Inquiry about the Fig-6

[yuhaoliu7456]

Does anyone know how to generate the visual results in Figure 6? I see that they extract SSL representations from image samples, and the authors don't seem to describe how they combine these features with randomly generated noise in the RDM.