Listing-ready trading card images from a single phone photo
CardMagic finds every trading card in the picture, then crops, straightens, flattens, and color-corrects each one. No scanner. No toploaders. No custom imaging hardware. Just the phone in your pocket and the card in your hand.
Ditch the expensive rig. Sell more cards in the same afternoon, or build CardMagic into your app and hand that speed to every one of your users.
2026 Topps 75th Anniversary, 1991 design
2026 Topps Series 1
- Cropped
- Deskewed
- Dewarped
- Auto-leveled
Tap or click any image to see it larger.
Four fixes, every trading card, one call
A phone photo is close. CardMagic closes the gap between the picture you took and the image a buyer trusts. The two cards above are the actual files CardMagic returned for that photo, with no manual editing.
Crop
Finds every trading card in the frame and trims each one to its own edges. Two cards in the photo means two files out. The cloth, the table, and your thumb are gone. Crop mode stops here and hands you the card exactly as photographed.
Deskew
Nobody lays a card down perfectly square. CardMagic rotates each crop so the borders run true, the way a buyer expects to see them.
Dewarp
A phone held at an angle turns a rectangle into a trapezoid. Dewarping corrects the perspective and squares the card up to standard 5:7 trading-card proportions, flat, as if it had been scanned.
Auto-levels
Restores contrast and removes the color cast of indoor light so white borders read white and the photo matches the card in your hand. It applies to the whole image evenly, like fixing white balance, and you can turn it off.
Show buyers every corner of the trading card
Corners are the first thing a buyer zooms in on, and the first place condition shows. Ask CardMagic for corner close-ups and every card comes back with four magnified corners and a sheet that lays them out as they sit on the card, ready to drop into a listing as a single photo.
It is one option on the same call. Any app built on CardMagic can hand its users condition shots without building a thing.
Tap or click either image to see it at full size.
Four corners and one sheet
Each corner comes back as its own 600 by 600 close-up, plus a single sheet with all four laid out as they sit on the card. Drop the sheet into a listing as one photo.
Right to the edge, and past it
Every close-up shows 14 mm of card in from the corner and 2.5 mm beyond it, so rounding, whitening, and dings at the very tip are plain to see.
A 1 mm grid, and no labels
A light grid, heavier every 5 mm, lets a buyer size a flaw at a glance without numbers printed over the card.
Sharp, square, and true to color
Taken straight from your original photo and squared to the card, so corners stay as sharp as the photo allows even when the card image itself is small. Auto-levels is never applied, so the color is the photo's own.
Build it into your app
CardMagic is one endpoint on the API you already use, and it turns the phone camera your users already carry into an imaging station inside your product. Whatever they do with trading cards, the listing image stops being their problem.
One photo, three answers
Send the same photo to identification, price data, and CardMagic. Your user gets what the card is, what it sells for, and the image to list it with. The whole listing from a single snap, and none of it built by you.
What it is
Card identificationWhat it sells for
Price dataThe image to list it with
CardMagic
Listing creation in your marketplace
A seller snaps one photo in your app. You return the identification, the comps, and the listing image, and the listing writes itself. Sellers finish more listings, your catalog looks consistent, and nobody had to learn photography.
For marketplacesInventory for card shops
Point-of-sale and shop software: lay out the new arrivals, take one photo per batch, and every card lands in inventory with a clean image attached. Intake that used to eat an evening takes a coffee break.
For sellersIntake after a card show
Shops and collectors come home from a show with a box of pickups. Photograph them on the counter or the hotel bed, and every card is imaged and ready to catalog before the box is unpacked.
For collectorsCollection and portfolio apps
Binder-quality images from a phone snap. Users see their collection the way it looks in hand, and the same image is ready the day they decide to sell.
Collection managementBreaks, consignments, and bulk submissions
Image the hits the moment they are pulled, or let consignors send a phone photo instead of a scan. Every submission arrives with uniform images and no scanner in the loop.
AI agents that list for you
An agent takes the photo, identifies the card, pulls the comps, and drafts the listing with the CardMagic image attached. One more tool in the flow, nothing for the user to touch up.
For AI agentsGet the best results from CardMagic
CardMagic is forgiving, but a few habits when you take the photo turn a good listing image into a great one.
Fewer cards per photo means more detail
CardMagic handles up to a full 9-card binder page in one photo, and every card it finds comes back as its own file. The trade-off is pixels. A 12-megapixel photo of one card spends all twelve million on that card; the same photo of a full page splits them nine ways. Shoot a page when speed matters, and shoot one or two cards at a time when the listing deserves every bit of detail.
