Ah I understand. What I think you are saying is that the compression level is the same in both cases and that you can't say from the filesize otherwise; is that correct?
Basically a softer image will (usually) lead to less bpp with the same JPEG-compression-factor used.
As the HR-image is visibly softer per pixel it´s clear that filesize couldn´t be twice as much as DR/HR. If the filesize would be twice as high, they in fact would have decreased the lossy compression in HR-mode.
Well you may be right. But could it explain the lower quality of the HR pic?
There are several reasons why HR-mode can´t deliver the per pixel detail of the 6MP-mode.
The first, and I guess especially with the F70 most significant reason is the lens which simply isn´t sharp enough to project a clear image on the tiny photosites. That´s quite common with almost all tiny superzooms, most of them are put into shame even by the old S6500, even in their lowest sensitivity, although they have much higher nominal resolutions.
From what I have seen there seems to be hardly any more detail available on the F70 in HR-mode.
The F200 is a bit different, as the lens is capable to deliver more than the sensor can capture in 6MP-mode. There is visilby more detail in HR-mode, especially in ISO100-200 and high contrast detail stays more defined up the ISO-range (low-contrast detail is being eaten up by NR after ISO400), but the F200 also can´t deliver as much detail as the double resolution would be able to do theoretically.
Next thing is the low pass filter, which of course stays the same. Therefore in HR-mode it is stronger "per pixel" (1 "pixel" in the binning-modes is the area of 2 photosites)
Also NR is stronger in HR-mode, but this shouldn´t affect high contrast detail much, only low contrast detail is blurred away. But of course this improves compression a lot, as the uniform areas the NR produces are much easier to compress.
In situations where mainly one primary color is involved HR-mode can´t add any real resolution, as 2 same colored photosites are always next to each other and the distance to the next pair in the same color is quite big.
The cause and effect chain is somewhat backwards.
It´s not the higher compression, which causes soft images and less per pixel detail.
It´s the missing per pixel detail which causes a higher compression ratio even at the same compression level.
You can also try to use Normal instead of fine compression. Although the compression is much higher you will struggle to see any difference. So it´s not the compression eating details away.
