Guides

    Geocaching events: cache still plus coordinate title card, not a treasure Omni

    Photograph the container you hid. Title-card the posted coords. QR the log URL. A cinematic hunt invents a park you do not permit.

    Versely Team6 min read

    Photograph the container you hid. Title-card the posted coords. QR the log URL. A cinematic hunt invents a park you do not permit.

    A geocaching event that books is a photograph of this container — ammo can, bison tube, magnetic nano, the hide you actually placed — then a title card that matches the coordinates you already posted, and a QR that opens the log URL you already host. It is not a cinematic treasure film of a park you do not have a permit for. It is not an Omni hunt through woods the model invented. Shoot the cache. Burn the posted coords. Encode the log.

    Photograph the container you hid

    Photograph the container. Same light as the hide if you can; a table still of the closed can if the hide itself cannot be shown without spoiling it. Do not generate a prettier cousin of a bison tube you do not stock. Text-to-video “cinematic treasure hunt at golden hour” invents a property. Average is a stock woodland, not your posted coords. Motion is allowed — a slow orbit of the closed can. Invention is not.

    Do not generate hunters in the woods. A photoreal body on a trail is a private figure you invented. Empty hands on the container, or the container alone. Do not generate a child finding a cache.

    If you even need motion, image-to-video starts from that still. Prompt camera only. “Slow orbit of the closed ammo can, no people, no new park.” Stop. Re-describing “a magical geocache in a misty forest” is how the model starts drawing a second hide in a place you cannot send people.

    Title-card the posted coordinates

    The card carries the offer and never more: event name, posted coordinates (same datum, same precision), date and meet time you already posted, cache size if that is on the listing.

    Add a title card to a video is a centered headline with an optional background block, positioned as an intro rather than as a caption. It is add_video_captions used on purpose: one short line, center, a block behind it so it reads over the can. There is no dedicated motion-graphics intro. Coords plus the event name is a title, not a paragraph. Swap only the copy when the listing changes.

    A persistent badge for the whole clip is add a text overlay: one authored string, whole duration, top / center / bottom. Do not ask a video model to letter the coordinates into the leaves. Prompted type is a pin you cannot edit.

    Qwen Image 3 Text to Image is the stills row if you need a printable coord card before it hits video: complex text in Chinese and English, intelligent prompt rewriting on by default, 4 credits, 2K max, qualities 1K and 2K. Quote the coordinate string. Proof at 100%. If the model “improves” a digit, you still proof the string. A cache that ships N 51° 30.123 and a card that shows N 51° 30.132 is two hides. Do not ask Qwen to invent a working map. If the digits are wrong, regenerate the still.

    QR the log URL on the device

    The log URL is already on the listing. Encode that string. Do not prompt a video model to “draw a QR code.” Prompted codes do not scan.

    QR Code Generator encodes text or a URL as a Model 2 QR in byte mode, with the four standard error-correction levels (L / M / Q / H). It draws locally. Nothing is uploaded. The quiet zone is the spec’s four modules. Invert the colours or crop that border and many cameras will not read it — leave both alone. For a log URL, stay ASCII.

    Drop the PNG on the end card. When the log URL changes, re-encode. The container still does not have to change. Text to image is the launcher if the coord card is a Qwen still. It is not a QR encoder. Use the free tool for the code. Use Qwen for the letters.

    A cinematic hunt is a park you did not permit

    Gemini Omni, Veo, and any reference-to-video “treasure hunt through a misty park” will invent a location. That location is not your hide. It may be a park that requires a permit you do not have. Shipping it as the event film is a claim about a place you cannot send people.

    Do not Omni a hunt. Do not generate a trail you did not walk. Photograph the container. Card the coords you posted. QR the log. Recur the format when the event moves; swap the still and the strings. Overlay the event rules as type if you already print them. The coords belong on the card you typed.

    Pause on the container. If a finder would not recognise the hide from the freeze-frame, you used text-to-video or someone else’s park. Real can. Posted coords. Scannable log. No invented woods.

    FAQ

    Can I generate a prettier woods because the real hide looks thin?

    No. A generated park is a claim about a place, and often a permit you do not have. Photograph the container you hid. Overlay the posted coordinates. An honest ammo can beats a fake grove.

    Why a title card instead of burning the coords in the generate?

    Because the posted coordinates are the offer. Prompted type is a pin you cannot edit. Add a title card burns the line already on the listing. When the cache moves, change the card, not the woods.

    Will a generated QR in Qwen or Omni scan?

    Do not count on it. Encode the log URL with the QR Code Generator: byte mode, local, quiet zone intact. Composite that PNG. Prompted modules are decoration.

    Is this an Omni reference-to-video hunt if I attach a park photo?

    No. A park still as reference is how the model invents a trail through a place you may not permit. Keep the container still. Title-card the coords. QR the log. Leave Omni off the event.