|
Post by ampingu on Apr 24, 2009 11:41:15 GMT -4
Id suggest that its exactly because apparent brightness does vary a lot with viewing angle, even on a flat surface, that you see these highlights. ie, the surface is not a 'lambertian reflector' Far from it, and you can get an impression by observing the moon from earth. A full moon is much more than twice as bright as a quarter moon, and fairly evenly bright across the disc. This shows that the light reflected from the moon peaks in a direction straight back to the source. I dont understand your far from it comment - you havent replied with anything that conflicts with the quote. Anyway, if I understand correctly, these are all very simple concepts of the kind we're exposed to in everyday life. (for example, in shining a spot on a wall from a makeup mirror), regardless of the amount of jargon we're throwing around. I suppose it doesnt hurt to be thorough. If youre considering a full moon, of course, given that the moon is spherical, most of the light will be reflected back in the direction it came from. the question of whether or not we see all that reflected light, though, depends on our viewing angle. see? This is slightly removed from the 'wet car park' (or specular highlight) effect we see in the pictures, but the concept is the same. (forgive me if the diagram seems patronising. Just wanted to be clear).
|
|
|
Post by gwiz on Apr 24, 2009 14:59:23 GMT -4
I dont understand your far from it comment - you havent replied with anything that conflicts with the quote. I meant the lunar surface is far from being a Lambertian reflector. To also spell it out, with Lambertian reflection, a full moon would be twice as bright as a quarter moon, because you can see a reflector that is twice the size. The fact that the full moon is much brighter than that, perhaps ten times as bright, confirms the non-Lambertian nature of the lunar surface reflection. It's nothing like a specular highlight either. The full moon doesn't have a highlight, it's evenly bright, which means that even the edges of the illuminated part are reflecting a lot of light back towards the sun. The phase variation is more like the rear reflector on a bike, but obviously not as extreme.
|
|
|
Post by ampingu on Apr 24, 2009 16:50:32 GMT -4
"I meant the lunar surface is far from being a Lambertian reflector."
ah, with you.
"It's nothing like a specular highlight..."
you dont agree the moon could be described as having a 'specular nature'?
im out of my depth with some of this optical jargon, to be honest, so im happy to concede. semantics arent important.
"The fact that the full moon is much brighter than that, perhaps ten times as bright, confirms the non-Lambertian nature of the lunar surface reflection"
i think the varying (macro level) albedo you describe is better explained by the topography of the surface (as the other guys here mentioned), with varying phase angles altering the total shadow cast by mountains and craters etc, giving a net reduction in apparent brightness.
in any case, the phase of the moon is straying a bit from the topic (even if its interesting). im happy that its all covered, long as you guys are.
thanks again.
|
|
|
Post by JayUtah on Apr 24, 2009 19:09:09 GMT -4
you dont agree the moon could be described as having a 'specular nature'?
No. The Moon is clearly a diffuse (Lambertian) reflector, not a specular reflector. However, that dichotomy is too coarse a description of what properties combine to create the final impression of lunar surface brightness.
|
|
|
Post by George Tirebiter on Apr 24, 2009 23:09:28 GMT -4
ampingu, to put it in CGI terms, the simple model that the surface of the Moon most closely resembles would be the Oren-Nayer diffuse model (with non-zero roughness). The change in brightness with viewing angle is caused by small-scale self-shadowing of the rough surface (which the Oren-Nayer model approximates), not specular reflection.
|
|
|
Post by JayUtah on Apr 25, 2009 1:18:16 GMT -4
Backscatter can be modeled on top of that with a generalized Torrance-Sparrow model. But we haven't gotten there yet.
|
|
|
Post by gwiz on Apr 25, 2009 5:04:04 GMT -4
i think the varying (macro level) albedo you describe is better explained by the topography of the surface (as the other guys here mentioned), with varying phase angles altering the total shadow cast by mountains and craters etc, giving a net reduction in apparent brightness. The shadowing effect isn't confined to macro-level topographic features, it works at all scales right down to the surface irregularities. That's why the same phase angle effects can be seen in Apollo pictures and in eyeballing the moon from earth. A smoother moon would be closer to a Lambertian reflector, but there is a second factor that would still give the enhanced straight-back reflection. The lunar dust contains small glass spheres, the result of the smaller molten particles ejected from meteorite impacts solidifying before they fall back to the surface. These spheres act as retro-reflectors, the cat's-eye effect.
