MIDJOURNEY V6.1 RELEASED + Runway Img-2-Video RELEASED.

Fam,

Well…I had plans this week.

Then Midjourney dropped V6.1 and Runway dropped Img-2-Video.

They’re here…and they’re impressive.

This is a long one → I’m sorry.

Here’s what we got for today:

1. MIDJOURNEY: V6.1 UPDATES → WHAT YOU NEED TO KNOW

2. RUNWAY GEN-3: IMG-2-VIDEO RELEASED

3. VIRAL VIDEO BREAKDOWN: MJ + RUNWAY MAKING MAGIC

4. AI CREATIVE: DESIGN SYSTEMS AT SCALE

5. BEYOND MIDJOURNEY:

6. MIDJOURNEY OFFICE HOURS

Let’s Boogie.

MIDJOURNEY VERSION 6.1 RELEASED:

We’ve been waiting…and Midjourney delivered again.

This isn’t a “massive update” per se…but it’s similar jump from V5-V5.1

Drew and I broke it down on Midjourney Fast Hours ↓

V6.1 UPDATES:

• Improved Image Coherence (arms, legs, hands, etc)

• Improved Image Quality (it’s good)

• (Reduced pixel artifacts, enhanced textures, skin, 8bit retro, etc)

• More precise, detailed, small image features (eyes, small faces)

• New Upscalers with much better image / texture quality -

• Roughly 25% faster for standard image jobs

• Improved text accuracy

• New Personalization model with improved

• NEW Q2: More texture and depth

We’ll touch a few here quickly…

LONGER FORM CAROUSEL → READ HERE

IMAGE COHERENCE:

How much better can it get?

The small details make a difference…and we have incremental improvements.

PROMPT: photo, everyday overweight man, crossing his arms, squatting, wearing a worn out old t-shirt, ripped jeans, scruffy beard, sunglasses --ar 4:5 --v {6, 6.1}

SMALL DETAILS:

• This is a big improvement.

• More complex poses were hard with V6

• Small details like feet spread apart, arms resting on knees

• The weight of his body is evenly distributed.

Let’s take a look at something more complex.

PROMPT: Photo Type: editorial cover shot, Subject Focus: 2 men riding horses, Subjects Wearing: worn down and torn clothing, Color Tone: contrasting light and dark elements, Background Color: vibrant jungle palette Texture: natural jungle textures, silky water, Mood: early morning ambition, Environment: in a Borneo jungle river, thick vegetation, Lighting: ethereal morning light --ar 2:1 --seed 2071069985 --personalize [code] --stylize 200 --v {6, 6.1}

IMPROVED QUALITY:

• Again...Solid, sutble improvements.

• Textures are significantly improved.

• Color, contrast, saturation improved.

• Lighting + physics have improved.

Where it really shines…is with some fine art baked in.

PROMPT: portrait of woman with intricate comic book facial tattoos, in the style of pop culture mashup, neon colors, photorealistic, closeup, black background --chaos 10 --ar 3:4 --sref https://s.mj.run/FrgAyHX3pyg --sw 350 --stylize 600 --v {6, 6.1}

OVERALL THOUGHTS:

• Similar jump from v5-v5.1

• Not a massive upgrade

• Improvements are subtle

• Distant faces are still meh

• More complex compositions

• Texture is significantly better

• Sharper/finer details in images

• Contrast has definitely improved

• Personalize" is firing on all cylinders

• Color and saturation is more defined

• Upscaler nothing to write home about

• Prompt coherence is good but different

If you want a more detailed breakdown…

Check out the FULL LINKEDIN POST ↓

RUNWAY GEN-3 IMG-2-VIDEO RELEASED:

RW continued the IMPRESSIVE release of Gen-3. (By adding img-2-video)

We waited for this feature…and it’s worth it.

If you missed it…FULL TEST VIDEO ↓

CRAZY IMPROVEMENT:

GEN-2 was revolutionary for img-2-vid…(Gen-3 blows it away)

Before I2V was released…consistency was the biggest concern. 

BUT…If you can build consistent images in MJ…You can build consistent videos in Runway.

Here’s an example of a CONSISTENT CONCEPT ↓

I2V INITIAL THOUGHTS:

• Still subtle hallucinations

• Images don’t lose fidelity

• Color transition is consistent

• Works very well with MJ images

• Still expensive…but less random

• Awkward natural motion at times

• Big upgrade on object permanence

• Short/punchier text prompts work well

• Text prompting really aids the generation

• Complex movement is hard with Img-2-Vid

I’m guessing Runway’s next move will be keyframes…stay tuned.

AI CREATIVE: DESIGN SYSTEMS AT SCALE

I’m tired of the lazy “Ai Bad” narrative.

