GUIDES
Macro is a depth problem
People talk about macro as if it were a magnifying glass: get closer, see more. On a real copy stand the first thing you notice is not more detail. It is how little of the object is sharp. At 100mm, close enough that a pearl fills most of the frame, the plane of focus is a few millimeters thick. The front of the nacre can be crisp while the rim goes soft, and the post is already gone. That is not a software bug. That is optics.
At 100mm and a pearl that fills the frame, the sharp plane is thinner than the distance from nacre to post. One exposure cannot hold the front of the pearl, the rim, the setting, and the post. Marketplace jewelry photographs pretend otherwise. They look "all sharp" because several focus planes were stacked, or because the photographer stopped down until diffraction started to hurt, or both. The prompt has to ask for that result, or the model will pick one plane and let the rest melt.
Why the prompts say focus-stacked
The pearl stud in a dark-field stone setting is explicit: "100mm macro lens," close enough that the single earring fills the frame, and a "focus-stacked appearance with the front of the pearl, its rim, setting, post, and immediate case surface all cleanly sharp." That list is the depth budget. Four surfaces sit at four distances. The phrase is not decoration. It is the only way a single generated frame can imitate what a stacking rail does on the bench.
The solitaire on black velvet has a worse budget. An extreme macro of one stone, with a tiny water droplet on an upper facet, has to keep the droplet, the near prongs, and the rear facets in the same picture. The note says depth of field is thinner than the distance between the near prongs and the rear facets. "Macro focus stacking keeps all visible facets, prongs, and the tiny droplet crisply resolved." A single exposure would force a choice: the droplet or the cut.
I write the stacked look as a result, not as a process. The model does not need a lecture on helicon or a rail. It needs to know that the front and the back of the subject stay readable. If you omit that clause, you often get a pretty highlight on a soft lump, which is the opposite of a jewelry file.
The highlight is a depth cue too
A pearl is a sphere. The prompt asks for "a single crisp pinpoint circular highlight on the pearl that defines its roundness." That highlight is a lighting instruction, and it is also a depth instruction. One small catchlight sitting slightly off center tells you the surface turns away from the camera. A smear of glare does not. A rectangular softbox reflection is worse: it pastes a window onto the nacre and destroys the sphere.
The same prompt forbids "rectangular softbox reflections" and "square reflections." Those words are there because generators love a broad key. Broad keys are easy. They also print their own shape. On a ring shoulder a rectangle is merely ugly. On a pearl it is a lie about the form.
The solitaire solves the same problem with a different fixture. A "very large seamless diffused dome" wraps the polished metal in a curved band so the prongs read as cylinders, not as mirrors of a softbox. The stone then gets its own light: "controlled rear facet illumination enters the diamond from behind." Front light alone leaves a clear stone looking like glass. Rear entry light is what jewelry photographers mean when they talk about making the cut work. Two lights, two jobs, both named.
Contact shadow keeps the object on the table
Macro also exaggerates float. When the background is black velvet or dark stone, and the object is pale, a missing contact shadow makes the jewelry hover. The pearl prompt asks for a "realistic contact shadow aligned with the lighting." That is a small darkening where the setting meets the case insert. It should match the key: if the tiny round source is high and slightly forward, the contact is tight and a little rearward. If you leave the clause out, you often get a cutout.
Velvet is useful here because it swallows spill. The solitaire sits on "deep black velvet with visible fine pile." You want just enough pile to prove the surface is cloth, and not so much that fibers climb the shank. Dark-field treatment plus "narrow controlled rim highlights" keeps the metal from merging into that pile. Macro contrast is binary if you let it be: either the outline is drawn, or the object is a bright blob in a hole.
I also watch the case edge. The pearl page wants the earring to fill the frame while the edges of the case or tray remain identifiable. That is a scale check. Without a tray lip or a stone surface, a lone pearl can look like a planet. The contact shadow and the case together tell you this is an object you could pick up.
What not to add
When a macro frame fails, the instinct is to add words: "ultra detailed," "8k," "intricate nacre," "every facet perfect." Those adjectives do not move the focus plane. They encourage the model to sharpen texture everywhere, including places that should go slightly soft, and to invent extra highlights. You get a busy pearl and a still soft post.
Name the lens. Name the distances that must stay sharp. Name the one highlight. Name the contact shadow. Forbid the rectangular reflection. That is a macro prompt. The pearl stud and the solitaire already do this. Copy one of them, then change a single control if you want to see what the depth budget does when you move the camera, the droplet, or the fixture. Changing one thing is the other half of the method, and it has its own note.