HandiProducts Photo Tagger — Tag System FAQ

No. Brand is already encoded in the tag prefix — the two-letter namespace before the colon.

Every ht: tag is a HandiTreads product. Every hr: tag is HandiRamp. You can filter the entire HandiTreads catalog just by searching for ht: — no extra tag needed.

Brand names do belong in the Description column of the worksheet — that's what the AI reads to understand context. For example: "HandiTreads-branded non-slip tread — same product as hr:stair-tread but sold under the HandiTreads brand." This helps the AI distinguish between identical-looking products from different brands.

These are two distinct dimensions and they're handled differently:

1. Star rating (1–5) → native XMP:Rating field
Star ratings are a universal XMP standard, not a tag. The tagger writes them directly into the image file via exiftool alongside the product tags. Every DAM tool — Lightroom, Bridge, even macOS Finder — reads and writes XMP:Rating natively. No scene: tag needed.

2. Usability / quality → scene: tags
Image quality is a judgment the AI makes visually — blur, poor exposure, bad cropping, compression artifacts. Standard star ratings don't capture this well. Two tags cover it:

The useful searches are "give me only the marketing-ready ones" or "flag the bad ones for culling." A middle tier isn't worth the noise.

Where to edit ratings:

• Lightroom / Bridge — best for bulk culling and final editorial decisions (keyboard shortcuts, side-by-side view)
• Tag review UI — shows the AI's proposed rating so you can correct obvious errors before accepting a batch; not a full rating workflow
• macOS Finder — quick one-off edits via Get Info or column view

Practical flow:

1. AI writes initial XMP:Rating + tags during processing
2. Tag review UI shows the proposed rating — correct it if obviously wrong
3. Final editorial rating pass happens in Lightroom/Bridge where you can see images full-size

Weather and environmental context are a different dimension than product type — they describe the scene, not the product. We handle these with a separate scene: prefix.

For example, the AI might want to describe a stair tread in snowy conditions as "snow safety" — accurate, but not searchable as a tag. Instead, two clean tags cover it:

Both are independently searchable. You can find every snow-condition photo across all product lines just by searching scene:snow.

Current scene: weather tags

Current scene: setting tags

Current scene: usage context tags

Historical photos are valuable for reference but need to be findable and distinguishable from current product shots. We use two scene: era tags:

Why not decade tags (like scene:decade-1990s)? The AI can detect that a photo looks old — film grain, color cast, image quality, dated product styling — but it can't reliably pin the decade. A 1994 and a 2001 photo look nearly identical. Decade-level tagging creates false precision in the data.

The practical search need is binary: "old photos for historical context" vs. "current photos for marketing." Two tags cover that cleanly.

Color is handled as part of the variant tag — but only where color is a real product differentiator (i.e. a separate SKU that customers choose). The AI is reliable at distinguishing colors visually, so this works well in practice.

HandiTreads come in 4 colors, each a real purchase decision. Color is encoded directly in the tag alongside size:

The four HandiTreads colors and their visual character:

• Chestnut Brown — warm medium brown
• Java Brown — dark espresso brown
• New England Gray — medium cool gray
• Obsidian Black — deep matte black

This applies to all three HandiTreads product types: stair treads (30"/36"/48"), ramp & deck treads (48" only), and nosings (30"/36"/48").

PetStep comes in two colors across both the folding ramp and pool ramp:

For pool ramps, color and leg length combine into the variant tag (standard 23" legs vs. long 30" legs × Graphite/Khaki = 4 combinations).

HandiRamp van ramps and modular ramps are always bare aluminum — no color variants, no color tag needed.

What about generic color as a scene tag? We don't use scene:color-* tags. Scene color (a gray deck, brown stairs) would create false matches with product color. Color belongs in the variant tag, not the scene namespace.