Visual attention is the process that enables us to select relevant visual stimuli in our environment to achieve a goal or perform adaptive behaviors. In this process, bottom-up mechanisms interact with top-down mechanisms underlying the automatic and voluntary orienting of attention. Cognitive functions, such as emotional processing, can influence visual attention by increasing or decreasing the resources destined for processing stimuli. The relationship between attention and emotion has been explored mainly in the field of automatic attentional capturing; especially, emotional stimuli are suddenly presented and detection rates or reaction times are recorded. Unlike these paradigms, natural visual scenes may be comprised in multiple stimuli with different emotional valences. In this setting, the mechanisms supporting voluntary visual orientation, under the influence of the emotional components of stimuli, are unknown. We employed a mosaic of pictures with different emotional valences (positive, negative, and neutral) and explored the dynamics of attentional visual orientation, assessed by eye tracking and measurements of pupil diameter. We found that pictures with affective content display increased dwelling times when compared to neutral pictures with a larger effect for negative pictures. The valence, regardless of the arousal levels, was the main factor driving the behavioral modulation of visual orientation. On the other hand, the visual exploration was accompanied by a systematic pupillary response, with the pupil contraction and dilation influenced by the arousal levels, with minor effects driven by the valence. Our results emphasize that arousal and valence should be considered different dimensions of emotional processing both interacting with cognitive processes such as visual attention.