Can someone explain PhotonsToPhotos graphs to me?

mike3996

Mu-43 Veteran
Joined
Jan 17, 2018
Messages
359
Location
Finland
DxoMark doesn't have the latest gear so PhotonsToPhotos.net has become pretty popular among gearheads.

But I don't understand what exactly is going on. What do the results shown in "Photographic Dynamic Range Chart" show what exactly? SNR? Sheer dynamic range at the ISO? Is there a (loose) correlation between these findings and what can be found on Dxomark's graphs?

Do the "shadow" improvement charts tell how pushable are the shadows at given ISO?

The explainers that the site has don't really "ELI5". I get that "big number good". I even get that the graphs are logarithmic on both axes. M4/3 cameras are often placed within one stop from FF cameras. Does this fight against the paper theory or is the math indeed somehow less pixel-peep oriented so that the results reflect more of the real world?
 

mike3996

Mu-43 Veteran
Joined
Jan 17, 2018
Messages
359
Location
Finland
  • Thread Starter Thread Starter
  • #3
Crickets :)

I gave a yet another try to search the subject. Now I found an hour-long podcast, a conversation with Jeff Harmon and the site owner Bill Claff.

https://phototacopodcast.com/photographic-dynamic-range/


So I gave it a listen on my lunch break. What I picked up from it, plus some personal takes.

Firstly, PDR is indeed a measure of dynamic range in the sense a dictionary would define it. It's the difference of the whitest white (just before clipping) and the darkest black (where the camera can still maintain a reasonable SNR) expressed as log2 value.

The "photographic" in PDR comes from the fact that Bill incorporates circle of confusion to the numbers as a factor. The theoretical numbers are what pixel peepers can test to but having COC as a nice way to use mathematics to evaluate the overall picture most likely makes for more realistic numbers.

My take: in DXOMark you can choose between two sets of numbers: Screen (pixel peeping) and print (files normalized to 8 megapixels before measuring). This "print" normalisation kind of has the similar idea. In any event, packing more pixels means the COC changes and that generally reflects on the graphs in showing same-size, high-resolution sensors performing better. (Because you are not supposed to be pixel-peeping!)

Because a PDR chart after all is a DR chart, DxOMark's "Landscape" values are the ones that somewhat correspond with PDR numbers. Because of different math, factors and so on, there's a pretty constant difference of 3 EV between the measurements.

The discussion towards the end tangented on the subject of low-light sports shooting and how measured PDR might reflect on noise and image clarity----sadly Bill sidestepped the matter so it was left unanswered. I think this matter can be further researched independently of PtoP.

What did come up, is the fact that much like noise tolerance in general, what is an acceptable PDR is a matter of individual taste. Bill suggested the following: study the high-ISO images of cameras you know well, then check what PDR values these images correspond to. If you for example think that Panasonic GX85 at ISO 3200 is usable in some light, you can check the value being PDR 5.25. Coincidentally Panasonic G9 maintains the same PDR of 5.25 at ISO 6400.
 

Mark73

Mu-43 Regular
Joined
Sep 13, 2019
Messages
62
I was about to refer you to the above podcast myself.

The site is an excellent resource and goes much deeper than dxomark.
 

Danny_SWE

Mu-43 All-Pro
Joined
Apr 30, 2013
Messages
1,507
Location
Sweden (Gothenburg)
Check out this astro photo guide, he has some ideas of how to use the SNR chart.
(but it was difficult adapting to m43, I landed at 2500 ISO for my E-M5.3)
 

Latest posts

Links on this page may be to our affiliates. Sales through affiliate links may benefit this site.
Mu-43 is a fan site and not associated with Olympus, Panasonic, or other manufacturers mentioned on this site.
Forum post reactions by Twemoji: https://github.com/twitter/twemoji
Copyright © 2009-2019 Amin Forums, LLC
Top Bottom