Look closely.
Camera and sensor variation can be studied as signals. The experiment below samples one observatory camera, conditions its JPEG bytes with system randomness and shows the source used. A changing signal alone does not establish a calibrated entropy rate.
Explore earth ↗Make a record.
Choose a source before drawing and retain every result, its time and reported source. Camera acquisition takes up to two seconds and 128 KiB. Compare repeated samples and document processing; a visual pattern or short sequence does not establish randomness quality.
Make a draw. Keep a record.
Generate hexadecimal bytes from the shared sampler.
How this draw works
The server takes up to 128 KiB of JPEG image data from one fixed stream, with a two-second timeout, and mixes it with fresh system randomness using SHA-512. The browser uses the same source module as the standalone key tool and mixes this response with its own random seed and uses fresh browser random words for each result. Rejection sampling avoids modulo bias. If the sampler is unavailable, results use browser randomness and the status says so.
Samples are experimental: no physical entropy rate has been established. Camera delays and repeated frames are possible. No predictive meaning is assigned to results. History stays in this page and clears when you leave or reset.
Recent draws
One source, many possibilities.
Explore SSH pairs, hexadecimal and other secret encodings, UUIDs, TOTP setups, QR codes and ASCII seed art using the shared camera and browser source.
Open the local laboratory ↗Share a photograph or a record.
A photograph, drawing or dated record can be sent privately for a personal reading. Include the procedure and the question.