Pick a background that contrasts with the border
CardMagic runs on the same AI as CardSight AI's card identification, so it finds cards in any real-world setting: a kitchen table, a card show, a storage unit. The background matters for your buyer, not the AI. The finished image keeps a small margin of the real surface around the card, and when that margin contrasts with the border, the edges and corners read clearly and the buyer can judge condition for themselves. Put a white-bordered card on a darker surface and a black-bordered card on a lighter one.
Leave a little room between cards
When you lay out several cards, keep a small gap between them and keep every edge inside the frame. Each card comes back with a slim margin of the real surface around it, so a neighbor pressed against the edge ends up in that margin, and an edge that runs off the photo cannot be recovered. A finger's width of space is enough for every card to come out clean, with all four edges visible to the buyer.
The old ways cost you time or money. Usually both.
We have fed stacks of cards through a sheet-fed scanner one toploader at a time, and we have priced the purpose-built rigs. CardMagic exists because neither should stand between a raw trading card and a listing.
| Imaging | CardMagic The phone you already own | Sheet-fed scanner | Custom imaging hardware |
|---|---|---|---|
| What you buy | Nothing new. The phone in your pocket and an API call | A sheet-fed scanner, plus a sleeve and toploader for every card that goes through it | A dedicated device, a subscription, and the vendor's software, plus the bench space to keep it all set up |
| Per card | Lay it down, take the picture | Sleeve it, topload it, feed it through, pull it back out | Position, capture, repeat, inside the vendor's software |
| Cards per capture | Every card in the photo | One at a time | Depends on the device |
| Where it works | Card show, kitchen table, storage unit | At the desk the scanner lives on | Wherever the device and its software are set up |
| What the buyer sees | The card as it looks in hand, framed and leveled | A scan taken through a toploader's plastic. Chrome and refractors lose their shine | Good images, at hardware-plus-subscription prices |
| Condition detail | Close-ups of all four corners on a 1 mm grid | Whatever the scan resolution shows, through the toploader's plastic | Depends on the device |
| Growing to volume | More calls, no more gear | Another scanner per station | Another unit per station, and the subscription keeps running |
What you buy
CardMagic
Nothing new. The phone in your pocket and an API call
The old ways
Sheet-fed scanner: A sheet-fed scanner, plus a sleeve and toploader for every card that goes through it
Custom imaging hardware: A dedicated device, a subscription, and the vendor's software, plus the bench space to keep it all set up
Per card
CardMagic
Lay it down, take the picture
The old ways
Sheet-fed scanner: Sleeve it, topload it, feed it through, pull it back out
Custom imaging hardware: Position, capture, repeat, inside the vendor's software
Cards per capture
CardMagic
Every card in the photo
The old ways
Sheet-fed scanner: One at a time
Custom imaging hardware: Depends on the device
Where it works
CardMagic
Card show, kitchen table, storage unit
The old ways
Sheet-fed scanner: At the desk the scanner lives on
Custom imaging hardware: Wherever the device and its software are set up
What the buyer sees
CardMagic
The card as it looks in hand, framed and leveled
The old ways
Sheet-fed scanner: A scan taken through a toploader's plastic. Chrome and refractors lose their shine
Custom imaging hardware: Good images, at hardware-plus-subscription prices
Condition detail
CardMagic
Close-ups of all four corners on a 1 mm grid
The old ways
Sheet-fed scanner: Whatever the scan resolution shows, through the toploader's plastic
Custom imaging hardware: Depends on the device
Growing to volume
CardMagic
More calls, no more gear
The old ways
Sheet-fed scanner: Another scanner per station
Custom imaging hardware: Another unit per station, and the subscription keeps running
Sell trading cards faster on every marketplace
A CardMagic image is a standard JPEG, or PNG if you prefer, sized for the zoom every marketplace offers. It works wherever you list.
When imaging drops from minutes per card to seconds, the math changes. The two-dollar card is worth listing. The 500-card lot gets listed this week instead of next quarter. Volume sellers move more inventory with the same hands.
- eBay
- Fanatics Collect
- CollX
- StockX
- Your own storefront
A day of listings, two ways
- Scanner
- Sleeve it, topload it, feed it through. Then take the card back out of the toploader so the next one can go in. Every card handled twice, and the buyer sees it through plastic.
