This makes little sense. Analysis shows that when at 3840x2160, Assassin's Creed Origins is using CBR. According to the developer, it is not always at 3840x2160. My assumption was that, no matter the buffer size, it always uses CBR. I believe this because that's what other dynamic CBR games do.
CBR is a generic term for starting with a checkerboard pattern to render pixels and then doing something to fill in the gaps. How one fills in the gaps can vary depending on what the devs want to do. I'm describing a 2 step process for filling in those gaps. It is no different in principle than using a non-checkerboard sampling as the base and reconstructing the rest. Literally the only difference is choosing what pattern to use for sampling. What games use dynamic CBR that you are referring to?
I also believe it because the alternative you're proposing seems perverse.
It's nothing more alien than any other pixel reconstruction technique. Most devs don't bother using those approaches at all, but it is far from unheard of. All I am discussing is the process of starting with some sampling pattern to render some subset of the pixels, then using reconstruction to fill in the gaps. In the first part of the process you can do a checkerboard pattern and then fill in some portion of the missing pixels before using reconstruction to fill in the rest.
You say the game might not use CBR when the buffer is below 3840x2160, say at 3760x2115. But think about what you're actually claiming there: that the game can natively render 7.95m pixels, but instead of upscaling the last 342k, instead clicks on CBR to natively render 4.15m pixels and then reproject the other half. I mean, this isn't physically impossible, but I can't think of a single benefit should a developer make this choice.
We may be talking past one another. I'm saying one can render half the pixels a la a CBR pattern, then render some of the missing pixels on top of that, then for what is left over use reconstruction. The portion of pixels rendered in the second step can vary with processing load dynamically from frame to frame. If the second step doesn't take place at all and skis right to reconstruction for the other half of the pixels, you get what you are calling CBR (2160c, with half the pixels rendered and the other half reconstructed).
There is nothing preventing a dev from using CBR + dynamic GPU load balancing to scale upwards towards native 4k. It is no different in the procedure than scaling downwards as you assumed was happening. And we have no info yet to suggest to us one way or the other which was happening (sub-2160c or somewhere between that and native 4k). All DF told us was that there seemed to be visual artifacts suggestive of CBR used at some point in the rendering/scaling process. We know it uses some scaling technique similar to CBR due to the artifacting DF reported. We know it uses some form of dynamic res scaling. We have no info whatsoever about the upper or lower bounds on that dynamic scaling.
I would love portable Halo mp but I have a slight concern about the face buttons on the switch as it relates to halo controls especially mobility. I think splatoon will be a good test. Did anyone here try the beta? Is it a valid concern?
I don't think this is true. Do you have links to developers using the term in that manner? In any case, for the game we're actually talking about, it uses reprojection in a checkerboard pattern, so that's what I'll use "CBR" to mean from here out.
All I am discussing is the process of starting with some sampling pattern to render some subset of the pixels, then using reconstruction to fill in the gaps. In the first part of the process you can do a checkerboard pattern and then fill in some portion of the missing pixels before using reconstruction to fill in the rest.
But this doesn't make any sense. CBR can only double the number of pixels, so if you shade more than half, you'd be throwing away some of them just to replace them with reprojected versions, a waste of processing power.
But for the actual game we're discussing, CBR explicitly means reprojection of half the pixels, not some generalized class of methods. Given that fact, your suppositions and imagined techniques are superfluous.
Just to simplify, this is how your argument runs:
1. This game has dynamic resolution.
2. This game uses CBR.
3. There's no proof it uses both at the same time.
This is all true as far as it goes, but there's also no proof that it doesn't use both at the same time. Given the lack of deductive conclusiveness, we should fall back on inductive reasoning from precedent. Dynamic CBR is a real technique, used already in shipping games. There's no apparent rationale to dispute that it's also used here, except from a desire to avoid concluding anything.
If all you intend is to say that the assessment is necessarily provisional and subject to change if more information comes out, then I have no problem with that. But that doesn't mean "Assassin's Creed Origins is dynamic CBR 4K" isn't still the best, most well-supported conclusion we can make right now.
I was thinking this...but can they...so many of the networking issues were to do with the game being "in the wild"...can they truly fix it without some sort of soft launch to check the fixes *actually* work?
Can you imagine the internet meltdown if they launch a broken "fixed" version?