IMAGE AI·APR 24·11 MIN
Z-Image Turbo vs Flux Schnell is not a drop-in swap
Z-Image Turbo vs Flux Schnell: same prompt, different brain. Rich layered prompts that land in Flux collapse in Z-Image. Empirical evidence across rounds.
READ →
Topical cluster
Local AI Image Generation
The pipeline behind a solo operator's image stack. A Mac writes bespoke prompts and reviews outputs, a Windows box with the dGPU runs the diffusion models, a feedback ledger turns hundreds of reviews into a prompt taxonomy. Notes on Flux Schnell vs Z-Image Turbo vs Qwen, swap-pressure safety, cluster-uniform visual systems, and the tooling that keeps the loop from eating the week.
4 postsFor: Creative technologists and solo operators generating production imagery locally
Go deeper
IMAGE AI·APR 24·14 MIN
mflux ate my Mac: the safety wrapper I wish I'd written first
Running mflux in a loop crashed my Mac mid-batch. Here is the mflux memory safety wrapper I wrote after: swap gate, cooldowns, timeouts, pkill escape.
READ →
IMAGE AI·APR 24·12 MIN
The image feedback ledger that became a prompt taxonomy
Across 21 review rounds and ~500 AI image generations, a small JSON ledger became an AI image prompt taxonomy. Here is what it caught and missed.
READ →
Why this matters.
Running diffusion models locally became a real option around mid-2025 once Flux Schnell and Z-Image Turbo landed with open weights. By 2026 the math is straightforward for a creative-tech operator: a one-time GPU investment beats a monthly cloud-API bill within the first quarter, and the privacy and iteration speed of a local pipeline is a different category of working entirely. The catch is the surrounding tooling. Prompt engineering, output review, ledgering, taxonomy work, and the orchestration between the prompt machine and the GPU box make or break the actual velocity.
This cluster is the journal of building that pipeline as a one-person studio. Mac as the prompt and review surface. Windows or Linux box with a dGPU as the diffusion runtime. A feedback ledger that turns hundreds of reviews into a documented prompt taxonomy. Decisions between Flux Schnell, Z-Image Turbo, and Qwen for different use cases. Swap-pressure safety so a model upgrade does not silently regress the visual system. Cluster-uniform image styles for editorial use.
If you generate production imagery and the cloud-API bill is creeping past your iteration budget, start with the local-pipeline hub piece before the next render queue.
Put this to work
Running Flux, Z-Image, and Qwen locally without the cloud-API bill.
> See the Operator's StackOther clusters
Attribution
Attribution & CAPI
Server-side tracking, dedup, and the math behind DTC attribution.
Shopify
DTC Shopify Infrastructure
Theme architecture, metafields, agent-orchestrated builds, Hydrogen decisions.
Healthcare
Healthcare & Compliance
HIPAA-aware Next.js, audit trails, and regulated DTC patterns.
Solo brand
Creative-Tech Solo Brand
The hybrid creative-director-who-codes operating system.
Pricing
Services Business & Pricing
Productizing services, pricing strategy, and the retainer exit.
Lifecycle
Email & Lifecycle Marketing
Klaviyo flow architecture, retention math, and the LTV playbook.
Agents
AI Agent Engineering
Claude Code sub-agents, MCP servers, skills, and the orchestration stack.
Analytics
Analytics & Data Infrastructure
GA4, BigQuery, and the warehouse-first analytics rebuild.
Conversion
Ecommerce Conversion & UX
PDP patterns, cart decisions, checkout extensions, and the CRO stack.
Programmatic SEO
Programmatic SEO & Content Ops
Scaling content without burning author brand authority.
Paid Social
Paid Social Performance
Meta, TikTok, and Google Ads for DTC operators who also own the data.
Brand
Brand Architecture & Design Systems
Visual identity, naming ladders, and design systems that scale with the business.
Shopify Apps
Shopify App Ecosystem
Selecting, configuring, and replacing the third-party stack on Shopify.
Fractional
Fractional Ops & Service Leadership
Running a high-leverage fractional practice without becoming an agency.
Let’s fix
some problems.
Three short steps below. I read all of these, it’s just me on the inbox. Usually you get a real reply within a day, sometimes the same day if I’m at the desk.
or email direct hello@michaeldishmon.com
01 / 03WHAT YOU NEED
$ cat lead.json | mail -s 'new signal' michael
