Scene gist refers to the ability to rapidly process the dominant meaning of a visual scene within the first ~100 ms of presentation. It is assumed that the gist especially relies on global and holistic information, such as the spatial layout. Research has shown that modified images (e.g., by the algorithm by Oliva & Torralba, 2006), designed to represent this “unique shape” of a scene can also trigger above-chance categorization of scenes. Our study examines whether emotional scene gist is preserved in global gist information. Therefore, valence connotations of initially neutral scenes were induced by evaluative conditioning. For conditioning, scene images were paired with five-second music clips evoking positive or negative emotions. Explicit ratings confirmed conditioning for these scenes. Most important, in an evaluative priming task, both the unmodified scenes and algorithmically transformed versions created with the Olivia and Torralba algorithm (OT6) served as 100ms primes for valence judgments of adjectives. We found an evaluative priming effect (shorter RTs for congruent prime-target pairs) for unmodified images, corresponding to the conditioning effect assessed by ratings. OT6 images did not cause a priming effect in the overall analysis. A post-hoc analysis, however, indicated that for those participants with a clear conditioning effect (according to ratings), there were significant priming effects for both unmodified images and OT6 images of equal strength. For the rest of the sample, priming effects for unmodified images were marginal and non-existent for OT6 images. Thus, we cautiously conclude that initial evidence for the basic hypothesis was found.