Sideline Captions sends a preview of each frame to its AI service. GalleryID Desktop identifies the athlete on your own machine, on Mac or Windows, with nothing uploaded and no connection required. Facial recognition and jersey numbers, both local.
Sideline is straight about this in their own FAQ: “Sideline sends a small JPEG preview (~0.25MB) to our AI service for captioning — your full-resolution files stay on your machine.” Fair disclosure, and the full-size files really do stay put. A preview of the frame still travels.
GalleryID Desktop doesn’t send anything. The models sit on your drive, the identification runs on your processor, and the names are written into IPTC and XMP right where the file already is. No preview, no queue, no service on the other end of it.
That matters in two places you actually work. Press box wifi that’s fighting eight thousand phones, where a cloud round trip per frame is the bottleneck. And the athletic department whose counsel has an opinion about images of athletes leaving the building at all.
Identification runs locally on macOS 12 and later or Windows 10 and 11. No preview leaves the machine, and we store nothing on our servers because nothing gets to them.
A connection is for downloading the app and your rosters. The identification pass itself doesn’t need one, so a dead uplink costs you delivery time, not tagging time.
Sideline’s download page lists Windows as coming late 2026, and their Mac build needs macOS 15. Desktop has shipped on Windows 10 and 11 since launch and runs back to macOS 12, so the machine you already own is likely enough.
Neither app charges you per image. The difference is what unmetered costs to run: ours is your own hardware, so there’s no service behind it to price later.
Sideline’s published identification methods are jersey numbers, team color, duplicate-number handling and your in-camera voice notes. Facial recognition isn’t among them. The voice note is a genuinely good idea and it does cover a hidden number, but only for the frames you thought to talk over while you were working. The celebration you didn’t call, the bench reaction, the number swallowed by an arm: those come back to you later with nothing attached.
Desktop reads the number too. It also reads the face. MatchID was trained on a large archive of sports photography rather than ordinary photos, so it holds up through motion blur, distance, turned heads and helmets, and it falls back to the jersey when there’s no face to read.
We’re not going to tell you it gets there on the first pass. Some frames need a person to look at them. Those land in a review queue, where one press from the roster finishes the job. Every name is confirmed. Never guessed.
You know the moment. The number is hidden, the player could be one of two who look alike from across the field, and you’re scrolling back through the take to narrow it down.
Click Who Is This, and MatchID checks that one photo against the headshots scoped to the folder and hands you its best match. You confirm and keep moving. No batch to run, no settings to open, and nothing sent anywhere to answer it.
Checked against each vendor’s published pricing and documentation.
| Capability | GalleryID DesktopMac & Windows | Sideline CaptionsMac only | Caption ProMac & Windows |
|---|---|---|---|
| Identification runs on your own machineNothing leaves the building | Nothing is uploaded | Sends a JPEG preview to their AI service | —Not published |
| Works with no internet connection | Captioning needs the connection | —Not published | |
| Runs on Windows | Windows 10 and 11 | Their download page says late 2026 | |
| Minimum macOSThe Mac you already own has to be new enough | macOS 12Monterey, 2021 | macOS 15Sequoia, 2024 | —Not published |
| Facial recognition | Trained on sports photos | Jersey numbers, team color, voice notes | Celebrity database, plus your own |
| Reads jersey numbersFor helmets, and faces turned away | |||
| Names the athlete when the number is hiddenBack turned, number blocked, no digits in frame | The face is the second route | If you recorded a voice note at capture | If the face is in their database or yours |
| Ask about a single photoIdentify one frame against the folder’s headshots | Who Is This | ||
| Tag from the roster in one pressStarting from blank, not correcting a draft | Name + caption line together | Correcting AI output, not tag-from-blank | |
| Caption line tied to the playerThe long form belongs to the person, not the field | One press writes both | ||
| Roster follows the folderNo swapping between events | Subfolders inherit | ||
| Rosters built in | 1,295 teams, 348,656 athletes | 1.6M+ athletes, 340+ leagues | Lists you build |
| College sports past the revenue sportsWrestling, field hockey, rowing, track, and the rest | 90 sponsored sports across D2 and D3 | 16 sports, across more levels of play | |
| Bring your own roster, any common format | Or build one in the app | CSV or a code replacement file | |
| Wire-service caption presetsAP, Getty and the agency house styles | Your own templates: app, org, folder | AP, Getty, Imagn, Icon, ZUMA | —Not published |
| The codes you already built keep workingYour code replacement file, typed the same way | Nothing was taken away | Reads a code file as roster data | |
| Ingest, cull, ratings, color labelsRAW browses from embedded previews | Measured: 4,664 files, 52 GB, in 1:19 | Grid and lightbox, star filters | Watch-folder ingest only |
| Crop and tone adjustments | Crop, straighten, histogram | ||
| Offline catalog of your drivesBrowse and search with the drives unplugged | Catalog, plus iPhone and Android companions | ||
| FTP delivery from the export step | FTP and FTPS, with watermarking | FTP, sFTP and Dropbox | |
| Unmetered, no AI credits | Runs on your machine | Unlimited on paid plans | Metered at every tier below Enterprise |
| Price | $20/mo+$10 per extra device | $14.99/moor $249 lifetime | from £7.99/moto £600/mo for teams |
Verified August 2026. If something here is out of date, tell us and we will correct it.
Four places they’re ahead of us today. You should know them before you pick.
Sideline publishes 1.6M+ athletes across 16 sports and 340+ leagues, spanning levels of play our library doesn’t reach. Ours is 1,295 teams and 348,656 athletes: NCAA Division I for FBS football and men’s and women’s basketball, Division II and III across 90 sponsored sports, and ten pro leagues. Deeper inside college, narrower outside it. If your season sits outside the NCAA and the US pro leagues, their library does work ours leaves to you.
Sideline ships named presets for AP, Getty, Imagn, Icon Sportswire and ZUMA Press, so the format comes off a menu. Desktop expects you to build the template once, at the app, org or folder level, and from then on CaptionID writes the athlete name and the full caption line from the bound roster in one press. Same filed caption, different setup cost. If you file to several houses in a week and would rather configure none of them, that menu is worth something.
Sideline sells a $249 lifetime license, and their pricing page offers lifetime updates with it rather than the usual year-and-then-upgrade ladder. Read the one condition attached: that page also says lifetime licenseholders bring their own API key, so the AI service behind the captioning is yours to arrange and yours to pay for. We don’t sell a one-time license at all. $200 a year, always current, nothing to arrange. That’s a position, and you’re allowed to disagree with it.
Their subscription bundles wireless tethering, remote FTP auto-download and live FTP galleries alongside captioning. Desktop is the local half of our answer and GalleryID Cloud is the hosted half, sold separately. If one bill for both halves is the requirement, count that in their column.
Identification that never leaves your machine and needs no connection. Windows, today. A face as the second route when the jersey number is hidden. Who Is This on a single frame. Rosters bound to the folder so events don’t mean file swapping. A library that goes 90 sponsored sports deep across Division II and III, not just the revenue sports. An offline Catalog of your drives with iPhone and Android companions. And the code replacement values you built yourself, still expanding as you type, untouched.
Different question, separate page: GalleryID Desktop vs Photo Mechanic and Lightroom Classic.
Names in the file.
Nothing in the cloud.
14 days free on Mac or Windows.
$20/month for the first device, $10 for each one after.
Sideline Captions and Caption Pro are trademarks of their respective owners. GalleryID is not affiliated with, sponsored by, or endorsed by either of them, and the names are used here only to identify the products being compared. Feature details and prices reflect each vendor’s published documentation as of August 2026 and may have changed since.