What is the main difference between a "Normal" SQL Injection and a "Blind" SQL Injection vulnerability?

The visibility of requests to the server administrator is irrelevant to the classification of SQL injection as normal or blind; the distinction is about what information is returned to the attacker in the response.

Blind SQL injection does not imply that the application properly filters input; the application is still vulnerable, but it simply does not echo results or errors back in the response.

Whether error messages are shown to the administrator is not the defining criterion; the key factor is whether injection-related output is displayed to the attacker making the request.

In a normal (in-band) SQL injection, the application reflects query results or verbose error messages directly in the HTTP response, giving the attacker immediate feedback. In blind SQL injection, the application suppresses this output, so the attacker must ask true/false questions or introduce deliberate time delays (e.g., using SLEEP or WAITFOR DELAY) to extract data one bit at a time. The distinction is entirely about whether injection results are visible in the response to the attacker, not about filtering or logging behavior.