Could be. However I would expect the physics engine component to suffer much much much earlier, even at a 1/5 scale planet. A diameter of ~2.500 km would still be too much at least if they used the engine and the static coordinate system as it was meant to be.
Which they didnt do as far as it seems since they told they moved on to floating origin approach where you have to think differently about scaled physics and some other stuff anyway. Maybe you’re right if they use a rawer floating origin or lets say segmented approach (I wouldnt call it floating origin then anymore) in a custom partially double coordinate system where they don’t reset the coordinates that often like you would do in a completely floating point based engine. They were definitely fighting with floating point precision at the point of their first planet generation iteration at least, the stuttering in some of the videos was a clear indication. But my thought was that this was not physics related but more general due to overall (re-)position problems in combination with FO. But at least that seems to have become better, the later builts run notable smother. But I am pretty sure there still are floating point restrictions around the implementation, the design of speeds (slower speed for battles and static warp drive in a certain direction) seems to be a workaround for network or float problems (or an indication for instancing?), with FO and some techniques speed handling could have been much more flexible (at least to handle the floating point limitation)?
My impression of planetary scales is maybe more in the direction that the sizes are a design choice but also (even more?) for performance reason of their planet generation as a combination of procedural and artist-driven custom surface generation. In the videos they seem to use a relative low number of terrain patches. Eventually there is a reason for that and they don’t want or cant traverse too much into finer sized patches. Despite the float limitation that still might be here and there in their reimplementation maybe also the number of detail they put onto the surface comes with a price.
We don’t know, I think it really depends on how much of the engine was customized, the (noise/custom-based planet generation) and if/how floating origin was applied.
I am fine with the design choice if they keep going the route to put more gameplay elements in like the city shown in the demo and than for example available in ED, until the budget ends…
Thanks for the example calculation, it seems pretty valid to me. I did a similar rougher one a few weeks ago, and came to ~30M USD, but with 330 developers, not checking todays number. IMHO the budget restriction is a good thing. They will have to come out of the comfort zone somewhen. The SC story will end somewhen probably not too far away, I don’t think next year but somewhen. Good or bad, but it definitely ends. And as positive sideeffect hopefully that ugly DS face will disappear in non-relevance again.
I think 3.0 will be interesting too see in terms of release date and, if available, also as an indication of the implementation strategy chosen discussed above.