EP 034 · 70 MIN HIS TRANSPARENT LOOK: 2 MOODBOARDS DEEP
Letting AI Cook & The Final Countdown for v7 Feat. Allar Haltsonen
The short version
Episode 34 features Estonian AI artist Allar Haltsonen breaking down the moodboard-and-personalization stack behind his signature "transparent" Midjourney aesthetic, plus his fashion workflow in the try-on tool Caimera. Rory and Drew press him on aspect-ratio strategy and close with predictions for Midjourney V7: stronger prompt coherence and Omni-Reference character consistency, but no dramatic jump in raw image quality.
Key takeaways
- 01
Stack two moodboards, one built from style-reference images for a specific look and one from your own best archived work, then mix both through your personal code.
- 02
Generate at 3:4 first to nail the main subject in sharp detail, then expand to 9:16 or 16:9; starting wide from the outset scrambles the composition, Allar says.
- 03
Stacking personalization on top of moodboards fights Midjourney's plasticky "smoosh face" skin, but Allar says results still read synthetic up close on fashion work.
- 04
For fashion product shots with varied body types, Allar uses Caimera: one reference image for style, one product image, then its "try on" tool composites them.
- 05
Rory's take: gatekeeping shouldn't decide who counts as a real artist, since not knowing Photoshop's thousands of buttons doesn't disqualify someone with strong taste and ideas.
- 06
Expect Midjourney V7's biggest gains in prompt coherence and Omni-Reference character consistency, not raw image quality; the hosts predict it'll be solid but not a crazy leap.
- 07
Square (1:1) is the only aspect ratio with true symmetry, which Rory calls the default choice for centered logos, buttons, or any perfectly symmetrical object.
Receipts · said on the record
“The hard part is when I was trying to build my perfect fashion mood board stuff, I got pretty nice results, but the skin is somehow not there from the mood boards yet.”
“I know what's in my head, how I want it to look like, so I can just use AI tools to achieve it.”
“I honestly think it's creating more artists than we've ever had. Everyone has ideas. Maybe you just don't have the technical skills to use Photoshop.”
“People hold on to power no matter what. So from that aspect, I get the gatekeeping. I also get the other aspect, obviously.”
“It's probably one of the biggest … hidden factors of mid-journey, but the generation aspect ratio can change the entire trajectory of where you're going with stuff.”
“I hate it. This is where I'd love to take it in the Midjourney and just edit it. But why is that so easy in Midjourney and so hard everywhere else? It's not like there's not one million data images with a brick wall.”
Chapters
- 0:00:00 Introduction to Allar
- 0:01:36 Exploring aesthetic styles
- 0:05:36 Caimera tool walkthrough
- 0:11:57 Fashion and design processes
- 0:14:05 Concepting and visual design
- 0:20:43 Technical strategies
- 0:25:06 AI in the creative industry
- 0:29:46 Managing social media output
- 0:34:00 Organized archives and discovery
- 0:39:01 Prompting and aspect ratio insights
- 0:54:57 Future of Midjourney and wrap up
Questions this episode answers
Who is Allar Haltsonen?
Allar Haltsonen is an Estonian AI artist known on X, Instagram and LinkedIn for fashion- and automotive-themed Midjourney work. He told Fast Hours he has no formal design, fashion or photography background and builds his aesthetic from moodboards, style-reference codes and personalization.
What did Fast Hours predict for Midjourney V7?
Drew guessed V7 would ship around the last week of March 2025 based on office-hours chatter, and both hosts expected its real upgrade to be prompt coherence and Omni-Reference character consistency rather than a dramatic jump in raw image quality.
How does Allar Haltsonen get his "transparent" Midjourney look?
He mixes two moodboards, one built from style-reference images for a frosty, transparent tone and one drawn from his own best archived work, then runs both through his personal profile code.
What is Caimera and how does Allar use it for fashion work?
Caimera (also called Chimera AI) is a fashion-focused image generator Allar has tested for about two years. He feeds it one reference image for style and one product image, then uses its "try on" feature to composite the two, and says it renders a wider range of body types than Midjourney typically does.
What aspect ratio does Fast Hours recommend starting with in Midjourney?
Both Allar and Rory say to generate at 3:4 first to lock in the main subject's fine detail, then expand outward to 9:16 or 16:9; starting in a wide or tall frame from the beginning scatters the composition, they say.
Show notes
Fast Hours Podcast Ep. 34—Letting AI Cook & The Final Countdown for v7 🎧 Listen now on YouTube, Spotify, or Apple Podcasts.
Drew and Rory sit down with Estonian AI chef, Allar Haltsonen, to talk about his creative process, spontaneous idea generation, and all the other tools in his proverbial toolbox.
They wrestle with prompting, the never-ending battle for realistic fashion textures, and the fine line between aesthetic genius and absolute chaos.
The trio dives into the weird intersection of fashion, automotive design, and whatever else happens when you let AI cook. They also nerd out over aspect ratios, predict the future of AI in video storytelling, and try to figure out what Midjourney V7 might have in store.
