✦ SVG Studio

The LLOS Visual Grammar is a design constitution that governs how icons encode meaning. Unlike traditional icon libraries where icons are illustrations, LLOS icons are semantic operators — they carry meaning through structure, accent placement, and compositional grammar. This ensures that over 36,000 icons share a unified visual language while each remains uniquely identifiable. The grammar defines laws, dialects, part-of-speech visual modes, and semantic species.

A literal icon draws the recognizable form — a tiger looks like a tiger, a car looks like a car. An interpretive icon encodes the concept's deeper essence through visual metaphor — a tiger might become "compressed energy exploding along a parabolic arc." Concept-to-SVG generates 1 literal + N interpretive by default so you always get both: instant recognition and semantic depth.

Each dialect is a visual worldview, not a parameter tweak. Saffron Accent is surgical and restrained, like a sutra carved in sandstone. Dhūp is atmospheric and layered, like verse by candlelight. Yantra is geometrically precise, like a theorem at dawn. Mudra is gestural and organic, like ink on a leaf. The same grammar governs all four — dialects change voice, not rules.

The saffron accent is the meaning token — the single most important visual element in any LLOS icon. Dark strokes provide structure and grammar scaffolding; the saffron accent carries the semantic payload. If you blurred everything else away, the accent alone should still whisper the concept. It is never decorative — every saffron element must carry meaning.

You need an OpenAI API key with billing enabled (paid tier). The Concept-to-SVG workflow costs approximately $0.05 per 4 icons using GPT-4o. Image-to-SVG is slightly more due to vision API calls. Always use a paid-tier key — the free tier permits OpenAI to train on your data, which is an intellectual property risk for original icon designs.

Silhouettes are shape-faithful — they capture the actual outline and proportions of the object in your uploaded image, like a high-quality shadow cutout or technical contour. Interpretive icons extract the semantic meaning from the image and create abstract metaphorical representations. You get 2 silhouettes for recognition plus your chosen number of interpretive variants for depth.

Yes. Every generated icon has "Copy SVG" and "Export JSON" buttons. The JSON contains the concept name, SVG code, dialect, part of speech, semantic species, and timestamp — structured to feed directly into the Selection Workshop for curation. You can also "Export All JSON" to copy the entire batch at once from the results toolbar.

Three elements, always. Scene: one clear visual metaphor, fully committed to — not a collage, not a dot-and-line. Feel: emotional elegance through gradients, proportion, and motion — the icon should evoke feeling before recognition. Identity: semantic distinctiveness from neighboring concepts — "Loneliness" must be unmistakably itself, not generic "alone." An icon missing any one element is incomplete.