PullCue's cover-photo feature can read visible text and propose likely series-title and issue-number hints. It can help you begin a catalog search without typing every word. It does not compare cover artwork, recognize an exact cover variant, prove a printing, or identify grade, condition, authenticity, rarity, or value. Confirm the physical comic and the catalog candidate before relying on a duplicate warning or buy decision.

That distinction is the whole feature boundary: text hint first, candidate search second, collector confirmation third. A clear photo can shorten identification, but it cannot turn incomplete evidence into an exact edition.

What does PullCue actually do with a cover photo?

In the current PWA beta, the PullCue cover-photo tool follows a bounded process:

  1. You take a photo or choose an existing JPEG, PNG, or WebP image.
  2. PullCue prepares a bounded temporary JPEG copy in the browser, resizing it when needed.
  3. An on-device text reader extracts printed words and number-like text.
  4. PullCue proposes up to five possible title hints, with an issue hint when one is readable.
  5. You choose or correct a hint, which opens the normal title-and-issue search.
  6. You review the returned candidates before requesting a recommendation.

PullCue uses the browser version of Tesseract for the text-reading step. The pinned Tesseract.js 7.0.0 documentation describes it as a JavaScript library that extracts words from images and runs in a browser. That is optical character recognition, or OCR. OCR answers “what text may be visible here?” It does not answer “which cataloged cover has matching artwork?”

Candidate retrieval happens only after a title or title-and-issue hint is sent into search. The cover reader itself does not silently select a comic or produce a buy decision.

Which cover details can become useful hints?

The reader is most useful when the cover shows a large, readable series logo and an explicit issue number. It may extract:

  • a prominent series title;
  • a number marked with #, No., or Issue;
  • an issue-like number printed by itself;
  • a limited-series label such as 2 of 5; or
  • several adjacent title lines that appear to belong together.

These are possible search terms, not verified metadata. A large event name, character name, creator credit, publisher logo, price, month, or promotional phrase can compete with the actual series title. A number on the cover might be an issue number, a price, an anniversary mark, or part of the artwork.

The label Strong Text Match means the title-like text crossed PullCue's strongest rule threshold. It is not a probability, a comparison with every other hint, a catalog match, Data Confidence, Collection Fit, or a Should Buy decision. Even the strongest text hint remains editable and requires confirmation.

Why can a clear-looking cover still produce the wrong text?

Comic covers are designed as art and advertising, not as clean document scans. Logos can be stylized, tilted, partly hidden, or printed over detailed artwork. Foil, glare, bag reflections, dark backgrounds, small trade dress, creases, price stickers, and perspective can make letters or digits difficult to separate.

The official Tesseract image-quality guidance notes that skew, noise, uneven backgrounds, resolution, borders, and page segmentation can materially affect OCR output. For a cover photo, that means a straight-on, well-lit image with the full title area visible is a better input than an angled, reflective, tightly cropped, or distant photo. It still does not guarantee correct text.

If the first result is poor:

  1. Move out of direct glare and hold the comic straight.
  2. Keep the full title area and any visible issue number in frame.
  3. Retake the photo rather than accepting a familiar-looking wrong hint.
  4. Use title-and-issue search manually when the printed logo remains unreadable.

Do not photograph only the artwork and expect the current feature to find a visual match. Image similarity is not part of this release.

What must I confirm before choosing a catalog candidate?

Confirm these details against the comic in your hand:

  1. Formal series title. Cover text can emphasize an event banner or character instead of the title recorded in the indicia.
  2. Issue number. Check that the number belongs to the issue rather than a price, year, cover count, or promotional mark.
  3. Series run. When titles restart, compare publisher, series start year, volume, and publication details. The series-run identification guide explains why the same title and issue number can describe different comics.
  4. Cover or variant designation. The current photo tool always leaves exact-variant confidence unknown.
  5. Printing. A later printing can reuse much of the first printing's design while remaining a different edition.
  6. Full barcode and supplement, when present. Use them as additional evidence only when the mapping is reliable.

Those checks matter because PullCue compares the selected identity with your collection. The duplicate-or-variant guide shows why issue-level ownership and exact-edition ownership are not always the same fact. A cover-text hint cannot safely support “you already own this exact variant.”

What happens when cover and barcode evidence disagree?

If the flow also has a comic barcode supplement, PullCue may use the supplement's issue-number pattern as another search hint. When that number conflicts with the number read from the cover, the app shows an evidence warning and preserves both possibilities for review.

Neither source should silently win. The photo may have captured an unrelated number, the barcode may be partial, or the catalog mapping may be unavailable. Compare the full code, the cover, and the indicia before searching again. A conflict is a reason to reduce confidence, not a reason to average two numbers into a new answer.

The data-sources and safeguards page keeps active, cached, uncertain, and rights-limited evidence separate. A returned candidate does not prove that every possible issue or edition was searched, and an empty result does not prove the comic does not exist.

Is the photo uploaded or saved?

In the current PWA beta, PullCue resizes and reads the cover image on the device. The photo is not uploaded, sent to analytics, or added to the collection, and the temporary image reference is released when reading ends. The public OCR runtime files—not the cover pixels—may be cached by the browser so the reader can be reused while those files remain available.

There is one important boundary: when you choose a text hint, its title and issue text can enter PullCue's normal catalog-search flow. “The photo stays on the device” does not mean a later search makes no network request.

First-time OCR setup generally needs a connection to load and verify the reader files. Later text reading may work offline if the browser retained them, but browsers can evict site storage. A fresh catalog candidate also needs either a network response or relevant cached metadata. Local ownership evidence can remain useful even when those external steps are unavailable.

When should I use Wrong Comic?

Use Wrong Comic? as soon as the recommendation screen shows the wrong title, issue, series run, cover, or printing. Do not continue because the collection-fit reasons look plausible. Those reasons apply to the selected candidate, not automatically to the physical comic.

Return to search and correct the identity. If the right catalog candidate is unavailable, preserve the title and issue you can verify and use a limited-confidence local record instead of choosing a false match. The broader comic-identification guide explains the safe fallback from photo to search, barcode evidence, or entered details.

How should cover-photo evidence affect Data Confidence?

Cover OCR is useful evidence about possible text, but it is not exact-edition evidence. A recommendation should remain cautious when the run, cover, variant, or printing could change the ownership result.

The Limited Data Confidence guide separates evidence quality from Collection Fit and Buy Decision. A cheap asking price does not make the OCR more accurate. A strong collection connection does not resolve a different printing. Conversely, a collector-confirmed local exact duplicate can remain decisive even if story metadata is missing, because the ownership fact does not depend on the cover reader.

Can PullCue identify the exact cover variant from my photo?

No. The current PWA reads text; it does not compare artwork or claim exact-cover recognition. Confirm the cover designation and printing yourself.

Can it identify a comic when no issue number is visible?

It may still offer a title hint. You must enter or confirm the issue number before a trustworthy issue-level search and recommendation.

Can it grade the comic or tell me what it is worth?

No. Cover-text recognition does not evaluate condition, interior completeness, restoration, authenticity, market value, or future price.

Is native iPhone or Android cover recognition included?

Not as verified current behavior. Cover-text reading is available in the PWA; current native builds direct collectors to title-and-issue search until a separate privacy-reviewed native recognition adapter is implemented and tested on physical devices.

The reliable rule is simple: use the photo to reduce typing, use search to find candidates, and use the physical comic to confirm identity. Only then should duplicate protection, collection fit, and asking price decide whether the comic makes sense for you.