"CAN YOU DO A BULGARIAN REVOLUT?" – WE DID. FIRST DRAFT, NO EDITS
The request came in short: a photo of a bank card, Bulgaria, Revolut. We said yes and asked the only question that matters – what data. The client was on the road, asked for a couple of hours, then sent the lot: number, expiry, CVV, payment system, bank, cardholder name.
Then came the part the client never sees:
ran the number through Luhn – the check digit has to add up, otherwise the card is dead the second anything automated looks at it matched the BIN to the issuer: Revolut in Bulgaria isn't a "local bank", the cards run through Revolut Bank UAB. The number range has to actually belong there, not just start with the right digit built the current design, not a five-year-old picture: vertical layout, flat laser print instead of embossing, correct chip module, contactless mark where it belongs the payment system logo – with the right gradient where the circles overlap and the wordmark sitting exactly where it should. A small thing that kills half the fakes out there and finally, physics. Plastic has to catch light: a highlight along the edge, a micro-shadow under the chip, a faint fingerprint. A phone shot, not a render
Sent the draft for approval. The client looked it over, signed off with zero corrections – final files went out. One pass, no rework.
A card is a deceptively simple document. Looks like a rectangle and sixteen digits. In reality it's maths in the number, a tie to the issuer, a current design and light on plastic. Miss any one of the four and the file gets exactly two seconds of attention.
Need a card for a specific bank and country?Message us. We'll tell you straight what's realistic for your case – then build it.
{
all data published with the client's consent}› TG: @mustang_service_ms ( https:// t.me/ mustang_service_ms )
› Channel: Mustang Service ( https:// t.me/ +JPpJCETg-xM1NjNl )
#drawing #card #photoshop #revolut #case
Mustang Service – when approval matters more than blind testing.
🪙 BANK AND CRYPTO CHECK DIFFERENTLY
🪞
YOU HAVE THREE NAMES AND NONE OF THEM MATCH
In the machine-readable zone your name follows the ICAO standard. In the passport's main field – a national transliteration scheme. In your bank – whatever was typed when the account opened.
Sergei, Sergey, Serhii. Iurii, Yuriy, Yurii. Formally one name, to the system three different people.
Matching runs character by character. A single letter apart creates a mismatch between form, document and payment method. Genuine data, genuine document, result: rejection.
Extra traps: apostrophes in surnames, double given names, and the order of fields in the MRZ – surname comes first there, and that gets mixed up constantly.
Before submission we align the spelling across all three points and match the form to the MRZ version.
Name spelled differently on your card and your passport? Write before submitting, not after the refusal – this takes minutes to fix.
AN ALGORITHM FLAGGED HER PHOTO – SHE SPENT NEARLY SIX MONTHS IN CUSTODY
A fake military ID surfaced in a bank fraud case. Facial recognition returned a social media photo of a 50-year-old Tennessee woman as a possible lead. She was arrested.
The system suggested a face. Humans were supposed to establish whether it belonged to the person behind the crime. The cost of getting that wrong was far more than a rejection screen.
A DOCUMENT'S SECURITY WORKS AGAINST YOU
Holograms, kinegrams, colour-shifting OVI ink, microtext, guilloche patterns. All designed so a human at the counter spots authenticity instantly.
To a camera, those same features are interference. A hologram throws a rainbow blob across the number. OVI ink shifts hue between frames and breaks template matching. Microtext collapses into mud under compression.
The paradox is brutal: the more modern and protected the document, the harder it is to shoot without the machine reading its security as tampering.
The fix lives in geometry, not in an editor. Tilt angle, distance, quality of light, position of the source relative to the laminate. Security features must be moved off the data zone, not suppressed.
A modern smartphone processes your face before you press the button. Skin smoothing, contour shaping, lifted shadows, denoising, HDR merging of several frames. All by default, unasked.
Deepfake detectors hunt for exactly those traits: missing pores and texture, unnaturally clean gradients, symmetry, noise that doesn't match across the frame. Beauty mode produces the same artefact profile as generation.
A ring light finishes the job: a perfect circle in the pupils, flat shadowless illumination, plastic skin. Precisely what synthetic models output.
A real live selfie looks rougher. Pores visible, one side softer in shadow, uniform grain, an asymmetric face.
🩸 A WARM LAMP REPAINTS YOUR DOCUMENT
Every document has a reference colour: paper tone, background, guilloche pattern, plastic. Providers keep sample images and compare your frame against them.
Phone settings finish it off:
Worst case: mixed light. A window pushes cool, a lamp pushes warm, and two colour temperatures live on one document at once. Nothing physical explains that.
We bring the frame back to neutral without touching a single data field.
Spanish police took down a gang that found a hole in a major telecom operator's online store. A script told the warehouse to ship before payment went through. The haul: over 150 smartphones worth more than €50,000.
The cherry on top: gang members passed video KYC themselves, posing as the elderly document holders. The "customer's" age didn't bother the antifraud.
Services run files through ELA – error level analysis of compression. JPEG resaves every zone differently: untouched areas glow evenly, edited ones burn brighter.
