Low-resolution online videos are less likely to influence opinion and also more likely to dissuade viewers from engaging with future content, research by Oregon State University scientists shows.
The study carries major implications for the design and delivery of video content and suggests that deviations from high-quality presentations can create repercussions regarding the video's content, according to Christopher Sanchez of the OSU College of Liberal Arts.
Sanchez says it's not surprising that the findings, published in the journal Displays, suggest people prefer high-quality videos and will opt for them if given the choice.
"Where this gets interesting, in my opinion, is that viewers don't always have that choice," he said. "Anyone who does much streaming will have encountered scenarios where the streaming service scaled down resolution to meet bandwidth requirements - for example, when you're on your phone with a spotty connection. Same thing with videoconferencing providers. But when a scale-down happens, ostensibly to preserve the connection or prevent buffering, it has downstream consequences that are unrelated to the material itself per se."
In this project, Sanchez showed study participants a five-minute video of a news program type of discussion on assisted suicide; half of the participants watched a comparatively low-quality version of the video, the other half a comparatively high-quality version.
"It appears we have a strong preference for high-resolution media," he said. "This preference seems to guide our engagement, both with future material and also with the content at hand. We become less engaged with what's in front of us, have less reaction to it emotionally and become less receptive to the opinions being expressed."
Participants who viewed the low-resolution video remembered the information presented just as well as those viewing the high-resolution video, but their attitudes toward the video content or message, in this case support of assisted suicide, shifted less.
"If you are in charge of messaging for legislation supporting assisted suicide, for example, this would be a bad thing, in the form of time and money wasted," Sanchez said. "And the degraded video also made people less likely to engage with such material in the future, which is potentially even worse - the current ad might be a miss, but now people won't even come back for a future one that might be better for any number of reasons."
He added that while this study didn't examine artificial intelligence's role in video production, it can nevertheless inform future study on the effects of AI-generated video.
"AI is used very much nowadays to generate video content, and our work seems to suggest that more realistic, higher-definition, higher-quality video productions are best if one wishes to influence viewers or at least keep them watching," Sanchez said. "AI has enabled the creation of high-quality video snippets with a simple prompt, which as viewers potentially opens us up to being influenced or increases our likelihood of doom scrolling; higher-quality video produces higher engagement in multiple ways."
Doctoral student Nisha Raghunath helped conduct and author the study, as did Chelsea Ahart, a former undergraduate member of Sanchez's research group.
