"Best guess is he is using Midjourney to render the images. It is driven by prompts which combined with the language, switches and seed, one can get to a very specific look. He may have even used a reference image for the app to use. The inserted text and its formatting is generated by the app. I messed around in MJ this AM and was able to generate imagery very similar with minimal fuss; e.g. "an online ad for the Atari barcelona gamer chair by Atari, leather, paneled wood, 70's aesthetic, Atari branding". If you replace "Barcelona" with another style or append the prompt with the designer you can reach even more specific results. "