In Spectrogram view, at some sizes, you can see certain frequencies being there. Then when you resize the view, they will disappear, and others will appear. I don’t know how to read the spectogram if changing its size changes which data it displays.
Picture examples follow. Highlighted in red are the changes in data shown. Top and bottom are the same part of an audio file, but resized differently.
In comparison, Image-Line’s Edison always shows you the same spectral data no matter the size. Here is a recording of the comparison: https://imgur.com/PZhuQn8
I’m hoping for an improvement in this area so it always shows the actual data.
This is true, WaveLab functions similarly to Edison in Linear scale. But if Edison is set to Logarithmic, it still doesn’t have the problem. How I see it is that they calculate the image at a fixed size, then scale it down to the desired size. That way there is no re-calculation/re-averaging needed, and the only loss of detail is the scaling of the image.
WaveLet view does look excellent. Because it needs to re-build it is quite disorientating for me however, and because of the speed it can’t be used to view a file while it’s played back live. I have an idea for this: if the image is built once for the entire length of the file, then it can just be shown as a picture afterwards, and it doesn’t need to be built while scrolling.
As a final comment, because there is no anti-aliasing/interpolation in Linear and Wavelet views, when it is scaled down it aliases, which also obscures some detail.
because there is no anti-aliasing/interpolation in Linear and Wavelet views, when it is scaled down it aliases, which also obscures some detail
There is anti-aliasing (compare to WaveLab 9 to see with aliasing!), but the smaller the display, the less details. Interpolation helps but can’t do miracles.
Some higher degree of interpolation would be possible, at the expense of speed.
As some additional information: the issue is especially prominent at a resolution like 1080p, and is less noticeable at 4K. So to see how much improvement is needed, I think it is best tested at a lower dpi.