The most optimal situation is to be able to hear the mix as the mixing engineer heard it. However, this is not possible in many/most situations.
I originally designed the upmixing portion of the True Image processor with a rather simple aim:
- If you are sitting in the correct location (commonly called the “sweet spot” – although this is misleading) with a correctly-configured system, then upmixing with True Image should not change your perception of the mix, so it is redundant – or at least unnecessary.
- If you are NOT sitting in the correct location, then upmixing with True Image should help to deliver a perceived mix (primarily the spatial attributes) that is more like what you would hear in the correct location.
Note that the downmixing portion of True Image has a different purpose – but that’s not what we’re talking about. Note as well that this was my aim – real-world results may vary with different loudspeaker configurations, listening rooms, and mixes.
For example: If you play Suzanne Vega (or listen to the lead vocal in any pop tune) over a correct 2.0 stereo configuration, but you sit slightly to the right of centre, then you’ll hear the vocals on the right. If you upmix that track to a correctly-configured 5-channel output, then you’re more likely to hear the vocals in the centre loudspeaker.
The Dolby Atmos decoder in Beosound Theatre (just like all other consumer-level Dolby Atmos-compatible devices that I know of…) can decode/render a maximum of 7.1.4 channels. It is typically safe to assume (but it is an assumption) that a Dolby Atmos-encoded signal was monitored in 7.1.4 when they made the mix.* Note that this does not necessarily mean that the mix actually uses all of those channels. Some mixing engineers choose to omit some channels (by sending silence on them) for technical or artistic reasons.
Based on that assumption, it is also safe to assume that, if the mix was in 7.1.4, then upmixing it with additional output channels is unnecessary, since the reason to make any surround mix is to make the perceived spatial attributes in the mix more similar over a wider listening area. (For example, if you have one chair and no friends and a good 2.0 channel stereo system, you have no need of a centre loudspeaker because phantom imaging works very well…)
Finally, to come back to your earlier comment (maybe “complaint”? 🙂 ):
The Speaker Role options that were originally available in Beovision 11 -> Harmony were based on a master set of multichannel formats that never actually happened in consumer-level distribution. For example, in the early days of 7.1-channel signals, DTS had 7 different variants of layouts. Sony’s SDDS system supported 5 channels across the front and only 2 surround channels. Since the Speaker Roles had to be ready to support all of those channels in all possible then-future formats, the True Image upmixer was made to deliver them when they were not present at the input. For example, if you chose to build an SDDS configuration instead of a now-normal 7.x configuration, then True Image had to be compatible with that choice, and therefore it would create outputs appropriately. Since these format configurations never actually materialised in distribution media, there is no real need to continue supporting them with upmixing just for upmixing’s sake. As you said: more isn’t necessarily better – sometimes it’s just more.
It’s important as well to separate the x.1.z vs. x.2.z vs. x.3.z difference from this, since that is a separate issue that is handled by bass management, not up/down-mixing.
* This may not be necessarily true for big-budget movies, since Atmos-compatible cinema systems can support up to 24.1.10 channels, which might actually be present in a large-scale re-recording theatre.
.
