EP 010 · 68 MIN MIDJOURNEY WON'T LET YOU SAY 'BIKINI'
Midjourney AMA: Answering Your Most Burning Questions (Part 1)
The short version
Episode 10 is part one of a two-part Fast Hours AMA answering listener questions about Midjourney prompting. Rory and Drew cover multi-prompting syntax for wild, unpredictable results, dodging content-moderation keyword blocks with literal phrasing like 'environment-appropriate attire,' and troubleshooting a real-photo-to-cartoon conversion step by step. They also recap office-hours news on V7, an incoming 3D model, and Midjourney's first video model, both due before year-end.
Key takeaways
- 01
Multi-prompt with the :: operator to blend two unrelated ideas, e.g. 'documentary photo of a healer :: red and orange color scheme,' instead of one combined phrase.
- 02
Dodge content moderation with literal phrasing: 'man on the left kissing a man on the right' instead of 'same-sex couple kissing,' or 'environment-appropriate attire' instead of 'bikini.'
- 03
Juxtapose two contrasting style codes in one prompt, dark plus vibrant or photorealistic plus illustrative, for a more interesting blend than either code alone.
- 04
Blend two unrelated images directly with no text prompt for a 'juxtaposition' effect: Drew's example is a glass vase sitting on a stone slab.
- 05
Fixing an over-illustrated image: drop image weight to 0.5-0.8, raise style weight to 150-200+, or add '--no cartoon, illustration' instead of re-rolling the whole prompt.
- 06
Click any Explore-page creator's name to browse their full public portfolio and prompt structures for inspiration; there's no follow or DM option, just browsing.
- 07
Midjourney can't exactly replicate an external clothing pattern, but a pattern generated inside Midjourney transfers onto other images reasonably well via image weight.
Receipts · said on the record
“Really the dumbest thing ever. Who even talks about it anymore? … we can't, because they just pumpfaked us on being able to use it and then disappear for nine months.”
“I still think permutations are so underrated … it's not only an efficient tool to get to the right images faster, but I think it really just kind of scratches the itch of like, did I get everything I wanted to get out of this”
“Any glass touches stone, it's probably breaking, right? So another unique way to use something like a blend is to take juxtaposing things, mash them together … People make livings off of juxtaposing styles, right?”
“It's why every woman came out looking like a model for the first couple iterations of Mid Journey … whatever they were scraping off the internet, whatever is the most readily available photos, it's gonna be model photos, fashion photos, brand photos.”
“Environment-appropriate attire gets around everything for me when I can't do it. Also, whenever I ask for women in fitness attire, they give me a bikini. I was just like, that's not fitness attire or athleisure. That's a bikini.”
Chapters
- 0:00:00 Episode Introduction
- 0:01:08 Budapest Trip Recap
- 0:05:55 Office Hours Updates
- 0:10:10 Image Editor Features
- 0:15:15 Video Model Discussion
- 0:19:15 3D Model Plans
- 0:22:32 VR and AR Future
- 0:26:43 Hardware Projects
- 0:30:00 AMA Time
- 0:33:10 Creative Techniques
- 0:47:30 Content Moderation Workaround
- 0:52:45 Subscription Plans
- 0:54:00 Explore Page & User Discovery
- 0:59:24 Style Transfer
- 1:07:00 Episode Wrap-up
Questions this episode answers
What is multi-prompting in Midjourney, and how do you use it?
Rory explains multi-prompting as splitting a prompt into separate weighted ideas using a double colon (::) -- for example separating a scene description from a color scheme -- so Midjourney generates each part somewhat independently and blends them, producing more unpredictable, creative results than one combined phrase.
How do you get around Midjourney's content moderation filters?
Rory and Drew describe scenes in plain, literal terms rather than loaded keywords: 'man on the left kissing a man on the right' instead of 'same-sex couple kissing,' or 'environment-appropriate attire' instead of 'bikini.' There's no official banned-word list, so it's trial and error, and Rory sometimes asks ChatGPT for alternate phrasings.
Can Midjourney exactly replicate a clothing pattern or product?
No -- Drew and Rory agree Midjourney can't get an exact match on an external item like embroidery or a specific plaid. A pattern generated inside Midjourney, though, can be transferred onto other images fairly well using image weight and image prompting.
How do you fix a Midjourney image that's too cartoonish when you wanted realism?
Working through a listener's Charmander example, Rory and Drew suggest lowering the image weight of an illustrated reference to around 0.5-0.8, raising style weight to 150-200 or higher, adding a '--no cartoon, illustration' parameter, or simply using an already-realistic reference image instead, optionally stacking character reference to preserve the character's identity.
Can you see who liked your Midjourney images or contact another creator?
Drew says no: there's no notification system for likes and no way to follow or message another user, but you can click any Explore-page creator's name to browse their full public portfolio and study their prompt structure for inspiration.
Show notes
We're back with a bang, answering your hottest Midjourney questions, AMA-style.
First, we catch you up on all the key updates from Midjourney's Office Hours, including batch generation, personalization, the upcoming video and 3d models, and more.
Then we'll dive head-first into listener questions in an AMA format, sharing insights on prompting techniques, juxtaposition, pattern matching, content moderation, community engagement, image transformation techniques, the importance of specificity in prompts, and the challenges of achieving exact matches in AI-generated images.
