p.4 #1 · Utility of dynamic range beyond a certain point
Lee Saxon wrote:
I understand why DPR and others test dynamic range in terms of examining the shadows in a severely pushed exposure, but IMHO that (or, relatedly, doing some kind of HDR tone-mapping, which OP alludes to disliking [as do I]) is not the real-world use case for extra dynamic range. The real world use case is being able to Expose To The Right as much as possible--to avoid *needing* to push shadow exposure, aka to capture as much clean data in the shadow areas as possible--while leaving the highlights not blown out and able to be pulled back down into non-HDR range. My last two cameras have been the D850 and GFX 100, and the difference in their ability to do that has been meaningfully useful. Now, sure, the A7r5 is somewhere between those cameras in DR, so you're not looking at as much of an upgrade, but I still sometimes use grad NDs and exposure bracketing at times with the GFX 100 so I'm a big proponent of there not being such a thing as too much dynamic range....Show more →
Agreed, which I why I recently started a thread on Dpreview arguing for a change to their "Exposure Latitude" methodology that factors in the highlight headroom the camera's default middle-gray metering / raw ADU placement, which is the information an ETTR strategy would use. Here's a link to the thread:
p.4 #2 · Utility of dynamic range beyond a certain point
sbay wrote:
That's really interesting, I assumed the opposite that prints would have lessor dynamic range.
I think the reason a low dynamic range print/painting can look really good is that our eyes are adaptive. We adjust to the brightest white and blackest black within our visual field of view.
Once I was exhibiting a matt image of mine and placed it right next to a glossy. That didn't turn out well as the glossy made the matt image look flat and horrible, but by itself it was fine.
I think that part of what might have given you that impression about the relative dynamic ranges of paintings and photos is the way in which paintings can convey color that has more depth and luminosity than photographs do. For one thing, the colors in paintings are literally 3-dimensional because they are composed of brushstrokes that are raised above the surface, however slightly. So, they have some of the qualities of a solid object.
Also, traditional oil painters work with layers of colors that a viewer looks into and through. They start with a chosen color for the underlayer, which covers the entire canvas, then they build up layers of substantive paint over the underlayer (which colors everything) and then multiple colored glazes over the paint that the viewer looks into and in the combined effect of all the layers. This is how Vermeer gets beautiful color shadings in his shadows and highlights as well as in the middle tones. If you look at close-ups of Vermeer's painting, you see multiple colors and shades occurring together but separately in very close proximity, virtually never a single color by itself. This is the effect of the multiple layers overlapping each other to different degrees of overlap and transparency, some thin and some slightly thicker. The image is built of all of these multiple layers.
And because the glazes are semi-transparent to a greater or lesser degree, light can enter into them and then reflect from the deeper layers. With Vermeer's paintings, the light often seems to come from within the painting.
I think there must be ways to do something at least partly equivalent with photographic prints, but I am still not skilled enough in post-processing to approach it.
p.4 #3 · Utility of dynamic range beyond a certain point
snapsy wrote:
Hmm, in my last reply I shared a PTOP link I generated showing the a7rVI's PDR advantage is from its lower read noise. I also previously described how DR/PDR is a full-range measurement and that shadows vs highlights is only a metering/exposure construct, so not sure what to make of your latest reply.
I believe you presumed that that post was about the A7RVI, while the post didn't mention this particular camera at all (I think this was Dave's problem with the post as well).
Regarding the A7RVI, I am aware of what you refer to, as I noted in post #4 on p.2 of this thread: "the PDR 12.55 of the A7RVI... is larger than that for A7RV (11.69), yet this is not achieved by capturing more light with the A7RVI. To the best of my understanding, this better PDR of A7RVI is the result of applying selectively a lower gain toward the "shadow signal" from the sensor."
When dealing with the general (disconnected from the specifics of the A7RVI) question on whether a larger dynamic range is useful ("Utility of dynamic range beyond a certain point", as in this thread title), my view is that it is useful, when it is achieved by increasing the ability of a camera to capture more light. Thus, the post that you refere to in "not sure what to make of your latest reply" was expressing my liking of the ability to capture more light with the extended dynamic range (such as that of the GFX100S II, for example).
p.4 #4 · Utility of dynamic range beyond a certain point
Cliff L. wrote:
I didn't read the entire thread, but the impression I got is that everyone is talking about dynamic range at a single point, typically the camera's base ISO (where most cameras are now "good enough"), but there's more to it than that.
What I'm more interested in is how well the camera's dynamic range holds up as ISO increases - since you can can now easily deal with high ISO noise in post-processing, but you can't bring back dynamic range that isn't there. That's where the real advantage of a sensor with high dynamic range lies.
The usual case when sensor A has a better dynamic range than sensor B is at the base ISO, or not far from that. This is because only at the base ISO sensor A can capture more light than sensor B (for example: the GFX100S II at ISO 80 vs A7RV at ISO 100).
At high ISO, the same two sensors should have practically the same dynamic range under the conditions of photographic equivalence (for example, GFX100SII at ISO 12800 and A7RV at ISO 8000), for the reason that these conditions ensure the same amount of light reaching the sensors.
All of the above, naturally, assumes that the camera noise is not reduced digitally (e.g. no denoising applied in raw, or no dual/two-level gain is used) in one of the two cameras.
At high ISO, there's relatively little light reaching the sensor; thus, the signal in the signal-to-noise ratio is small, hence the dynamic range is intrinsically low. The only way to improve on this is by reducing the noise. I am not aware of a consumer-oriented camera that has this advantage of significantly low noise at high ISO. If there is some - I would be interested in knowing that.
p.4 #5 · Utility of dynamic range beyond a certain point
RoamingScott wrote:
This is a good point that has been mostly glossed over. We are already well into the range of DR on modern cameras that only benefits the photographer in the editing room...your final JPEG that gets posted to Fred Miranda or IG or Flickr won't have as much range, and if you print, you're already in 3x overkill territory.
In practical terms, regarding how photography has been done for ages, we have what we need now. Unless some new form of taking and consuming photos emerges, there's no urgent "need" for more DR, just the desire of the pixel peepers and people who can't otherwise make an interesting image....Show more →
I fully agree with "We are already well into the range of DR on modern cameras that only benefits the photographer in the editing room". I think this is a correct answer to what is the "Utility of dynamic range beyond a certain point". In other words, the added dynamic range is of interest to those who are shooting at the base (or close to) ISO, editing raw files, and dealing with the extremes (shadows and highlights) in the raw files.
On the other hand, "your final JPEG that gets posted to Fred Miranda or IG or Flickr won't have as much range, and if you print, you're already in 3x overkill territory" is irrelevant(!) in a discussion of the dynamic range of a camera.
The camera dynamic range and the bit depth of a container (e.g. jpg) used to fit the visual data are unrelated parameters. There is no need for the final product (jpg, print) to have a wide dynamic range for viewing, as this has been aptly illustrated by the example of painted art. However, a large dynamic range in a camera is needed to capture as much of the the visual content as possible, prior to fitting it into a smaller digital container jpg.
I don't find this useful when a discussion of the camera dynamic range is conflated with the discussion of the bit depth of the final product in print or viewed on an electronic device. Also, it is not useful to apply the terminology "dynamic range" to painted images. One can measure the tonal range of a painting, but referring to that as the "dynamic range" is confusing at best.
