A number of violent far-right attacks in recent years have revealed an apparent connection with `chan culture', not just in the tangible examples of attackers uploading manifestos, final messages and livestreams to chan sites themselves, but in the widespread community support exhibited in some corners of this online subculture where violence is both trivialised and glorified. Commonly, this is manifested in the visual culture present on chan sites, particularly memes, which may be used to promote extreme or even violent narratives under the guise of humour and irony. This paper seeks to understand how the visual culture of chan sites contribute to, and/or encourage violent discourse. In particular, we combine quantitative data scraping, ethnography and visual analysis across tens of chan sites ranging in popularity like 4chan, 8kun, or neinchan between March -- June 2020. Over all we collect a dataset of 135K images across different chans and provide first qualitivative characterization of the most popular images shared across different chans.