
Original document
GPT Image 2 · en
Scene:
A square false-color SEM microscopy plate with muted amber and teal tinting over an otherwise electron-microscope tonal base. The sample field is viewed from a slightly elevated angle, with a shallow topographic shadow and a clean black bottom information strip. A lower-left scale bar area and small calibration blocks are present as visual structures only, with no readable characters.
Subject:
A dense field of pollen grains resting on a rough carbon tape surface. The grains vary in shape: round spiny spheres, oval grooved capsules, and a few collapsed shells. Each pollen grain has species-like microarchitecture, including pores, spikes, ridges, perforations, and powdery residue. Some grains overlap or sit in tiny depressions, casting delicate SEM-like shadows.
Important details:
Make the image scientific and plausible, not decorative. False color should be restrained post-processing: warm grains, cool shadows, neutral substrate, and preserved grayscale relief. Keep the instrument layout structured with a bottom strip, scale bar geometry, and micro UI marks that remain unreadable. Surface detail must be crisp at multiple scales: grain silhouette, pore pattern, spines, dust, and substrate granularity. Avoid turning pollen into fruit, beads, or planets.
Use case:
SEM Microscopy single study 2/8, testing false-color pollen surface rendering with structured scientific plate layout.
Constraints:
No readable text, no watermark, no logos, preserve exact 1:1 aspect ratio, no bright candy colors, no macro photo bokeh, no plant petals, no floating grains, no cartoon spikes, no labels, no decorative border, no biological gore.How to use
- Copy the complete original prompt, including any parameters.
- Replace the subject, text and composition with your own requirements.
- Use a compatible external tool; check its current access conditions and review the result.
Before you start
- Check whether a reference image is required. A compatible model label does not mean this example was generated with that model.
- Keep names, figures and visible text accurate; results can vary between model versions.