- CardMagic
- Lay out the cards, take the picture, send it. Every card comes back framed and leveled. The time you save is the time you spend listing.
Send the photo. Get the cards back.
One request with the photo attached, as a multipart upload or the raw image bytes. One card comes back as the image itself. Two or more come back as a ZIP with one file per card in reading order, each already cropped, straightened, flattened, and leveled.
The defaults are the listing recipe: squared up to standard 5:7 card proportions with a small margin and auto-levels on. Ask for crop mode to keep the card exactly as photographed, set your own margin and fill color, or turn auto-levels off. Add corner close-ups and each card is followed in the ZIP by its four corners and a corner sheet. The same API key you use for identification works here, and the OpenAPI spec is generated from the live API.
import { writeFile } from 'node:fs/promises';
const form = new FormData();
form.append('image', photoFile); // the phone photo, as taken
// Defaults are the listing recipe: straightened, squared
// to 5:7, 5% margin, auto-levels on. ?mode=crop keeps the
// card exactly as photographed. ?corners=true adds four
// corner close-ups and a corner sheet per card, and the
// reply is then always a ZIP.
const response = await fetch(
'https://api.cardsight.ai/v1/cardmagic/process',
{
method: 'POST',
headers: { 'X-API-Key': 'your_api_key' },
body: form,
},
);
// One card comes back as the JPEG itself. Two or more come
// back as a ZIP of card_0.jpg, card_1.jpg, ... in reading order.
const count = response.headers.get('X-CardMagic-Count');
const type = response.headers.get('Content-Type');
const isZip = type === 'application/zip';
const bytes = Buffer.from(await response.arrayBuffer());
await writeFile(isZip ? 'cards.zip' : 'card_0.jpg', bytes);
console.log(`${count} card(s) ready to list`);Explore the platform
CardMagic makes the image. The rest of CardSight AI tells you what the card is, what it sells for, and where it sits in a collection.
Card Identification
Snap a photo, get the card. 99.5% accuracy across 8,250+ sets.
Slab Identification
Grading company, grade, and qualifiers read straight from the label.
Price Data
Real marketplace sales and live listings, raw and graded.
Collection Management
Collections, binders, and want lists on top of the full catalog.
Frequently asked questions
What trading card sellers and developers ask about CardMagic, hardware, scanners, and marketplaces.
No. CardMagic works on the photo, not the camera. Any phone from the last several years captures far more resolution than a marketplace listing needs. There is no device to buy, no hardware subscription, no separate software to learn, no stand to set up, and nothing extra to carry to a card show.
Yes, for listing images. A sheet-fed scanner needs each card sleeved and toploaded so it survives the trip through the rollers, run through one at a time, then pulled back out of the toploader afterward. CardMagic takes a photo of the raw card on any surface and returns it framed. The buyer gets a better picture too: the card itself, not a scan taken through a toploader's plastic.
Every card comes back as its own image, together in one ZIP, named in reading order from top to bottom and left to right. The sample on this page is one photo of two cards, and CardMagic returned two files. Lay out a lot, take one picture, and each card is ready to list.
CardMagic handles up to a full 9-card binder page in one photo, and every card it finds comes back as its own file. What changes is detail: a 12-megapixel photo of one card gives that card all twelve million pixels, while the same photo of a full page splits them nine ways. Shoot a page when speed matters, and one or two cards at a time when the listing deserves the crispest image.
No. CardMagic runs on the same AI as CardSight AI's card identification, so it works in any real-world condition. The background is for your buyer, not the AI: the finished image keeps a small margin of the real surface around the card, and when that margin contrasts with the border, the edges and corners stand out so the buyer can see the condition for themselves. Use a darker surface for a white-bordered card and a lighter one for a black-bordered card.
Yes. Turn on corner close-ups and every card comes back with four 600 by 600 close-ups of its corners, plus one sheet that lays all four out as they sit on the card. Each close-up shows 14 mm of card from the corner and 2.5 mm beyond it on a light 1 mm grid, so rounding, whitening, and dings are easy to judge. The close-ups keep the photo's own color, since auto-levels is not applied to them, and the grid is scaled for a standard 2.5 by 3.5 inch card.
