While text input remains the primary interaction method with most computer applications, it faces diverse challenges with spatial computing devices becoming successively more wearable and mobile. We present TipTopTyping, a novel computer vision-based thumb-to-finger text input method for virtual and augmented reality that uses pinch gestures between thumb and fingertips for easy and more intuitive text entry. With OPTI, we further propose a new character layout specifically designed and optimized for this input modality. The system performance is evaluated in a user study (N = 20) of two mobile scenarios: standing and walking. After only 12 sentences of practice, the participants quickly achieved mean text entry rates of 6.15 and 5.69 words per minute and mean accuracies of 1.27 and 1.43 keystrokes per character while standing and walking, respectively. Furthermore, OPTI not only shows a 4.36 % better typing accuracy but also the potential to outperform QWERTY layouts in writing speed with a little more practice.
