Price only has to be proportional if you want to try to estimate the model well using Main Effects. If you are going to estimate interaction effects between the two attributes (brand and price) then you can do whatever you want in terms of the price range covered for different brands.
But, as soon as you have price dependent on 3 or more other attributes, then making the price ranges proportional and formulaic is extremely important. Otherwise, you might need a 3-way (or 4-way, etc.) interaction term to unravel the data and fit it properly, and our software doesn't (automatically, without power tricks and data reformatting) provide a way to estimate 3-way (or 4-way, etc.) interaction effects.