It changes the photo, not the card's condition. Cropping, deskewing, and dewarping fix framing and perspective, and auto-levels adjusts contrast and color cast across the whole image so white borders read white. Every correction applies to the whole image evenly and nothing is painted over, so corners, edges, and surface flaws stay exactly as captured. That keeps your listing honest and your buyer confident. If the photo's own color is the point, turn auto-levels off.
Any of them. The output is a standard JPEG, or PNG if you ask for it. The two sample crops on this page came out of a 12-megapixel phone photo at roughly 1400 by 2000 pixels each, which is large enough for zoom on eBay, Fanatics Collect, CollX, StockX, and your own storefront. Nothing is watermarked or locked to a platform.
Yes. Every CardSight AI plan includes every feature, and CardMagic follows the same rule, starting with the free tier. Pick the plan that matches your monthly volume and CardMagic is already there.
Yes, and that is where it pays off most. CardMagic is one endpoint on the same API as card identification and price data, with the same key and the same SDKs, so a marketplace, shop system, collection app, or AI agent can offer listing-ready images from a phone photo without building any imaging of its own. Most integrations are a day of work: send the photo, save the files that come back.
Yes. Send the same photo to card identification and to CardMagic. One capture gives you the player, set, year, card number, and market comps, plus the framed image to list it with. Together they turn a stack of raw cards into finished listings in one pass.
Tap any question to expand
No. CardMagic works on the photo, not the camera. Any phone from the last several years captures far more resolution than a marketplace listing needs. There is no device to buy, no hardware subscription, no separate software to learn, no stand to set up, and nothing extra to carry to a card show.
Yes, for listing images. A sheet-fed scanner needs each card sleeved and toploaded so it survives the trip through the rollers, run through one at a time, then pulled back out of the toploader afterward. CardMagic takes a photo of the raw card on any surface and returns it framed. The buyer gets a better picture too: the card itself, not a scan taken through a toploader's plastic.
Every card comes back as its own image, together in one ZIP, named in reading order from top to bottom and left to right. The sample on this page is one photo of two cards, and CardMagic returned two files. Lay out a lot, take one picture, and each card is ready to list.
CardMagic handles up to a full 9-card binder page in one photo, and every card it finds comes back as its own file. What changes is detail: a 12-megapixel photo of one card gives that card all twelve million pixels, while the same photo of a full page splits them nine ways. Shoot a page when speed matters, and one or two cards at a time when the listing deserves the crispest image.
No. CardMagic runs on the same AI as CardSight AI's card identification, so it works in any real-world condition. The background is for your buyer, not the AI: the finished image keeps a small margin of the real surface around the card, and when that margin contrasts with the border, the edges and corners stand out so the buyer can see the condition for themselves. Use a darker surface for a white-bordered card and a lighter one for a black-bordered card.
Yes. Turn on corner close-ups and every card comes back with four 600 by 600 close-ups of its corners, plus one sheet that lays all four out as they sit on the card. Each close-up shows 14 mm of card from the corner and 2.5 mm beyond it on a light 1 mm grid, so rounding, whitening, and dings are easy to judge. The close-ups keep the photo's own color, since auto-levels is not applied to them, and the grid is scaled for a standard 2.5 by 3.5 inch card.
It changes the photo, not the card's condition. Cropping, deskewing, and dewarping fix framing and perspective, and auto-levels adjusts contrast and color cast across the whole image so white borders read white. Every correction applies to the whole image evenly and nothing is painted over, so corners, edges, and surface flaws stay exactly as captured. That keeps your listing honest and your buyer confident. If the photo's own color is the point, turn auto-levels off.
Any of them. The output is a standard JPEG, or PNG if you ask for it. The two sample crops on this page came out of a 12-megapixel phone photo at roughly 1400 by 2000 pixels each, which is large enough for zoom on eBay, Fanatics Collect, CollX, StockX, and your own storefront. Nothing is watermarked or locked to a platform.
Yes. Every CardSight AI plan includes every feature, and CardMagic follows the same rule, starting with the free tier. Pick the plan that matches your monthly volume and CardMagic is already there.
Yes, and that is where it pays off most. CardMagic is one endpoint on the same API as card identification and price data, with the same key and the same SDKs, so a marketplace, shop system, collection app, or AI agent can offer listing-ready images from a phone photo without building any imaging of its own. Most integrations are a day of work: send the photo, save the files that come back.
Yes. Send the same photo to card identification and to CardMagic. One capture gives you the player, set, year, card number, and market comps, plus the framed image to list it with. Together they turn a stack of raw cards into finished listings in one pass.