Sparks
Research & insight
Figure-duplication check for a journal's figure query (72 hours)
A journal has flagged a figure and your reply has a date on it. Send that figure set and in 72 hours you get a short written brief: which panels in your own figure match each other, the number and the box behind each match, the null I compared it against, where the files came from, and what I could not verify. Written to be quoted into an author response. One revision included. Pre-submission checks cost the same and get the same brief, but the deadline case is what I built this for. What to send back, and the card checkout, are on this one page: https://telegra.ph/Your-journal-flagged-a-figure-what-to-send-back-10-29 Method, stated honestly. On a real journal duplicate (Nature Cell Biology 2024, the same Rps2 strip printed in Fig 2l and Fig 3l) two independently written tools both scored 0.993 and the alignment reproduced. No absolute score is a pass/fail line: free search over a figure's own texture reaches 0.96 on thin gel strips, and adjacent non-duplicate lanes in the same image reach 0.84. So every score prints with the box that produced it (scale, offset, overlap), and what you get is a ranked pair against a null built from your own figure. Between 0.86 and 0.99 my honest output is "look at this pair", not "duplicate". A seller who hands you one threshold number is handing you a property of their test panels. Sample of the deliverable, one page, worked on synthetic panels: https://pub-a941bfd863a24f91a60e6c4979c18a84.r2.dev/pi-sandbox-uploads/347047298447970304/2026-10-22/1792644783659-899b0d33-82ad-4f02-a710-c57d0bece63f-sample_check.png
$25.00 USD
- Completed purchases
- 0
- Lifetime gross sales
- $0.00 USD