Prompting Eyewear and Glasses

Glasses frames seem simple until you try to describe them to an AI model. A standard optical or sun frame involves at minimum: two lens shapes with defined curvature, a bridge connecting them, two temples that hinge and taper toward the ear, two end-pieces connecting the front to the temples, a pair of nose pads (on metal frames), and various screws and hinge hardware. Each of these elements has a distinct material, finish, and reflective character. AI models receiving a prompt like 'tortoiseshell sunglasses on white background' collapse all of this information into an averaged approximation: the acetate becomes a smeared gradient of brown and amber with no real depth, the lenses become flat dark ovals with no convincing transmission or tint, and the temples simply fade to nothing behind the frame front. The fix is to build your prompt in layers. Start with the frame type and shape: 'large rectangular pilot-style frame, slightly wider than tall, with pronounced upper brow line.' Then material: 'rich tortoiseshell acetate front, hand-polished surfaces, genuine depth of layers visible through the material — amber, honey, deep brown, and espresso tones layered beneath a glossy surface finish.' Then the temples: 'matching tortoiseshell acetate temples, straight profile tapering from 6mm at the hinge to 4mm at the tip, clearly defined and sharp-edged throughout their length.' Then lenses: 'gradient brown-tinted mineral glass lenses, dark espresso at the upper quarter fading to honey amber at the midpoint and nearly clear at the lower edge, slight transmission allowing the white background to be faintly visible.' Breaking the frame into discrete zones with specific descriptors for each zone is the single most reliable way to produce convincing eyewear imagery.

The two dominant frame materials behave completely differently under AI generation and require entirely different vocabulary. Acetate (sometimes called zyl or cellulose acetate) is an opaque layered plastic with organic color depth. Good acetate prompt phrases include: 'opaque layered acetate with visible depth of color through the thickness, high-gloss polished outer surface, slight translucency at the thinnest edges catching light, no metallic quality whatsoever.' For specific acetate patterns: 'matte black acetate with a subtle micro-texture surface, flat and even in color without any transparency' or 'crystal-clear havana acetate, amber base with dark brown irregular patterning distributed across the entire frame, some patches denser than others.' Metal frames require the opposite vocabulary set. Whether the material is titanium, stainless steel, monel, or gold-filled, the prompt should emphasize thin profiles, mechanical construction, and specular reflective behavior: 'thin-wire titanium frame, rectangular lens shape, minimal cross-section of approximately 2mm for the top bar and 1.5mm for the bottom rim, mirror-polished gold PVD coating catching a sharp line of specular highlight from the studio light, mechanical hinges visible as small barrel hinges at the frame-temple junction.' Nose pads are a key detail that immediately distinguishes metal from acetate frames: 'clear silicone nose pads on adjustable angled nose pad arms, small oval pad shape, the translucent pad material slightly compressed against the imagined nose of the wearer.' For keyhole bridges on vintage-style frames: 'keyhole-shaped bridge with the distinctive narrowing at the center opening, polished stainless steel construction, the inner curve of the keyhole casting a small shadow onto the nose area.'

Lenses are where most eyewear AI prompts completely break down. A lens is not simply a colored oval — it is a curved optical element that partially transmits the scene behind it, reflects the scene in front of it according to coating type, and refracts light at its edges. For clear optical lenses: 'optically clear CR-39 lenses with anti-reflective coating, lenses are nearly invisible in the frame — just a slight blue-green AR sheen at an oblique angle, the background fully visible through the lens, lens edge showing the characteristic slightly thicker rim of an optical blank.' For polarized sunglasses: 'dark grey polarized lenses with a slight mirror flash, the upper half of the lens showing a subtle surface reflection of the overhead light, the lower portion more transmission-visible allowing a hint of the white surface through the lens, polarization causing a slight darkening of the background reflection.' For mirrored lenses: 'silver flash mirrored lens coating, the entire lens surface acting as a convex mirror reflecting a wide-angle distorted view of the studio environment — softboxes appear as two oval white blobs in the upper portion, the studio ceiling visible as a curved panorama across the mid-lens.' For photochromic lenses: 'photochromic lens showing mid-transition from clear to grey, approximately 40 percent tinted, the tint more uniform gray and evenly distributed across the full lens area.' For colored tints: 'amber tinted lens, the background visible through the lens as warm amber-shifted rather than neutral, the tint consistent across the full lens surface, deepening slightly at the lens edge due to the additional glass thickness there.' Always specify what is visible through the lens — whether that is the white background, the studio environment, or a product surface — because this transmission detail is what makes lenses read as transparent rather than as flat colored panels.

