MIDJOURNEY's YUGE UPDATE + Runway’s Impending Domination

Fam,

I can’t even keep up.

You step away for one day and you’ve missed like 15 updates.

This might have to be a 2x-week newsletter.

No time for chit-chat…let’s do it.

Here’s what we got for today:

1. MIDJOURNEY UPDATE: Multiple Style Codes

2. MIDJOURNEY: Multiple --P Codes

3. New Partnership

4. RUNWAYS’s Impending Domination

5. The Annoying SORA Playbook

6. Beyond Midjourney: Googles Video-to-Audio

7. Midjourney Office Hours Update.

Let’s Boogie.

MIDJOURNEY UPDATE: MULTIPLE SREF CODES

Midjourney was on fire last week.

And these updates were much needed.

Here’s What’s New:

1. Multiple SREF Codes

2. Style Code Weighting

3. Multiple Personalization Codes

4. Personalization Code Weighting

If you missed it…Here’s the full LINKEDIN POST.

HOW TO USE MULTIPLE STYLE CODES:

1. Create a prompt

2. Add “--sref” to end of prompt

3. Add a style code

4. Then a space 

5. Add another style code

6. Generate.

Multi SC Format: Prompt --sref [code1] [code2]

Multi SC Prompt Example: Award winning photo of an epic ancient warrior --ar 5:6 --sref 1497823114 2360446404

Note: You can use as many as you want (2, 3, 4, etc)…go wild.

Prompt: Award winning photo of an epic ancient warrior --ar 5:6 --sw 200 --stylize 550 --sref 1497823114 2360446404 2346775807

STYLE CODE WEIGHTING:

Best part of the update?…You can weigh the style code's influence.

This is where the possibilities become infinite.

WTF DOES THIS MEAN?

1. Weighting will add more/less influence 

2. From one code or another

3. Meaning if a code is weighted 2:1

4. (Example --sref [code(x)]::2 [code(y)]::1

5. Code (x) will have 2x the influence vs Code (y)

6. Meaning it will come through stronger

HOW TO WEIGHT STYLE CODES:

1. Create prompt + add style codes.

2. After style code type two colons “::”

3. Then add a number (ex. 1, 2, 3, etc…)

4. You can add to both codes or just one.

5. Hit generate.

EXAMPLE PROMPT: Prompt: Award winning photo of an epic ancient warrior
--ar 4:5 --sw 200 --stylize 550 --sref 1497823114::1 2360446404::1 2346775807::3

QUICK THOUGHTS:

1. You can get super creative…and generate unique looks

2. Utilize SW parameter to add more/less style code influence

3. Style codes take over an image

4. What you gain in style you lose in control

5. If MJ isn’t listening to your prompt…drop the SW value.

MORE INFO ON STYLE CODES + HOW TO TEST THEM:

MJ UPDATE #2: MULTIPLE PERSONALIZATION CODES

Ok we’ll keep this one short…but it’s worth noting.

In the same fashion…you can now utilize multiple personalization (--p) codes.

If you missed it: CHECK OUT LAST WEEKS NEWSLETTER

HOW TO USE MULTIPLE PERSONALIZATION CODES:

1. Create a prompt

2. Add your (--p) code to the end

3. Add a space

4. Then add a second code

Format Example: a man in a park --ar 5:6 --p [code1] [code2]

NOTES:

1. You only have (1) P-code

2. So you’ll need to find a friend or someone who’s shared theirs

3. You can also weigh the P codes

4. Use the same “multi-prompting” technique as the last section.

Luckily my buddy Drew Brucker gave me his code to test this.

Example Prompt: Award winning photo of an epic ancient warrior --ar 5:6 --sref 1497823114::3 2360446404::4 2346775807::3 --personalize [my code]::1 [drews code]::2 --sw 250 --stylize 700

NEW PARTNERSHIP: DFY CREATIVE W SUPERSIDE

Last week we pulled back the veil on what we’ve been working on.

