By
Jamon
Thu Sep 17, 2009 8:22 am
Think of it like this, you take a high-resolution digital photograph of a model. You load it into your PC and can zoom in close and still see detail. It isn't like actually seeing her with your eyes, but it looks similar when you zoom out and everything comes together.
To upload it to the web, you want to reduce the amount of data because bandwidth/storage costs money and is limited. You compress to JPEG, a lossy format. If you zoom in on both the original and compressed versions you'll see big blocks on the lossy image.
But those blocks aren't very noticeable unless you zoom in, due to the way the human eye/mind perceive sight. Still, the data is missing, so even if humans can't notice easily, machines performing calculations do, and have less resolution to work with.
You don't like the blocks, so you take a picture of the jpeg image using your 35mm camera. You develop it and look at the enlarged print. How does it compare to the original lossless digital photograph? It is an analog representation of a digital image.
It's true that the blocks are not represented in the same way, not as defined, there's now some noise introduced by the film grain. But the data is still missing, and it is not anything like if you took a photograph of the model using the 35mm to begin with.
Now when you scan, and recompress to jpeg to upload to web again, you lose even more data. The result is a blurry, blocky, poor quality image compared to the original. You might not notice any quality loss in your mix, due to psychoacoustics.
The problem becomes more apparent when compressing for the web. Since it is a recompression, you lose more quality than if you had the original audio to sample. Now what happens when someone samples your song and repeats?
I know factually that sometimes audio on the web, even on the official website of a label/band, was already recompressed. They sometimes give the web developers MP3s to work with, so they have no choice. Then people download that, and recompress again.
It's impossible to know where your sounds have been if you just grab them from unknown web sources. Some people even use pro samples they purchased to make loops, upload as MP3 to web, then someone samples that, chops to get individuals sounds, etc.
That's why it's best to try and keep a high-quality lossless signal from beginning up until final distribution, where if you need to compress lossy at least you know it's as clean as possible. Hopefully soon when tech increases we can kill lossy formats and only use lossless.