EP 037 · 66 MIN RANK 200 IMAGES, HANDS FIX THEMSELVES

Midjourney V7 Showed Up in Crocs and Cooked

April 13, 202566 min Rory Flynn & Drew Brucker

Watch on YouTube Spotify Apple Podcasts Audio file

The short version

Episode 37 is Fast Hours' second look at Midjourney V7 after a week of real use. Rory and Drew argue the backlash was premature, since ranking the first 200 personalization images visibly fixed hands, colors and composition, remix and tile are back, and one-shot character sheets already preview the still-unreleased Omni-Reference feature. They also flag a new --no-parameter inversion trick and a side-by-side comparison showing V7's images are noticeably brighter with more shadow and highlight detail than prior versions.

Key takeaways

01
Rank your first 200 V7 personalization images before judging the model: Rory saw immediate gains in hands, colors and composition.
02
Stuck for ideas? Prompt what you don't want instead: a long --no list (no humans, no flowers, no text) produced some of the best surprises of the episode.
03
For controlled results, build the composition with a plain prompt first, then hit Remix Strong and drop in a style reference or mood board on top.
04
Pair a personalization profile with mood boards on V7: Rory found moodboards alone looked 'jacked up,' but the profile code seemed to bridge the gap.
05
V7's brighter, more exposed look isn't a settings tweak — it's more real detail rendered into shadows and highlights that earlier versions hid in silhouette.
06
One-shot character sheets in base V7 already hold consistent hair texture, jewelry and even a mangled necklace detail across four panels, ahead of Omni-Reference shipping.
07
Don't judge a new Midjourney model in the first 48 hours: Drew argues real evaluation needs weeks of testing stylize, aspect ratio and personalization together.

Receipts · said on the record

“The first 200 images I rated, my images immediately got better. Hands got better, appendages got better. … And the more I keep ranking, the better my stuff keeps looking.”

Rory Flynn Why ranking 200 personalization images changed his results

“Mid Journey is basically a victim of their own success there. … Everyone has that sort of short-term memory where it's like, well, we might need to be a little bit patient, or figure this thing out again, because it's a whole new infrastructure.”

Rory Flynn Why V7's rough launch triggered an overreaction

“There's a few things. I'll go to the actual character sheets though first, which I was super impressed with. There's no upscaling on this outside of Mid Journey's Upscaler.”

Rory Flynn Character-sheet consistency achieved with no upscaling

“The dynamic range is how you would like sort of describe that. The brightest brights, the darkest darks, the details within them. … That's the difference between a camera that costs $2,000 and a camera that costs five to $10,000.”

Drew Brucker Explaining V7's improved dynamic range with a camera analogy

“It's like inversion, you know, like invert the problem here. If you're feeling lack of creative inspiration or you just wanna really put some tension and constraint on the tool, why not?”

Drew Brucker Reframing the --no parameter as a creativity tool

Chapters

Questions this episode answers

Was the backlash to Midjourney V7 justified?

Drew and Rory argue no: V7 was an alpha-stage release from a company that, unlike its competitors, isn't VC-funded, and most critics judged it within 48 hours without testing stylize, personalization and aspect ratio changes together. They suggest judging any new model after a month or two of real use.

How much does ranking images for personalization actually matter in V7?

A lot, according to Rory: rating his first 200 images produced an immediate, visible jump in hand and appendage quality, color and composition, and he recommends anyone unhappy with V7 rank at least 200 images before judging it.

What is the --no parameter trick discussed on the episode?

Rory found an Explore-page prompt using an extensive --no list (no humans, no flowers, no text, and more) that produced strikingly good, unexpected results. Drew reframes this as inverting the prompt: when you're short on positive creative direction, specifying what you don't want can unlock results a positive prompt wouldn't.

Is Midjourney's character consistency (Omni-Reference) released yet?

Not at the time of this episode — Midjourney had only given a rough one-to-four-week timeline. But Rory's one-shot, four-panel character sheets in base V7 already held consistent hair, jewelry and clothing texture, which both hosts took as a strong early signal.

Why do Midjourney V7 images look brighter than earlier versions?

Rory theorizes it reflects more rendering detail, not just an exposure change: dark, silhouetted regions need less generated detail, so earlier versions leaned dark to hide their limitations, while V7 can now render a fuller dynamic range, holding detail in both shadows and highlights.

Show notes

Fast Hours Podcast Ep. 37—Midjourney V7 Showed Up in Crocs and Cooked (with Drew Brucker & Rory Flynn)

Drew and Rory are back for their full Midjourney V7 breakdown. What starts as a casual catch-up spirals into 60 minutes of real strategy, subtle discoveries, prompt experimentation, and a surprisingly profound rant on image brightness. Also, yes, there’s an alien smoking a joint.

From remix tricks to personalization secrets, from prompt inversion to tonal contrast, this one’s got gold buried in the chaos.

Coverage: Midjourney V7 personalization techniques Character consistency updates Remix & tile shipped Draft mode impressions How to use “no” prompts like a wizard Using the Explore page like a Pinterest board for freaks

Ps—If you’re still rating strawberries on V6, we’re not judging… but we are worried about you.

⏱️🔥 Midjourney Fast Hour [00:00] Breathing room with Midjourney V7 – finally [01:48] Revisiting old prompts: cringe or still fire? [03:04] Stylize settings: what’s working now [06:09] Personalization isn’t optional anymore [07:08] Remix & tile are BACK [08:34] Remix strong vs subtle: what’s better? [09:04] Why style refs work better after remix [11:38] Character consistency timeline + visual proof [13:40] Why Midjourney’s slow drip strategy is good [15:12] Model overreactions and spoiled user syndrome [18:06] Object reference (Omni Ref) is being slept on [19:31] Midjourney for freelancers = sweatpants marketing [20:27] Scrappy creativity and custom 3D icons [22:31] Visual consistency across details (teeth, hair, necklaces) [23:08] Poor man’s Rory Flynn appears [24:09] Why BoltCam + Luma is ridiculously fun [26:34] Draft mode is quietly improving [28:38] Mood boards only work with personalization [30:03] “The work is mysterious” = prompt gold [31:13] New “rate this image” feature – does it help? [32:20] Creepy portraits and single-eye stunners [34:25] Album art experiments and inspiration [35:16] The Explore Page finally doesn’t suck [36:49] Remixing prompts from top creators [38:19] Prompt inversion: the power of “--no” [40:13] Composition unlocks from excluding objects [43:42] Image blending, style refs, and banger prompts [44:51] Tattoo realism and forgotten saves [46:57] Personalization tips and speculative theory [49:20] Just prompting “Midjourney…” for kicks [52:13] Midjourney is finally brighter (literally) [54:19] Brightness vs realism: what's changed in V7 [56:55] Dynamic range: the secret sauce [59:47] Midjourney’s stylize slider [01:01:30] Nuzzle, tonal balance, and V7 token talk [01:03:43] Chill TF out—it’s an alpha for a reason

Every word Read the full transcript