Originally Posted by Ballis
Hopefully the 4x pixel upscaling Ruined mentions is used for a low latency mode in these new pjs.
Scaling need not introduce (much) latency, Lumagen's, with some of the best scaling on the market, introduce almost no latency. Why so many projectors have such enormous lag, or have such huge drops in quality in "game mode" I have no idea.
Originally Posted by Ruined
The poster I responded to posted they were concerned that most content was 1080p Blu-ray and we would be watching that in 4k. In other words, a concern something might be lost watching 1080p content in 4k as was often the case watching a DVD on an HDTV with poor quality upscaler.
The biggest problem, by far, with scaling DVD (480i) to HD (1080p) was not the scaling, it's the fact that DVD doesn't have enough information for "HD" screen sizes. SD was OK when people sat 10-12' from a ~25" TV. When you've only got 240 lines of chroma information, there's only so much you can do.
Of course there's the other huge issue with DVD, in that it was stored interlaced, and often had terrible mastering, and on top of that there were more bad deinterlacers than good ones. People seem to conflate deinterlacing issues on DVD/SD with scaling issues.
This actually was an issue (and remains an issue, actually, with poorly engineered solutions) with the transition from DVD to Blu-ray partially because 1920x1080 is not a multiple of 720x480. There is truly no way you can replicate something approximating 720x480 with a 1080p panel.
This represents a fundamental missunderstanding of resampling/sampling theorem. The whole "even multiple" thing is a myth, there's nothing in sampling theorem that says anything about requiring multiples for resampling. In fact it's quite the opposite, so long as your destination sampling frequency (resolution) is higher than twice the bandwidth of the signal, the resampled signal will contain all the information of the original.
All this "even multiple" BS seems to come from people who test scaling/resampling using contrived test patterns that are only good for checking 1:1 pixel mapping and focus. These patterns were not sampled in the first place (they were created pixel by pixel) and cannot be resampled at all
without being damaged, even multiples or not.