Make the hero, then make him the reference for everything else. The hero prompt writes the identity once: a South Asian man in his early 30s, medium-deep brown skin, short textured black hair, thick black brows, short dark stubble, dark brown eyes, photographed straight after coming in from minus 20 degree wind and snow.
For shots 02 to 08 I attached two images: the hero I kept as the first reference (who he is) and a framing shot as the second reference (how the crop and light should look). Every prompt repeats a short version of the identity so he never drifts.
If a framing reference has different hair, say so: keep his hair exactly like the first reference.
Every shot names the same room: deep navy seamless backdrop, one large softbox from camera left, natural falloff into shadow on the right. Every shot names the same camera: medium-format film, 100mm (macro for the close-ups), f/8.
Repeating the set in each prompt is what makes eight separate generations feel like one real shoot.
AI loves smooth, warm, perfect faces.
Real cold only shows up when you ask for it by name:
Describe the frost precisely: fine, feathery, matte, needle-like. If you only say "frost crystals", the model tends to string clear glass beads along the hairs.
Every prompt ends with the same refusal line: No beauty retouching, no smoothing, no HDR, no waxy or oily highlights, no sweat, no plastic look, no CGI. Fine film grain. No text. For the frost, add what it is not: not round beads, not glass droplets, not glitter, not snow paint.
Ask for dry, matte skin, so the only sparkle in the frame comes from the ice.
Two image references per shot as described in step 01. I cropped the finals to 4:5 (1080 x 1350) for the carousel and put text only on the hero and the last slide.
QC zoomed in. Regenerate that one frame when:
Keep the rest.
The eight ready prompts are in the members pack below.
This guide is the full process. The ready-made version lives inside AI Creative OS on Skool, next to every other prompt pack I post.
