Affective image editing (AIE) is the task of altering an image so that it evokes a specific emotion in the viewer, without changing what the image depicts. A new preprint on arXiv proposes IECLA, a method that frames this as a problem of interpretable emotion contribution learning.
Existing approaches, the authors note, tend to improve either the emotional expressiveness of the output or its visual fidelity to the source. But the abstract suggests that focusing on the target emotion alone is not enough—implying that a more fine-grained understanding of how individual visual elements contribute to the emotional response is needed.
Because the abstract is truncated, the full details of the IECLA method are not available in the source. The paper is listed as a new announcement on arXiv under identifier 2610.25015.