|
|
|
Post by ampingu on Apr 25, 2009 6:03:18 GMT -4
Yes, dichotomy, quite so. The planetary (macro level) surface properties are bound to differ from those seen at the surface level (micro level). So... Are any of you guys experts in this particular area? I need to ask because your analyses seem to conflict a bit. If youre going to outright contradict my assumptions, I feel a need to defend them! (as much for posterity as for the sake of my own understanding). Assuming you dont mind discussing this further with an amateur, Im going to summarise my current understanding: 1./ The moon as a whole. An overall specular nature implies that if you target light at the surface, the odds are that the light will be reflected in a direction conforming with the first law of reflection. Demonstrated here (pinched from wikipedia): The moon has relatively consistent topography across its surface, so you can expect it to conform to the above criteria. Therefore I argue that the moon can be considered to have a specular nature. (Its a sphere)* 2:/ Surface level 'anisotropic variances' (I guess we can call them that): The hot spots visible in the photographs (other than those special case few that show an ellipse caused by the LM landing), are the result of the specular component of the reflected sunlight. Regarding the discussion on varying 'texure', it is the light and dark sides of the grains that make up this texture (on a micro level), relative to your viewing angle, that provide for the specular highlight. Its not an alternative effect - its the process through which the specular highlight occurs. At lunar ground level, its very clear from the photos, that the net surface properties do not amount to a lambertian reflector. I dont believe the different reflection models available in cg are relevant. Specularity is a general concept that transcends these models - with even relatively diffuse modelled surfaces having at least some specular highlight. *I have a reference here: "From repeated observations by radar it has been established that the reflection of radio waves by the moon and planets is primarily specular in character"radio waves are part of the em spectrum, just like visible light - so i think this makes a valid cite. adsabs.harvard.edu/full/1968SvA....12..489A
|
|
|
Post by JayUtah on Apr 25, 2009 13:08:04 GMT -4
Are any of you guys experts in this particular area?
Yes. In the early 1990s taught the subject at the University of Utah, which trained many of the Pixar and Adobe people.
An overall specular nature implies that if you target light at the surface, the odds are that the light will be reflected in a direction conforming with the first law of reflection.
No. If the surface is characterized as specular, it will reflect light away in a direction conforming to a mathematical reflection of the illumination vector through the surface normal. If a surface is characterized as diffuse, the reflected light direction is independent of incident angle; it will reflect equally in all directions, although the intensity of the reflected light will vary proportionally to the cosine of the incident angle.
Real surfaces exhibit properties grossly consistent with a linear blend of the two models when viewed at customary distances. However such a model does not account for all local illumination effects. In the case of the lunar surface, for example, it does not account well for backscatter. The model may be employed at the micromodeled level, as did Torrance and Sparrow, to build up higher abstractions of local illumination.
The moon has relatively consistent topography across its surface, so you can expect it to conform to the above criteria. Therefore I argue that the moon can be considered to have a specular nature.
No. The Moon exhibits almost no illumination effects consistent with the specular model. If it were specular we'd see it as very dark, illuminated only in spots where the surface happened to reflect the sun or Earth light back toward us. Instead we see a uniformly illuminated sphere or crescent. This is exactly consistent with a diffuse model.
In contrast consider the Earth's oceans as seen from space. They exhibit specular properties: they are generally dark except at the hot-spots where they reflect an image of the distant sun.
The hot spots visible in the photographs (other than those special case few that show an ellipse caused by the LM landing), are the result of the specular component of the reflected sunlight.
No. The specular local illumination model predicts brighter apparent surfaces if seen from near the reflection angle derived from incident angle and surface normal. In Apollo photography, the brightest apparent surfaces are seen instead when looking down-sun. The specular model spectacularly fails to predict the illumination as seen in Apollo photos, or in fact any illumination of the Moon. That is because the effect seen in Apollo photos is a global illumination effect considering also the geometric arrangement of objects in the scene.
Regarding the discussion on varying 'texure', it is the light and dark sides of the grains that make up this texture (on a micro level), relative to your viewing angle, that provide for the specular highlight. Its not an alternative effect - its the process through which the specular highlight occurs.
No. You're badly misusing the term "specular hightlight."
At lunar ground level, its very clear from the photos, that the net surface properties do not amount to a lambertian reflector.
Agreed, if by "net" effect you meant the global, macro perceptual effect. Because the texture scale is significant at the viewing scale, the global illumination effect is ad hoc. But nor do they amount to a specular reflector. The specular model is not just preferential reflection -- it is reflection preferential to a specific direction, which direction is not obeyed in Apollo photography. There are plenty of non-specular local illumination models that provide preferential reflection.