Eyewear product photography has established a set of standard angles that serve different communication purposes, and specifying the angle explicitly in your prompt is essential. The flat-lay hero shot shows the front of the frame: 'glasses laying flat on a white surface, viewed directly from above at a perfect 90-degree overhead angle, temples folded out fully extended behind the frame front, the bridge and entire front frame clearly visible, the open temples forming a slight V-shape behind, even soft studio lighting from overhead, product centered in frame.' The three-quarter angle shows the frame as it would appear on a face: 'glasses photographed at a 45-degree angle from the upper left, showing both the full front of the frame and a foreshortened view of the left temple, slight perspective making the frame appear three-dimensional and wearable, professional product photography.' The profile or side shot emphasizes temple design: 'glasses photographed in strict profile from the right side, the frame front visible as a thin vertical line, the full length of the right temple from hinge to tip completely visible and in sharp focus, temple length approximately 145mm implied by scale.' The front-on display shot simulates the glasses on an invisible face: 'glasses photographed front-on at face height, slightly elevated angle as if looking down at a person of average height wearing them, frame fully symmetrical in composition, both lenses and the bridge clearly centered.' For editorial shots on-face, specify the model and expression distinctly from the frame description to prevent them from merging: 'sunglasses worn by a person with warm medium brown skin and closely cropped natural hair, glasses sitting correctly on the nose bridge with temples hooked over the ears, the frame described separately — separate from the model — as [your frame descriptor].'

Hinges and small hardware elements are where AI generation most frequently produces obviously fake results — repeated patterns, blurry mechanical structures, or hinges that simply do not connect the frame front to the temples in any mechanically plausible way. For a standard barrel hinge: 'barrel hinge at the frame-temple junction, three-barrel construction with alternating male-female barrels, a small stainless steel screw visible through the center of the hinge stack, the hinge mechanism logically connecting the end-piece of the front frame to the temple, when the temple is open at 90 degrees the hinge shows the full barrel structure clearly.' For spring hinges: 'spring hinge mechanism on both temples, the spring cartridge visible as a slightly bulkier section of the temple just behind the hinge, the temple resting in a slightly inward-flexed position indicating spring tension, modern optical eyewear construction detail.' For screwless or hingeless flex frames: 'screwless rimless frame with drill-mounted lenses, the lens edge visible as a clean polished edge with two small holes drilled at the outer corners for the stainless steel mounting hardware, the monofilament lower cable passing through a groove at the bottom of the lens.' For rimless frames: 'fully rimless drill-mount frame, the lenses with no surrounding frame material, held in place only by the bridge hardware at the nose and two small mounting screws at the outer edges connecting to the temples, the lenses floating visually without any enclosing frame element.' Specifying 'mechanically plausible' in your hardware descriptions may seem redundant but it is a reliable cue that reduces the probability of completely non-functional hinge depictions.

An eyewear brand launching a seasonal collection typically needs consistent hero shots across 8 to 20 different colorways and frame styles, all appearing to be shot on the same surface under the same light. Manually prompting each image produces inconsistent scale, lighting character, and background treatment that undermines the coherent collection presentation. In the Floniks workflow editor, you solve this by putting all photographic constants into a shared prefix node: surface description, lighting setup, shadow type, image quality descriptors, and the compositional angle for all frames. Each product node in the workflow then contributes only the frame-specific variables — frame shape, material, color, lens tint, and any distinguishing hardware features. The batch runs all product nodes through the same photographic context, producing a set of images that share the visual identity of a single controlled studio shoot. For colorway variations of the same frame shape, a secondary parameterization node can accept an array of color variables and generate all colorways of a given style in a single workflow execution. This is particularly useful for acetate frames that come in 6 or 8 colorways using identical shapes: the shape and composition prompts stay in the shared node, and only the acetate color descriptor changes per output. The Floniks editor also allows a final finishing node to apply consistent post-processing — uniform shadow intensity, white-balance correction, or a gentle warm grade — across all frames in the collection simultaneously, so the entire set arrives with production-ready consistency without manual color correction on each individual image.