Ai CAUTION on Joe Rogan + Visceral Midjourney "Light Tokens"

Fam,

I am finally coming up for air.

The newborn is 3 weeks old…and we’ve now topped (4) hours of sleep.

Getting there…but that won’t slow us down here.

Here’s what we got for today:

1. Midjourney Mastery Update

2. Juicing your images with “Light Tokens”

3. Beginner’s Guide to Style Random

4. SREF + CREF for Storytelling

5. ElevenLabs massive update

6. Ai Caution on Joe Rogan

7. Midjourney Office Hours

Let’s Boogie.

JUICING YOUR IMAGES WITH “LIGHT TOKENS”

Lighting is severely underutilized in Midjourney prompting.

You take it for granted because MJ does a great job with it.

But it’s something you should know how to control.

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

WHY LIGHTING?

Because…it shapes perception, adds depth, and creates emotion.

And most people neglect it.

So consider it when prompting because it makes a difference.

Lighting makes everything pop more.

So here’s a few areas to focus on:

1. Prompt Structure

2. Light Quality

3. Light Direction

4. Time of Day

5. Descriptive Keywords

I won’t bore you all by going through every category.

So here’s a FREE DOWNLOAD of the CAROUSEL + PROMPTS + TOKENS.

But, we’ll talk about some advanced uses below.

PROMPT STRUCTURE:

This is NOT highlighted enough…but your prompt structure plays a major role.

MJ likes to weigh tokens HEAVIER at the BEGINNING of the prompt.

VS the END of the prompt.

So use this to your advantage.

If you want the lighting to come through strong…put it first.

If it’s inconsequential to you…put it last.

Example Prompts:

Lighting Last Prompt: editorial color photo, hardworking farmer in Bali rice fields, Leica M10-R, shallow depth-of-field, leading lines, backlight, --ar 5:4 --stylize 200

Lighting First Prompt: backlight, editorial color photo, hardworking farmer in Bali rice fields, Leica M10-R, shallow depth-of-field, leading lines, --ar 5:4 --stylize 200

MULTI PROMPTING + LIGHTING:

Now, I think most of you are beyond beginners…so utilize Multi-prompting to your advantage with lighting.

If you’re not familiar with multi-prompting…

Here’s the FREE DOWNLOAD for the BEGINNER’S GUIDE to Multi-Prompting.

Oftentimes, I’m not going crazy with the multi-prompting…you only need a subtle increase in weight to really make a difference.

Prompt Structure: ethereal lighting, deep textures::1.4 [shot type]::1.3 [subject + action], [environment], [camera details], leading lines::1.2 --ar 3:1 --stylize 200

Prompt: ethereal lighting, deep textures ::1.4 editorial color photo ::1.3 hardworking farmer in Bali rice fields, Leica M10-R, shallow depth-of-field, leading lines ::1.2 --ar 3:1 --stylize 200

LIGHTING + SREF:

If you get an image you like…use it across multiple images with SREF.

Typically, I’ll use the same prompt and adjust the [subject + action].

Also, play around with Style Weight [--sw parameter]…(Values between --sw 30-90 work really well here.)

Original Prompt: ethereal lighting, deep textures ::1.4 editorial color photo ::1.3 hardworking farmer in Bali rice fields, Leica M10-R, shallow depth-of-field, leading lines ::1.2 --ar 3:1 --stylize 200

SREF Prompt: ethereal lighting, deep textures, editorial color photo, {man riding a motorcycle in the bali jungle, woman kneeling feeding a dog on a bali backroad, woman swimming in an infinity pool overlooking the Bali countryside} Leica M10-R, shallow depth-of-field, leading lines --ar 5:6 --sref https://s.mj.run/AbsP16Q82KY --sw 50 --stylize 200

I could go on forever…but I’ll cut it off there.

BEGINNER’S GUIDE TO STYLE RANDOM:

Last week we talked about getting creative with Style Random.

So I did a full walkthrough.

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

SREF’S BIG ROLE:

SREF is becoming more and more important in MJ prompting.

In this case…once you have a style code that you like…you can use the images generated with the style code as an SREF.

Blending them is when the magic happens.

For example… here are (2) distinctly unique styles…

SREF CODES:

1. --sref 3474327402

2. --sref 3210011585

BLENDING STYLE CODES:

To blend style codes…simply write a new prompt.

Add the images generated with codes as SREFs…

And make sure there’s no style code parameter at the end of the prompt.

Example Prompt: street-style photo of a samurai --ar 4:5 --sref https://s.mj.run/Z4HI1jtqoQA https://s.mj.run/cOqUaD4CPZc --stylize 500

If you want to go down this rabbit hole.

Here’s the FREE DOWNLOAD of the FULL GUIDE.

SREF + CREF FOR STORYTELLING:

If you can master Style Reference (SREF) + Character Reference (CREF) there’s no reason you can’t create full stories with Midjourney.

This will become increasingly more important with the development of tools like SORA, Runway, etc.

A few quick tips:

1. Start a setting or environment first.

2. Then build your character to match the setting.

For example…If you want to create a scene of a volcanologist and an erupting volcano.

Start by generating the volcanic environment…then use SREF to build your character.

Then use CREF to transfer your character to multiple poses/actions.

Here’s a sneak peek of what I’m talking about on this X POST.

Prompt Structure + Prompts in post.

I’ll be doing a longer breakdown on it this week. (Stay tuned)

ELEVENLABS MASSIVE UPDATE:

Well…this is a good one.

Elevenlabs just rolled out “Text-to-Sound” and it’s super helpful.

The same way Midjourney rendered stock imagery useless…This will make stock sounds useless…

There’s literally nothing worse than sifting through stock sounds looking for something to fit your video…(video editors…I know…you know)

AI CAUTION ON THE JOE ROGAN EXPERIENCE:

Admittedly, I have been a JRE listener for 10+ years…because of episodes like this.

Long-form content is the only way to talk about subjects like Ai.

The topic is too nuanced…and there are too many variables to discuss in a 5-minute segment.

If you’re even remotely interested in AI…give this episode a listen.

Jeremie and Edouard Harris provide their cautionary stance on Ai.

I think it’s important to see the massive benefits and the potential downside of Ai as it develops.

LISTEN HERE

MIDJOURNEY OFFICE HOURS (5/29):

Midjourney Website Updates

• Testing onboarding process for users signing up without Discord

• Considering blending social and non-social spaces better in design

• Working on improving mobile experience

Midjourney Version 6.1 / 6.5 Release

• Close to releasing version 6.1, most major aspects working

• Final training run may take several weeks once all parts working

• Focusing on image quality, speed, coherence, faces, skin, hands, bodies

• Debating whether to call it version 6.5 due to the significance of changes

• 6.5 will be very consistent with v6 in style and prompts

Midjourney Version 7 Plans

• V7 to have different aesthetic due to new training data

• New aesthetic system will look very different from v6

• Many other features hoped for like batches and better prompt understanding

Personalization Feature

• Getting closer to releasing

• Adding more features before release, may delay by a week

• Combined with V6.5 will be very powerful

Have fun.

If you need help…DM me on Linkedin.

Until next time…

Rory