EP 003 · 56 MIN WHAT YOU DON'T SPECIFY, MIDJOURNEY PICKS
Prompt Theory in Midjourney: Our Best (Proven) Tips
The short version
Episode 3 is a proven-tips deep dive on Midjourney prompt structure: every detail left unstated in a prompt still gets filled in by the model, so the hosts write prompts as explicit fields, subject, position, lighting, background, instead of one long sentence. Rory and Drew share fixes for ignored details, an image-weight range near 0.4-0.6 for a subtle pose nudge, and real camera and lens language for a controllable look. They also flag a new post-upscale gray-distortion glitch.
Key takeaways
- 01
Break every prompt into fields, subject, position, lighting, background, instead of one long sentence: unstated details get filled in anyway.
- 02
Use documentary or photojournalism as a blanket token to stop generating people who stare straight at the camera.
- 03
Counterbalance your descriptors: asking for a blue-and-green palette while also specifying black-and-white film stock will just confuse Midjourney.
- 04
Dial image weight to roughly 0.4-0.6 when you only want a pose or subject nudged, not the whole color palette pulled over.
- 05
Fix an ignored detail by moving it earlier in the prompt or repeating the keyword: prompt hierarchy still affects what Midjourney honors.
- 06
Layer image references like Photoshop layers: generate a pattern and a person separately, then blend them instead of prompting both at once.
- 07
Watch for a post-upscale glitch: Drew found a gray, desaturated distortion appearing in upscaled images that wasn't in the original.
Receipts · said on the record
“I have an earring that's floating in space over on the right side that I just didn't catch.”
“I remember people being like, I think this one's worse. I'm like, I just don't think we understand the level of control that we have.”
“If you are listening to this, pause this and rewind that, because this is literally one of the biggest mistakes people are making in Midjourney: they're not focusing on the eyes.”
“No one looks at the camera, actually, when you're telling a story.”
“If you guys can take me off Magnific, please do. I would like to save some money. That'd be awesome.”
“Have you ever noticed when you're upscaling an image, sometimes you're getting like this gray distortion in the photo? … See, here's the original. Not there. Isn't that wild?”
Chapters
- 0:00 The Midjourney 6.5 Hype Train
- 3:00 The Realism Race: How Far Can AI Images Go?
- 6:00 Wishlist for Future Midjourney Updates
- 9:00 Glitch Hunt: The Gray Distortion Mystery
- 12:00 Prompt Theory: The Art of Structured Creativity
- 16:00 The Art of Specificity in Prompting
- 20:00 Eyes, Expressions, and Emotions: The Devil's in the Details
- 24:00 Composing in Layers & 3D
- 28:00 Camera Angles: Breaking the Symmetry Curse
- 32:00 Photography Crash Course for Midjourney Artists
- 36:00 Lens Logic: From Wide Angles to Telephoto Tricks
- 40:00 Camera Types and Their Unique Signatures
- 44:00 Photoshop to Prompts: Translating Editing Skills
- 48:00 Troubleshooting Techniques for Stubborn Prompts
- 52:00 Advanced Moves: Layering and Image Weighting Mastery
- 55:00 Closing Thoughts: The Road Ahead
Questions this episode answers
What's the core prompting rule from Fast Hours episode 3?
Rory argues that every visual element left unspecified in a Midjourney prompt still gets filled in by the model anyway, so he writes prompts as explicit fields, subject, position, lighting, background, rather than one long sentence.
How do Rory and Drew suggest fixing a prompt that ignores one of your details?
Rory checks for counterbalancing descriptors that contradict each other, like asking for blue-and-green tones while also specifying black-and-white film, and Drew repeats the keyword or moves it earlier in the prompt, since word order still affects emphasis.
What image-weight range do the hosts suggest for a subtle pose nudge?
Drew says he typically dials image weight down to about 0.4 to 0.6 when he only wants to borrow a pose or rough subject from a reference image without pulling its full color palette into the new image.
What Midjourney bug do Rory and Drew flag in this episode?
Drew shows a gray, desaturated distortion that appears in specific spots of an image only after upscaling and isn't present in the original generation; neither host had found a fix beyond re-rolling or repainting it.
What camera and lens knowledge do the hosts say helps with Midjourney prompts?
Rory and Drew say learning real photography basics, shutter speed, focal length such as 35mm versus 100mm, and named cameras like a GoPro, iPhone or Leica, gives prompts a more specific, controllable visual signature.
Show notes
Prompt Theory in Midjourney... What is it? How does it work? Why the hell does it matter?
We're back with a power-packed episode full of practical tips on prompting.
In this episode, we're sharing: ✔ Our top hacks for photo-realistic images in Midjourney ✔ Simple tweaks to dramatically improve your prompts ✔ Secret techniques for controlling facial expressions and poses ✔ How to think in 3D for better 2D compositions ✔ Photography basics that'll level up your Midjourney game ✔ Troubleshooting guide for when your prompts just aren't working + more
We're cutting through the fluff and giving you actionable advice that you can use right away.
Whether you're struggling with stubborn prompts or looking to add that extra oomph to your creations, we've got you covered.
Stick around till the end for some mind-blowing advanced techniques that'll set your art apart.
Ready to level up? Let's dive in!
