It still doesn't tell you what it represents, just what Bill's definition is. And why you'd want a SNR of 20 in the deep shadows is beyond me. Something like 2 does just fine.
July 15, 2023
33
-
IMHO the bigger issue, and not only going by the two screenshots posted in this very thread, is comparing PDR for the sensors of different sizes to compare the "state of the art".
IMHO 20 is fine for film because of the character of the grain in standard development, and more or less OK for JPEGs as of ~2 years ago.
-
Yes, that's certainly an issue.
20 as the lower bound? Do you need an SNR of 20 in the deepest shadows, zone 1?
-
i posted the figures for the 7 yold a6300 which seems to hold up even to this day.
-
at pixel level i will take the a7m4 larger pixels anyday of the week. theres even a sony ambassador quoting the same.
and shooting at high mags i have proven that the larger pixels can out perform smaller for detail. -
Where there is no texture, only slight tonal gradation, some like those to be smooth.
-
-
40 meg vers 14 meg
-
"Larger pixels, anyday" :)
-
xt5 3.04um a74 4.92um 4 or 10 x microscope objective min 5.0 um as they are effectivly f16 so xt5 would be useless for me.
smaller pixels cannot record finer detail period. -
-
checkmate 😎😁
-
Nope 😄, you have been proven wrong again 🤣😎
-
It’s definitely a moire beast compared to others.
-
What do you believe PDR proves? I have no problem with people creating sensor metrics, but you must interpret them in light of what they are actually capable of saying. PDR is not a measure of noise performance, relative to exposure. It is a measurement of how far below the highest number in the bit depth of the raw Bill's standard of minimal image-level SNR lies. It has no direct indication of noise vs exposure. DR and PDR at higher ISOs than base are fluff, with no substance. You can modulate the PDR or DR with the same sensor at high ISOs just by changing the amount of headroom in the raw, using the exact same sensor, just like Canon does with HTP vs non-HTP. If Bill graphed Canon's HTP measurements, instead of the non-HTP ones, Canon would come out as king of PDR or DR at ISOs above base, even though HTP is slightly noisier (2x the post-gain read noise) than non-HTP.
-
With such a high threshold of SNR, the metric is heavily weighted towards photon noise and is most relevant to those who don't want to see noise at all; not those who work in very limited light where there is lots of read noise guaranteed. Even when that is understood, though, there is still no direct connection between PDR and exposure-referred noise, because the headroom range is undefined. It can give "true negatives", though, as a very bad PDR for a specific camera at higher ISOs probably does mean either a lot of noise or insufficient headroom for normal exposure. A very good PDR at high ISOs, however, proves squat about noise because it may be due to extra headroom, alone.
-
Perhaps you meant, "AOTBE", 100% pixel levels views generally look softer and noisier with higher pixel densities, but that is actually a good thing if you want the most malleable, artifact-free captures. Proper sampling is not a defect or artifact..
