Originally Posted by armadillo
Chris, you are free to deny whatever you want. Several posts on this thread have given you references of scientific papers that arrive at limited contrast of the HVS in the range of 100-300:1. I am citing from the following links:
"The huge input range of the human visual system is largely the result of adaptation processes.
As summarized by Walraven and colleagues , several researchers have isolated the response of retinal photoreceptors from adaptation e ects by measuring cell responses to very brief flashes of light. Their measurements indicate that without adjustment by adaptation processes, responses vary only in a narrow range of light intensities covering about two factors of ten, or 100:1."
You expect people to believe you when the precise papers you quote clearly are discussing local
contrast perception limits? Your intellectual dishonesty in quoting these papers is staggering.
How about the first sentences in the abstract of this article:
"High contrast images are common in night scenes and other scenes that include dark shadows and bright light sources. These scenes are dicult to display because their contrasts greatly exceed the range of most display devices for images. As a result, the image contrasts are compressed or truncated, obscuring subtle textures and details. Humans view and understand high contrast scenes easily, \\adapting" their visual response to avoid compression or truncation with no apparent loss of detail."
Or Page 3:
"Synthetic and real-world scenes often contain very high contrasts. For example, a scene with dark shadows, visible light sources, caustics or specular reections is likely to contain contrasts as large as 100,000:1 or more."
Or Page 4:
"The ease with which humans view high contrast scenes suggests that models of visual perception may help solve the problem of displaying high contrast images on a limited contrast display. This paper presents two simple methods inspired by the human visual system. In particular, humans form separate but simultaneous judgments of lighting and surface properties as if the scene were perceived in multiple layers . The lighting layer contains most of the high contrasts while most of the image detail and texture is contained in the layers describing surface properties. The rst method, therefore, compresses the lighting layers of an image and leaves the surface properties unchanged. The second method mimics the directional nature of visual adaptation. Because the human visual system adapts preferentially to available light in the direction of gaze, this method adjusts the entire image for best display of a small neighborhood around the viewer's center of attention."
Or here which precedes the quote that you present:
"However, the response of retinal ganglion cells to large local contrasts is bounded by gradual, asymptotic limits. Signals from retinal cells are dificult to measure, but experiments by Sakmann and Creutzfeldt (1969) and others (summarized in ) have shown ganglion firing rates in the cat approach a fixed upper limit as local contrasts exceed about 100:1, and their plots of firing rates revealed a family of smooth asymptotic curves. Retinal ganglion cells may directly encode the small contrasts (<100:1) caused by reflectance variations in a viewed scene, but the huge contrasts possible at illumination boundaries must drive both ON-center and OFF-center cells towards their asymptotic limits."
This describes localized
limits at a high contrast boundary. Your assertion that the 100:1 limit applies across a scene is completely erroneous and flatly rejected by the very paper you cite.
Or continuing in the paper:
"Adaptation has a strong local character because the human visual system adjusts separately at different locations within a viewed scene or image. These adjustments allow simultaneous sensing of texture and detail in both strongly shadowed and brightly lit regions. As a result, human vision almost never clips" as a camera or display might. For example, trees silhouetted against a brilliant sunset may appear featureless black when photographed or rendered, but a human viewer will see leaf colors, bark textures, and other fine details of the tree if any of them subtends more than a few degrees of the visual field. Local adaptation allows us to recover the appearance of the tree within the scene.
Local adaptation depends strongly, but not entirely, on the image within the viewer's small, central fovea. For example, looking directly at the surface of an incandescent light bulb causes the remainder of the visual field to temporarily appear darker, indicating that the bright image on the fovea depressed perceived intensities everywhere. However, if the bulb is at least 20-30 degrees away from the direction of gaze, hand movements that reveal or block a view of the bulb have little or no effect on the apparent brightness of the rest of the scene."
Or again if it's not clear enough:
"Local adaptation is particularly useful when viewing high contrast scenes because small neighborhoods tend to be much more homogeneous than the entire image. Neighborhoods that include both shadowed and brilliantly lit features will have high contrast, but these regions are usually only a small fraction of the entire image."
"Furthermore, there is a limit to how much contrast can be perceived in a very small neighbourhood of the visual field. ... The threshold at which this occurs, the maximum perceived contrast, is reported to be around 150:1 ."
And again your intellectual dishonest is staggering. The full quote:
"Furthermore, there is a limit to how much contrast
can be perceived in a very small neighbourhood of the
visual field. That is, when the contrast between
adjacent spots on the retina exceeds a particular
threshold, we will no longer be able to perceive the
relative magnitude of that contrast (roughly speaking,
the spot on one side will appear white and the one on
the other - black). If you separate the spots in space,
you will again be able to see their variations in
brightness. The threshold at which this occurs, the
maximum perceived contrast, is reported to be around
Your attempt to take this quote out of context by removing relevant statements from the paragraph is unbelievable.
And yes, this has to do with small angles, because it gets even worse if you leave the foveal area. The black and white patterns you like to cite are irrelevant, because our retina has very limited color vision outside the fovea, because the cones (the color-sensitive photoreceptors) are concentrated there, whereas the rods (luminance detectors) are in the periphery. You simply fail to understand that the HVS has complex spatio-temporal processing that primarily results from adaptive processes within a narrow field of view. The contrast adaptation occurs dynamically as we move the foveal area across limited areas of interest with a huge 3D space. However, within each of these limited areas, the contrast does not exceed 100-300:1. As I said, large, immersive projected images will beenefit from higher CR, but if you look at a small image that is covered by your foveal area, you simply won't be able to resolve more contrast than the numbers the scientific literature gives you. So go ahead and deny, I couldn't care less.
I'm not denying that limitations AT a high contrast boundary are significant. Indeed, I've been saying that the entire time since the very beginning, and you only now are beginning to recognize that these figures are for extremely localized features. Further, color vision is irrelevant here as Darin pointed out. Lastly, we have the ability to fix our gaze at different points across a scene, so an image presented by a display device needs to match our abilities to see large contrasts across
a scene or image, not just in the extremely limited narrow angles of adjacent pixel groups.
The key point is that both of the papers you cite DIRECTLY and unambiguously support my point: ANSI CR capabilities significantly
beyond 100-300:1 are necessary to match or exceed the capabilities of human vision. Indeed, the figures of 100-300:1 are only applicable to extremely limited viewing angles and would be more appropriate to discuss in terms of display MTF and contrast performance at extremely high frequencies, not at the very low frequency of an ANSI Checkerboard pattern.