: Midjourney ran large-scale image rating parties — structured community sessions where subscribers compared pairs of V8.1-generated images at 2K resolution and voted on quality, detail fidelity, and aesthetic accuracy. These votes fed directly into model tuning, not just user research.

The V8 and V8.1 cycles completed multiple rounds (Round 2, Round 3, and a Final round), each running on 1–3 week iteration sprints. The result is a development cadence significantly faster and more community-driven than how OpenAI or Adobe tune their image models — which rely primarily on internal evaluation teams and curated benchmark datasets invisible to end users.

This creates compounding personalization: subscribers who participate in rating parties are effectively co-training the model on their own aesthetic preferences. The feature now extends to niji 7 (Midjourney's anime and illustrative-style model, optimized for cel-shading, manga aesthetics, and animated character work), which received full Personalization and Moodboard support in V8.1.

Participate in the next rating party at midjourney.com/rank-v8-1. Hotkeys: 1 = left image wins, 2 = right image wins, 3 = skip. Each vote refines the personalization system for the entire subscriber base, not just your own profile.

How to Access V8.1 Right Now

V8.1 is rolling out through a staged waitlist on alpha.midjourney.com (Midjourney's early-access testing environment, separate from the main Discord bot and the standard web interface). Standard, Pro, and Mega plan subscribers can now use Relax mode on V8 — unlimited generation credits with queuing — which wasn't available during V8's initial alpha due to server capacity constraints.

Key V8.1 commands:

--hd              HD mode — 3x faster, now default in V8.1
--iw [0-3]        Image weight — reference image influence strength
--chaos [0-100]   Variation and randomness in outputs
--weird [0-3000]  Push toward unusual aesthetic territory
--raw             Reduce post-processing for photographic look
--exp             Experimental mode, expanded creative range
--ar [ratio]      Aspect ratio, e.g. --ar 16:9 or --ar 3:2

V7 personalization profiles carry forward to V8 without rebuilding. Manage and create aesthetic profiles at midjourney.com/personalize — now using a scrollable click interface that replaces the older forced 1v1 comparison flow. This significantly reduces the time investment in building a personalization profile from scratch.

If you're on an active subscription and waiting for full V8.1 access, check the AI automation workflow guide for techniques you can apply on V8 standard in the interim. The HD default change alone makes V8.1 worth prioritizing — you'll get output quality that previously required 4× the credits, now included in your existing plan without any configuration change.

Related Content — Get Started | Guides | More News