Previous | Next --- Slide 44 of 47
Back to Lecture Thumbnails

It is like we lose some information for each resample step. Without new information, we will definitely degrade image quality.


Is this how the deepfried memes are made?


Is there a different way of downsampling which tries to make the upsample of the downsample look closer to the original, rather than trying to make the downsampled version look closer to the original?


It looks as though we lose some fidelity on the high frequency parts of the images as discussed in the earlier lecture on sampling.


Is there something similar to the Laplacian image pyramid in the graphics world, where we can reconstruct a downsampled mesh without loss?


Why does it make the edges appear sharper? Is this always the case?


Does this also occur with lossless techniques?