Maybe a dumb question, but why is “add 3-EQ” a default option when right-clicking on bus groups? I thought Multiband EQ was a more recent and optimized effect, so I don’t get why this older effect is suggested by default in this specific menu.
There’s two parts to this question, so I’ll address each of them separately.
It’s true that the Multiband EQ effect was added to FMOD Studio more recently, but that doesn’t make it better or more optimized than the 3-EQ effect.
The two equalizers are both equalizers, but they’re different types of equalizer that do different things, and so are useful in different circumstances. It’s like how the object spatializer effect doesn’t supersede the spatializer effect, even though the object spatializer was added later and is also a spatializer: Both effects are useful in different situations, and you’ll pick which one to use based on which best meets the requirements of the content you’re working on.
It’s also not true that that the Multiband EQ is more optimized or efficient than the 3-EQ. They’re both as highly optimized as we can make them (as are all FMOD effects). However, because a multiband EQ does something that’s inherently more complex than what a 3-EQ effect does, it’s a more expensive effect to run than the 3-EQ effect is.
If the Multiband EQ effect was an objectively-superior replacement for the 3-EQ effect, rest assured that we would have moved the 3-EQ effect into the “FMOD Deprecated” submenu.
For more information about the differences between these two effects, I recommend reading their entries in the Effects Reference chapter of the FMOD Studio User Manual.
It’s for compatibility reasons.
FMOD Studio supports plugging hardware mixing desks into your local network and using their physical controls to manipulate FMOD Studio’s UI. Because some hardware mixing desks feature dedicated 3-EQ controls on their buses, we added a convenience feature that adds a 3-EQ effect to a bus in order to simplify the process of making FMOD Studio projects fully compatible with those hardware control surfaces.
Thanks, I got it!
However…
… that’s not what I found in this little test, a few years ago:
Which is why I thought multiband EQ generally was the best choice.
As your past self correctly guessed, changing the settings makes a big difference. By default, the multiband EQ has one band set to a 12 dB lowpass filter and the other four bands disabled; enabling the other bands or setting enabled bands to more expensive filter types can significantly increase the effect’s CPU cost.
This is why our documentation describes the Multiband Equalizer effect as “low overhead to medium overhead” instead of just “low overhead” or “medium overhead”: Depending on its settings, it can be in the same cheap ballpark as the 3-EQ, channel mix, and gain effects; or in a similar league as the more expensive compressor, multiband dynamics, and reverb effects.
That’s not totally fair: as I showed, the multiband EQ with 3 bands enabled (low shelf, peak, high shelf, to emulate the 3-EQ) is still less expensive than the 3-EQ. And since it’s possible to enable only 1 or 2 bands on the multiband EQ (which the 3-EQ can’t do), I believe it is superior in any situation.
I just wanted to chime in here to provide some additional context.
The key difference between 3-EQ and Multiband EQ is that 3-EQ processes three bands in parallel whereas the Multiband EQ processes (up to) five bands in series. It’s the parallel vs series that ensures we keep both effects as part of FMOD. My advice would be always prefer the Multiband EQ unless you need the specific capability of parallel band processing provided by the 3-EQ.
For some technical details: the 3-EQ and Multiband EQ use biquad filters to achieve their goals, if you give each biquad filter the cost of “1” you can roughly compare the two effect’s performance. For instance, in 12dB (x-slope) the 3-EQ uses 4 biquads to isolate the bands, so that’s a cost of “4”. Each individual filter provided by the Multiband EQ that doesn’t list a dB value uses 1 biquad, 24dB uses 2, 48dB uses 4. This means the minimum cost of 3-EQ is already four times that of Multiband assuming you just want a simple EQ curve. The cost of 3-EQ is not strictly four times though since it operates in parallel we get some optimization efficiency back.
Hopefully that helps explain why we still have the two, and the key take away is use Multiband EQ unless you need parallel processing.
1 Like
Thanks for the insight!
What is “the specific capability of parallel band processing” that we might want? Could you give an example of what it might be used for?
The structure of the 3-EQ filters provides frequency isolation of the low, mid, and high bands. Conceptually, it does this by taking the input signal, splitting it 3 times and filtering it (with low/high pass filters) independently so when recombined there will be a flat gain response. This allows you to automate the low, mid, and high cleanly, and you can choose higher slope values for sharper frequency isolation.
To do the same with the Multiband EQ is tricky since the bands are processed in series, one after another. You cannot use the same approach of low and high pass filters since isolating the low band will leave you with none of the high band. You could try to imitate the 3-EQ by using a low-shelf, peaking, and high-shelf filter but it loses a lot of the precision (band isolation) that 3-EQ gives.
So, from a technical perspective, 3-EQ is for precise band control. Why might you want this? It’s a good option for user controls, exposed (automated) directly to the game. Whereas Multiband EQ is the workhorse for all your every-day filtering needs.
1 Like