If you’ve managed a team…and watched them struggle. (It’s a terrible feeling)

Ai is more than (images/videos/text)…it’s systems…that can assist your team produce quality work…(and not hate their lives)

That’s why I partnered with SUPERSIDE…to help creative teams and push boundaries.

DESIGN SYSTEMS AT SCALE:

Individual Ai tools are cool…but when you combine them…that’s when things get interesting. So we focused on building design systems that incorporate multiple tools and use/learn from data and brand materials.

Goal: Initial brief to first draft…in less than 30 seconds.

How We’re Doing It:

1. Break down a brand into components (ethos/copy/imagery/templates)

2. Analyze consistent patterns and performance

3. Organize knowledge base for each component

4. Build agents for siphoned and specific tasks

5. Connect various agents to produce first draft.

To produce a product that’s tailored to your brand and audience.

Triming a 24-hour process down to a 24-second process.

If you need a forward-thinking team of 500+ traditionally trained designers…who are now juiced with battle-tested ai-systems…

VIRAL VIDEO BREAKDOWN: MJ + RW MAKING MAGIC

I won’t lie to you…I knew this video would ruffle some feathers (it did)

MJ + RW is the power duo of the moment…and they’re impressive.

If you missed it…HERE’S THE FULL POST ↓

VIRAL APPEAL

This video had 4+ Million impressions across social platforms…And it has people divided. The goal was to test the capabilities with complex aspects like multiple people, multiple camera movements, dynamic motion, natural + artificial light, busy environments…and hands.

The craziest part?…

This video was created with about 3-4 prompts…in 3-4 hours.

Here’s some of the prompts:

MJ BASE PROMPT STRUCTURE: bokeh motion, [subjects + action], tulum {beach/jungle} rave, deep textures, long shadows, tattoos, vivacious crowd, 16mm, asymmetrical composition, [Time-of-Day], silhouettes, techno, --ar 16:9 --q 2 --p [code] --s 250 --v 6.1

MJ BASE PROMPT EXAMPLE: bokeh motion, female DJ playing turntables, tulum beach rave, deep textures, long shadows, tattoos, vivacious crowd, 16mm, asymmetrical composition, sunrise, silhouettes, techno, --ar 16:9 --q 2 --p [code] --s 250 --v 6.1

RUNWAY PROMPTING:

Keeping the RW prompts simple and easy will help.

There’s already enough going in the images…so keep the prompts manageable.

RW PROMPT: slow motion medium shot of dj playing the turntables at tulum beach rave, ethereal lighting, 35mm style

MJ + RW:

TIP: Build motion into your prompts.

The more initial motion you have the easier it is for Runway to roll with it.

Here’s what I mean

MJ PROMPT: bokeh motion, female techno fans cheering, black hair, trendy attire, tattoos, rave on the tulum beach, vivacious crowd, 16mm, asymmetrical composition, sunrise, techno, --ar 16:9 --quality 2 --sref https://s.mj.run/qMnVtR-WAGI --personalize [code] --sw 30 --stylize 650 --v 6.1

RW PROMPT: slow motion shot of woman cheering hypnotically, ethereal lighting, 35mm style

Solid Runway gens start with good base images.

That’s where complete control of Midjourney can help.

If you want complete control of Midjourney → MIDJOURNEY MASTERY

BEYOND MJ: OPEN AI TURMOIL:

The script keeps writing itself…and there’s drama at Open AI again.

Matt Wolfe breaks it down HERE

TL;DW

• OpenAI Leadership Crisis: Following Sam Altman's firing in November 2023, several co-founders and key figures have departed, raising concerns about the company’s direction.

• Recent Exits: Notable departures include Andre Karpathy, Ilia Sutskever, and John Schulman, with many citing a lack of focus on safety and alignment.

• Greg Brockman's Extended Leave: Co-founder Greg Brockman is on a long vacation, not resigning, amid speculation about burnout.

• Financial Woes: OpenAI faces projected losses of $5 billion, risking bankruptcy within a year, compounded by ongoing lawsuits.

MIDJOURNEY OFFICE HOURS (07/31):

Upcoming Releases

• Midjourney 6.2 (in ~1 month)

  • Last 6.x version

  • Balances and improves 6.1

  • Possible small depth control net

• Midjourney 7.0 (main focus after 6.2)

  • Improved aesthetics, speed, prompts, knowledge

New Features in Development

• 3D models/ Video capabilities

• Storytelling tool (possible public release)

• Web improvements:

  • Enhanced "Describe" feature

  • Combined reframe/repaint editor

• Zoom-in for cropped images (testing for 6.1)

Future Considerations

• Debating 6.x vs. 7.0 feature development

• Higher resolution output possible

• Video/3D prioritized over specific styles

• Potential model merging in 7.0 (e.g., niji mode)

Have fun.

If you need help…DM me on Linkedin.

Until next time…

Rory