Talk:Oren–Nayar reflectance model

On notability question
I'm not convinced of the notability of the "Oren–Nayar diffuse model." An anon IP editor claims that the one independent book ref is sufficient, but he has not said whether he has seen that book, nor has he offered to tell us what it says, which I do hereby request. It is rare to accept a topic as notable on a single mention; even more so when nobody can tell us what they mention says. If the source says "known as" or "referred to as the Oren–Nayar diffuse model" or something like that, then maybe we can accept that as sufficient evidence of notability, assuming the author is independent of Oren and Nayar. But if it just advances the name itself, or says something like "the model of Oren–Narar", then we have business annointing that as a name. That's my only concern. Anyone have that book handy? Dicklyon 01:08, 25 October 2007 (UTC)

Further to this, I just checked on who put the ref. It was the guy who created the original stub, who hasn't been on wikipedia since the day he did so; he gave no indication of what point in the article that ref was intended to support. Nobody has added a ref or any useful info since; it looks like some are findable, so I'll work on that. Dicklyon 01:21, 25 October 2007 (UTC)

OK, no book refs for "Oren-Nayar diffuse model", but plenty for "Oren-Nayar model", so I added a couple of those and removed the tag. I think the topic name with "diffuse model" is pretty lame, because the model is not diffuse, and this topic doesn't say it's about reflection, and serious writers don't call it that. I suggest we move to either Oren–Nayar model or Oren–Nayar reflection model. Comments? Dicklyon 03:36, 25 October 2007 (UTC)


 * It's specifically a model of diffuse reflection, thus the name. It's the "model" that is superfluous in a search, searching for "Oren-Nayar diffuse" finds far more hits. If you don't even know what this is, why are you trying to judge its notability? --88.195.54.90 16:30, 25 October 2007 (UTC)


 * I do know what it is, since I looked it up and read about it; and judged it to be notable, since the article contains citations to reliable sources about it (which I added). What behavior are you trying to criticize here? Have you done anything to help? Dicklyon 23:26, 25 October 2007 (UTC)


 * And I find only 3 book hits for "Oren-Nayar-diffuse" compared to 13 for "Oren-Nayar-model". I realize it's a model of diffuse reflection.  But what title should the article adopt?  A common one?  Or a more rare one?  Should we attempt to be parallel with Phong reflection model?  or maybe Blinn–Phong shading model?  Dicklyon 23:30, 25 October 2007 (UTC)

Torrance-Sparrow model
Why does this line link to an article that doesn't even mention Sparrow? The closest is has is the cook-Torrance model. Is this a problem in this line or the lined article? —Preceding unsigned comment added by 66.64.16.58 (talk) 20:40, 4 September 2008 (UTC)

Wrong pictures
Picture showing Lambertian Model is in linear color space, instead of the default for the Internet (non-linear) sRGB color space. So the picture (on typical monitor) looks completely different from an actual Lambertian Model rendering. Very weird for me that no one noticed it before. 83.7.234.61 (talk) 18:19, 19 December 2009 (UTC)

In the last picture you cant really tell which arrow is pointing to the "real" graph and witch one should be pointing to the Oren-Nayar graph. —Preceding unsigned comment added by 89.247.92.143 (talk) 02:29, 23 October 2010 (UTC)

Wrong definition of σ
σ should not range between 0 and 1 as said here, but from the Oren-Nayar paper (ref [1]), σ² is the variance of distribution of θa, the angle between the real (rough surface) normal and the mean normal (flat surface). Therefore σ should be in degrees. This is confirmed by other papers found in the external link mentionned.

Also, The picture with the sphere is labelled with σ=0, 0.1 and 0.3, whereas it is shown with σ=0, 20° and 40° in the following paper (found in the site referenced by "external link"): Generalization of the Lambertian Model and Implications for Machine Vision, S.K. Nayar and M. Oren, International Journal on Computer Vision, Vol.14, No.3, pp.227-251, Apr, 1995. — Preceding unsigned comment added by 83.205.191.25 (talk) 17:39, 12 November 2011 (UTC)

