Tonality, Noise and Micro 4/3

Started 3 weeks ago | Discussions thread
ForumParentFirstPrevious
Flat view
knickerhawk Veteran Member • Posts: 5,627
Tonality, Noise and Micro 4/3
12

In another thread on this forum there is some discussion about "tonality." Some argue that larger formats produce images with better tonality. Sometimes other terms like "creaminess" or "smoothness" or "richness" or even just plain "color" are used to subjectively describe a desirable image attribute that seems to be more present in larger format images. The question I explore here is just what is this ineffable quality. Why does it (arguably) seem to be more frequently and satisfactorily present in images produced by fullframe cameras compared to mFT cameras?

Tonality is not a particularly well defined term but it clearly has a lot to do with lightness/darkness of elements in an image and especially with how image elements range or ramp from darker to lighter values (or vice versa). The term "tone" is quite often used in conjunction with "curve" or "contrast" to describe how an image element progresses from darker (shadowed) areas to the lighter (better lit) areas. It's usually regarded as a desirable thing when the progression from darkness to lightness is achieved either linearly or logarithmically in the same ways that we directly observe things that are not uniformly lit. Sudden jumps or visible "steps" or "clumps" are perceived as wrong or bad (unless, of course, they accurately reflect objects that are not uniformly lit), hence the association of adjectives like "smooth" or "creamy" with the term, "tonality." And where there should be uniformity and consistency due to even lighting, neither large scale aggregations nor small scale grains of variability should be visible. We expect richness and solidity in the tonality of uniformly lit elements.

With that background in mind, let's consider the reasons why an image might not exhibit this desired ramping and uniformity. There are several possible culprits in digital photography:

  • Image noise
  • Bit depth
  • Coding errors/limitations of some sort (e.g., use of lossy algorithms to reduce file size or to resize the image display)
  • Differences in interpretations of the data (i.e., different profiles, digital filters, etc.)

Feel free to comment if you believe there are other causes of loss of desired tonality, but these four categories seem to me to incorporate most potential culprits. My interest here is identifying the major potential culprits directly related to format (i.e., sensor size) and how format size becomes a factor. Since bit depth, coding errors, and camera/converter profiles are not inherently related to format, I will not explore them in great depth here. However, regarding bit depth specifically, I will just note that most discussions on this and other forums about tonality relate to digitally displayed images. Since most displays (until recently at least) have not exceeded 8-bit depth and all images shared and discussed on these forums are 8-bit, it's reasonable to assume that any difference between 12-bit and 14-bit image capture is not a major factor in differences that are clearly visible in images output to and displayed at 8-bit depth. Either coding errors or noise or some other unknown must be magnifying any differences buried at 4 levels of less significance than captured in 8 bits if, indeed, bit depth is to be a factor. Suffice it to say that I haven't personally experienced (as a prior owner of cameras that can shoot at either 12 or 14 bit depth) or have otherwise seen compelling evidence that the difference between 12-bit and 14-bit capture plays a meaningful role in anything close to normal capture, editing and viewing conditions.

Ok...preliminaries aside, below I illustrate how noise plays a visible role in what I, at least, consider to be "tonality." If the differences you see in the comparisons are what you would also characterize as tonality-related, then I believe the burden has been shifted such that anyone arguing for causes other than noise, and any such doubters should offer their own clearly visible evidence of how the other factor(s) also/alternatively play a role. Once the evidence is in, we will be in a good position to discuss how (if at all) format size, as such, is the primary determinant of tonal quality in images. Let's get started with my submission of evidence...

Below are three variations of a gradient I've generated in Photoshop. They are all totally synthetic, meaning they are completely computer-generated and, thus, are not associated with any particular camera or format. To me, the ordering (top-to-bottom) is worst-to-best with respect to "tonality." Also, the differences I see are quite similar to the differences I see in real images in which someone claims a difference in the tonal quality (creaminess, smoothness, color, etc.) Do you agree?

For best evaluation of the differences click and view in the DPR Image Viewer.

And below is the same comparison with additional tone curve adjustments applied in Photoshop. To me, the curve adjustments, in addition to obviously changing the tonal ramp/contrast itself, also accentuates the difference in the quality of the tonal transitions. Do you agree?

For best evaluation of the differences click and view in the DPR Image Viewer.

Next consider the two following examples. Obviously these crops come from real images of shots I took (fyi, they're of the famous Haystack rock outcrop off the coast of Oregon). To me, the right-side version has superior tonal quality. Do you agree?

For best evaluation of the differences click and view in the DPR Image Viewer.

And here's the same two crops with the same tone curve adjustments as were applied in the synthetic example above. Again, I see accentuated differences in the quality of the tonal transitions. Do you?

For best evaluation of the differences click and view in the DPR Image Viewer.

At this point I want to solicit comments and feedback on the comparisons and my other claims here. I'll chip in later with an explanation of exactly what's going on in the images and how I think this explains not only the impact that noise plays (it should be obvious that there are differences in noise in these comparisons), but also how this relates to sensor size.

Please comment!

ForumParentFirstPrevious
Flat view
Post (hide subjects) Posted by
ForumParentFirstPrevious
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow