The CAM powers the FTDI, not the other way around?
Anyways, I have bad experiences with USB-UART adapters powering really anything. The ESP32 (and the camera) will pull a lot of current during startup and a brown-out may occur. That’s also noted there.
If you have followed their exact wiring (with your modification that VCC from the FTDI is passed to the VCC of the CAM) and methodology (unplug VCC, hold down button for GPIO0 or short it, repower board) and the upload still doesn’t work, you should try using a stable external +5V power supply (e.g. USB cable with the pure pins on the other side for +5V, GND, D+, D-, or using another dev board which passes the USB voltage directly through unto a pin header).
That’s a dilema actually, because the I/O voltage of the ESP32 is 3.3V. But you need 5V on the power pins. But I guess if you put the jumper on the 5V side of the FTDI, the VCC pin as well as the I/O levels become 5V, not being controlled separately.