The paper cited below gives results (text and plots) with sigma expressed in degrees (this is more intuitive for the reader). However, the formula should be applied with sigma expressed in radians. It is intuitive that the coefficients 0.09 and 0.33 would not make much difference in A and B with a sigma expressed in degrees and squared. Implementing the formula (in radians and degrees) and comparing it with the plots shown in the paper (for instance the plots in figure 14) shows that sigma in radians produces the good curve. Since the maximal possible facet slope is pi/2, sigma cannot excess this value. — Preceding unsigned comment added by Eheitz (talk • contribs) 18:56, 4 January 2012 (UTC)

Image Gamma
All rendered images appear to be gamma incorrect and thus are null and void in a realworld comparison. Unless one set of images, real or rendered, are corrected to either be gamma correct or gamma incorrect.... what I mean to say is both sets of images should have the same gamma to be a valid comparison. — Preceding unsigned comment added by 80.62.167.246 (talk) 16:43, 19 April 2012 (UTC)


 * I fully agree with that. Ideally, all images should of course be gamma correct, not just have the same gamma, as an image with an incorrect gamma still will be displayed non-photorealistically on the screen. But if several images have the same gamma, even if an incorrect gamma, you can still compare them to each other, which is somewhat useful.


 * So, does anyone know whether the images in this article are gamma correct? If not, I think the image comparing a photo with an image generated with a Lambertian model should be removed from the article as it becomes completely irrelevant, and even misleading. —Kri (talk) 13:59, 7 June 2015 (UTC)


 * I added to the top of the page to notify readers about this potential error. —Kri (talk) 22:58, 8 June 2015 (UTC)

My apologies. A scientific paper, due to size constraints, cannot provide all possible low-level details. The first image on wiki looks same as on page 1 in paper. The fact that authors did not specify some details about images, is not enough to call the images misleading.


 * My judgement is that the computer generated images may very well be gamma-incorrect (unless the author of the scientific paper had knowledge about gamma-correction); in that case I do consider the illustrations to be a bit misleading, especially those that make comparisons with photographs (since those are most probably gamma-correct and thus have gamma that deviates from the computer generated images). The fact that the images are featured in a scientific paper is not a guarantee that they are gamma correct. If on the other hand all images are gamma correct, I don't consider the illustrations to be misleading. In either case, I think the -tag should be preserved until the "specify"-tags have been resolved; the presence of this tag only specifies that the article be misleading, not that it actually is. —Kri (talk) 19:56, 28 July 2015 (UTC)

Light Scattering is mainly a single event effect
The Scattered light is considered in the literature as a diffusive light, light that passed a number of scattering events before it left the scattering material. Diffusely scattered light must obey Lambert's Cosine scattering law. In the case of unidirectional light scattered backward from a surface of a sphere, the meaning is maximum scattering intensity in the middle of the sphere surface, and a decline to zero toward the periphery by the cosine law. The full moon looks uniform and people continue to assume that the light is diffusely scattered from it. More than that. The nearly uniform sphere image is common to all the planets and their moons, including the earth as observed from the moon. Out of thousands upon thousands of true photos, there is no single true photo that obeys Lambert's Cosine law. The only photos that do obey the law are rendered photos, photos that are at least partly simulated. Contrary to all that, if the scattering is assumed to be mainly a single event, then all the scattering dipoles are directly stimulated by the light radiation on the illuminated scattering material. Then scattering by them must be coherent, and then the full moon and all the other illuminated bodies, with similar illumination geometry, must be uniform, at least approximately. The full moon tells us that single event scattering is dominant. Maybe with small corrections of multiple scattering. Why is the single event dominant? It seems that the effect is geometrical and statistical. If we consider one event scattering, two event scattering, multiple event scattering, then the event probability will decline with an increasing number of scatterings. The single event has a probability of at least 50% and it is the strongest event. Nearly all the background that surrounds us is a singly scattered light. A true diffusely scattered light is rather rare. Urila (talk) 01:59, 13 May 2020 (UTC)urila

Blue Marble
"Blue Marble", NASA's name for the full earth photo, shows uniform average image density from the center of the image towards its periphery. More than that: The earth's image includes large areas of gas-phase – clouds, liquid phase – oceans, and solid-phase – land, and the uniformity is nearly true for each phase separately. This feature is not compatible with Lambert's cosine law, and the Oren-Nayar model is not relevant to it. One photo – a thousand models. Urila (talk) 05:59, 17 October 2020 (UTC)