GLP-1 telehealth video after the FDA warning letters
FDA warning letters made compounded GLP-1 marketing a claim-language problem. The equivalence phrases to strip, plus a who-qualifies script that clears review.
8 min read
Good abandonment: when drop-off isn't failure
A viewer who leaves after getting the answer is not the same as one who leaves bored, but average percentage viewed scores them the same. What to use instead.
9 min read
Google Vids adds avatars and footage editing
Google Vids now does personal avatars and prompt-edits real footage, with SynthID on every clip. What always-on marking means for video that runs off-platform.
9 min read
Grain, flare and vignette against the AI sheen
Generated footage reads wrong because imperfection is absent, not detail. A finishing order for grain, vignette and flare, and the amounts that hold up.
9 min read
Answering four sponsor objections to AI content
Brand safety, audience backlash, disclosure duties, and doubt the audience is real. Four scripted answers, each backed by an artefact you show on the call.
9 min read
The hand deforms as soon as it holds the product
Contact with a product is where hands fail. A grip-region mask-and-inpaint recipe, then a composite fallback that keeps the real SKU geometry intact.
9 min read
What HappyHorse's 38-second render claim means
Alibaba quotes ~38 seconds per 1080p HappyHorse 1.1 clip on one H100. What a hardware-conditioned figure predicts about a hosted queue, and what it can't.
9 min read
Hard or soft: source size in lighting prompts
Shadow edge hardness is set by apparent source size, not by adjectives. A size-and-direction vocabulary that outperforms five style words in every prompt.
9 min read
Higher-ed admissions video and FERPA consent
Identifiable students in promotional video need a signed release, which is why dorm tour projects stall every July. A consent-tiered plan that keeps shipping.
9 min read
Video for yield season and summer melt
Between the deposit and move-in day the job is reassurance, not recruitment. A personalised-at-scale video plan that needs no new footage and no new consents.
9 min read
Caregiver recruiting video for home care agencies
Agencies compete harder for caregivers than clients. A one-shift recruiting reel built on shift, pay and route honesty, with no client home ever on camera.
9 min read
When a strong hook doesn't hold
A hook that wins the swipe and loses the viewer eight seconds later is a coherence failure, not a hook failure. The two-checkpoint read that separates them.
9 min read
How many episodes before you kill a series
Commit a pilot length, one metric and a written stop rule before episode one, so two noisy duds cannot cancel a series that has not been measured yet.
9 min read
How many reference slots your shot actually needs
Single-image, three-slot and nine-slot reference models compared: which shots improve past three references, which just get slower, and a shot-type table.
9 min read
AI Info Label on Instagram: Why It Appears, How to Avoid
Meta shows AI Info on Instagram and Facebook from C2PA or IPTC signals, or from your upload disclosure. It is not a detector. A real photo can be labelled.
9 min read
How to read a GPAI training-data summary
GPAI providers must publish a training-content summary. Six questions that separate a useful one from a decorative one, and what the document cannot prove.
8 min read
How agent tools route to six sub-agents
A flat list of 139 tools serves an agent worse than six domain specialists. Name the job so the router sends the turn to the right one first.
9 min read
Identity blending: a face that isn't anyone
Weighted mixes of faces a model already knows produce a novel character that reproduces. The syntax, the ratios that stay stable, and how to lock the result.
9 min read
The face changes when the character turns
A single front-facing still cannot constrain a profile, so identity collapses on the turn. Build a multi-angle sheet, and keep the turn as a cut.
9 min read
Image-to-3D vs traditional asset pipelines
Generating a mesh from one image covers props well, hero objects partly and characters not at all. Topology, textures, rigging and the cleanup nobody budgets.
9 min read
Scoring a video from one still image
Lyria 3.5 takes an image as a musical brief. A pipeline for scoring a cut from its key frame, and how to keep one cue consistent across a multi-clip edit.
9 min read
The sign is misspelled. Fix it without re-rolling
Do not re-roll a good image for one misspelled word. Quote the exact string, describe the font, set a mask margin, and know when to composite type instead.
8 min read
Inside the digestive system explainer workflow
Eleven scenes, one character reference used twice, and 396 credits. A scene-by-scene teardown of a published explainer recipe, and how to fork it.
9 min read
Instagram predicts audio page taps on Reels
Instagram predicts Reels reshares, watch-through, likes, and audio page taps. The tap is a create-intent proxy. The track decides whose audio page opens.
9 min read