A logo is invented. A crest is researched. The distinction matters more than it sounds. A logo is made for the brand; a crest is made for the lineage. A logo says 'we are Acme.' A crest says 'this is what our family has always meant.' The wine label sits in the second category, not the first, and that is why the studio's crest pass starts where it does.
The surname is the seed
Type a surname into the studio and the heraldic-research engine, a small language model pass tuned on the heraldry corpus, returns the elements traditionally associated with that name's origin: the animal, the motif, the secondary charge, the tincture, the shield shape. The Andersons of Norman descent get a different starting set than the Andersons of Scottish-Highland descent than the Andersons who arrived through Ellis Island in 1907. The research pass is opinionated about which Anderson it thinks is yours; you can correct it.
What it returns is not the final crest. It is the brief. The actual design, the proportions of the shield, the disposition of the elements, the foil treatment of the linework, the deboss depth, happens in the next pass, the rendering one. The brief constrains the rendering enough that the result reads as authentically yours rather than as one of ten thousand generic European armorials.
The palette restraint
The traditional heraldic palette includes gules (red), azure (blue), vert (green), purpure (purple), sable (black), or (gold), and argent (silver). The studio's rendering pass restricts itself, in practice, to two: antique gold and muted plum, with cream as the field. The restraint is not a heraldic principle; it is a label-design principle. A four-color crest on a wine label reads as a sticker. A two-color crest reads as foil.
If the heraldic-research pass returns an azure field for a family of Scottish origin, the rendering substitutes muted plum, with a note that explains the substitution. We have not had a customer object yet. The bottle reads as a wine bottle rather than as a herald's paint plate, which is the thing the customer is actually buying.
What to look for when the bottle paints
Three things, in order. First: does the crest sit centered in the gold medallion at the top of the label, with even visual weight on the two sides? An off-balance crest reads as off-balance from across the room, even when the off-balance is two pixels. Second: does the linework have apparent foil depth, bright on one side, shadow on the other, rather than reading as flat gold paint? The foil character is what separates the heirloom-grade crest from the souvenir-grade one. Third: does the family name on the banner along the bottom read in a font that matches the bottle's other foil text? A misaligned typeface, Trajan above, sans-serif below, kills the heraldic illusion in one beat.
When all three of those land, the bottle is ready. The crest reads as if it has always been the family's, even when the family adopted it three minutes ago. That is the point. The crest is not a discovery. The crest is a decision, made well enough to look like a discovery.
