Originally Posted by mark haflich
You guys are putting too much into the wording..
Could be. It might not be the author being sloppy, it might be not understanding the subject matter. Although I do think there is a non-zero chance that the writer is correct and JVC has done this in an inferior way to what is possible (for schedule or technological reasons).
Others can probably explain it better and it seems obvious to me that there is a difference between downscaling to 1080p first and then working from there vs extracting both 1080p images from the original 4K, but I'm going to try a simple example to explain a difference.
Let's assume that product A has to deal in integers and a new product B comes along from a competitor that can deal with resolutions down to 1/4th of an integer. It would be too expensive for product A to be modified to have 1/4th of an integer resolution, so the engineers come up with a way to deal in 2 integers with half as much time applied to each. To viewers this looks pretty much like 1/2 of an integer resolution. I'll call this new one product A2.
If product A2 can only take integer input then there is only so much that can be done to be similar to 1/2 of an integer resolution. However, they come out with a product A3 that can now take 1/4th of an integer inputs. They still have integer output, but with a system that can look like 1/2 an integer of resolution.
The engineers could take that 1/4th an integer input and convert everything to single integers, then extract their 2 integers from that, or they could apply a little more intelligence and realize that extracting their 2 final integers from the original makes a lot more sense.
As an example, if an input was say 3.5 the dumb way to deal with that would be to convert it to 3 or 4, then figure out how to split that up into 2 integers based on other data. The smart way to do it would be to convert it to 3 and 4, for an average of 3.5, then deal with it from there. If the input was 3.75 then they couldn't do that value, but could use 3 and 4 (for 3.5 average) or 4 and 4, or use one of those and keep some meta deta around in case they want to push it toward an average of 3.5 or 4 depending on other data.
I realize this is different and very simplified compared to JVC's situation, but if the above situation happened and a writer claimed that there was no advantage to 1/4th of an integer resolution input on product A3 I would say that they either don't know what they are talking about, or they do and the manufacturer has a dumb implementation that converts that 1/4th of an integer resolution input to a single integer resolution value (not 2 single integer resolution values straight from the input signal).
--DarinEdited by darinp2 - 10/21/13 at 4:51pm