The response was amazing…and I REALLY appreciate all the support.

If you missed it…WE’RE PARTNERING UP.

WHAT THAT MEANS:

• We trained a team of 100s of traditional designers to utilize AI

• We tested the systems + processes for (12) months

• And across 500+ client projects…

• We saw 200% faster completion from brief-to-delivery

• 40-85% reduction in hours spent on projects

• SAVING clients over $1.4 Million.

THE REASON I PARTNERED WITH THEM:

• They’re not a BS agency (that’s only using Ai)

• They’re traditional designers, creatives, animators, etc.

• They’re taking Ai seriously and solving complex problems

• And they have the horsepower to take on massive projects.

If you’re an ambitious brand looking to scale…

BEYOND MJ: RUNWAY’S IMPENDING DOMINATION

Runway has been creeping in the shadows for a while.

I was wondering where the hell they’ve been.

Turns out…they’ve been working on Gen-3 and it looks ridiculous.

Ai video generation will not be the same after they release it.

Check out the videos in THIS THREAD.

MASSIVE IMPROVEMENTS:

Runway has always had the edge in “control”.

Tools like LUMA and other generators lack this granular control…and that’s what separates Runway.

Now, if they’re able to pull this off…I think we’ll all stop talking about SORA for a while.

THE PROMISES:

With all these releases there’s promises…but let’s see if they keep them.

It’s looking like longer videos, increased visual acuity, higher resolution outputs, ability to handle text, increased prompt coherence, sharper and more dynamic camera movements, etc.

I’m personally putting my eggs in their basket (because of the control features).

And apparently it’s releasing “in the next few days”.

IMO, I hate the “in the next few days” strategy…see more about that below.

THE ANNOYING SORA PLAYBOOK:

I waxed poetic on a post this weekend…because i’m tired of the “SORA” shit.

I’m really hoping Runway doesn’t go down the same path.

TL;DR: SORA’S ANNOYING PLAYBOOK:

SORA jumped out of the gate with the most revolutionary tech we’ve seen.

Then they went silent…hoarding the tech for internal creatives to boost their profiles and create infinite sharability on social media.

Other tools took notice…and are doing the same thing (it’s annoying)

(Looking at you Magnific/Runway)

Here’s the FULL POST

THE PLAYBOOK:

• Big jump in tech/new features

• Hoard tech to a small exclusive group

• Blast the sh*t out of social with content

• Boost internal creators profiles/personal brands

• Build a “convertible” organic marketing channel

• Create the biggest collective scarcity mindset

Then…launch the product (at some point/TBD/coming soon)

With no date in mind…and make us all wait.

It’s transparent…and it’s annoying.

But…It works.

GOOGLE VIDEO-TO-AUDIO:

Another day, another problem solved with AI.

There’s nothing worse than searching for stock audio or sound editing a video. (Basically, stock anything)

So google solved that problem with Deepmind V2A.

Check out the VIDEO BELOW

Might seem inconsequential, but this solves a major problem.

Here’s the BLOG POST FROM GOOGLE.

MIDJOURNEY OFFICE HOURS (06/19):

Nothing crazy here that we didn’t cover above.

Training Midjourney 6.5

• Focus on improving image quality, addressing edge cases

• Expected to have really good image quality

• Also updating upscaler quality

• Release likely in a few weeks due to long training time

Midjourney Version 7 (V7) In Progress

• Individual parts are coming online and looking good

• Datasets being prepared for final training runs

• Expected to be a dramatic improvement over V6

• V7 aims to be smarter, have better image quality, aesthetics, and be faster and cheaper than V6

• Possibility of creating a slower, more expensive, higher-quality version of V7 after initial launch

Website Improvements

• Adding polishing touches to the website based on user testing

• Gradually allowing random users to access the website without Discord

• Adding a chat room for help on the website

• Experimenting with community features like trends and popular sref codes

Have fun.

If you need help…DM me on Linkedin.

Until next time…

Rory