Field notes for making AI video at scale
Workflows, prompt craft, and honest tool breakdowns for turning long scripts into finished video without losing your weekend.
Reference images vs text: how to lock one AI character across hundreds of shots
In 2026 the face lives in the reference image, not the words. Here is what each one is actually good at, the layered stack that beats either alone, and how to hold one character across a few hundred shots without re-describing them every time.
Latest
What long-form AI video actually costs, once you do the math
A five second clip costs pocket change. A three-hundred-shot video does not. Here is where the money actually goes, why the per-clip price is a trap, and the four levers that keep the bill sane.
How to route hundreds of AI prompts to your tools without the chaos
You wrote the prompts, locked the characters, planned the shots. Now you have to actually get them into your image and video tools, in order, by character, without losing an afternoon to copy-paste. Here is the system that scales.
How to keep an AI character consistent in video, not just stills
You locked the face in your stills. Then you hit play and it wanders. Here is how to carry one character into motion: why image-to-video changes everything, and the short-clip workflow that actually holds.
How to keep one AI character consistent across a hundred shots
Text gets you close. References get you most of the way. Training gets you there. Here is the ladder of character consistency, from the free habit to the heavy artillery, and exactly when each rung is worth the work.
The anatomy of a great AI image prompt for video
A vague prompt gives you a wish. A structured one gives you a shot. Here is the four-part formula we use for cinematic, consistent prompts, with examples you can copy.
Get the workflow before everyone else
Join the waitlist and see the full BatchFrames pipeline that turns your script into finished prompts.
Join the waitlist