I dont believe the different reflection models available in cg are relevant.
They can be, because many are physically modeled. However, you cannot account for all visible effects in a scene solely by local illumination models. You can apply an appropriate local illumination model from computer graphics, but you must also model the surface appropriately and render it from an appropriate distance. Then you will see the same effects as are visible in AS11-40-5903.
Nevertheless the study of local illumination models is rewarding because it teaches the basic behavior of light. Whether one believes any of those models applies to this case is irrelevant; the knowledge obtained in studying the derivation of local illumination often informs a line of reasoning that includes global effects.
Specularity is a general concept that transcends these models...
False. Specularity is quite clearly an effect of local illumination. Your problem is that you are trying to describe global illumination and perceptual effects only in terms of local-illumination concepts and terminology.
radio waves are part of the em spectrum, just like visible light - so i think this makes a valid cite.
No. Emissive, transmissive, and reflective properties vary by wavelength. The Moon is obviously non-specular in the visible wavelengths.
|
|
|
Post by dragonblaster on Apr 25, 2009 14:48:18 GMT -4
Always good to see a post from someone who really knows his stuff, Jay. Concise and clear exposition there.
|
|
|
Post by ampingu on Apr 25, 2009 15:55:32 GMT -4
Yes, agreed - thank you Jay.
"No. If the surface is characterized as specular, it will reflect light away in a direction conforming to a mathematical reflection of the illumination vector through the surface normal."
You seem to be saying that its an absolute term, ie that a surface is either specular or it isnt.
And then you go on to say...
"Real surfaces exhibit properties grossly consistent with a linear blend of the two models..."
With the two models being specular vs diffuse? So you agree that in most real world circumstances, surface properties are only ever 'more specular' or 'more diffuse'?
That sounds like a bit of a contradiction.
I was careful to use the term 'overall specular nature' (I do concede that this terminology is sketchy).
Im happy with the rest of your reply. Thank you for the insight.
"In Apollo photography, the brightest apparent surfaces are seen instead when looking down-sun."
Do you have an example showing a relatively uniform ('texture free') part of the lunar surface showing a 'hotspot' when viewing down-sun? (just for fun - id be interested to see it).
|
|
|
Post by JayUtah on Apr 25, 2009 18:11:23 GMT -4
That sounds like a bit of a contradiction.
No. The property of specularity is well defined mathematically. The property of diffusion is also well defined mathematically. The simplest realistic illumination model performs a parametric blend between these two properties at each point. Depending on the parameters, the result may be illumination which exhibits preferential reflection along the classic specular departure angle, but that is a property of the combined model, not of the specular component alone.
I do concede that this terminology is sketchy.
That's the problem. You're trying to use words that mean specific things to radiometrists but conflating your invented meaning with their conventional meanings.
If you want to say that some surface is "more specular," meaning that the specular component of a more inclusive model predominates, that would be understandable and reasonably correct, and may be more along the lines of what you're trying to say. Where you run afoul of terminology is when you say that specularity itself is probabilistic. That is a property of higher-order models.
Do you have an example showing a relatively uniform ('texture free') part of the lunar surface showing a 'hotspot' when viewing down-sun?
Take care: I didn't mention "hot spots." The general rule is that dictated by phase angle and generally doesn't create discontinuous hot spots within a field of view.
AS11-40-5854 is taken from the cockpit of the LM looking down-sun. -5882 shows an excellent example of intensity that varies strongly by phase angle in an anti-specular fashion. Perceived intensity is indeed preferential, but does not obey the geometry of the specular model.
The series 5935-5927 from AS11-40 show up-sun angles in which a specular-prevalent model would predict greater illumination. But these are somewhat darker, and they exhibit the non-uniform appearance of a textured surface at a large phase angle.
|
|
|
Post by ampingu on Apr 25, 2009 20:00:19 GMT -4
Thanks! And thank you again for putting so much time into this. I know youre not obligated - especially if its only to help with one guys (my) understanding. "You're badly misusing the term "specular hightlight.""Can you give me a concise definition of a specular highlight? Im starting to suspect that my wikipediesque internet sources are a little off. This is (what I would call) a specular highlight, as demonstrated using a phong reflection model on a planar surface in 3d studio max. Is this the same effect seen in my 'wet car park' example further back in the thread? (which does look similar, i think, to the effect seen in AS11-40-5940 - even if its caused by something else) Yes, thank you - thats exactly what I meant by 'overall specular nature'. I should have taken more care Isnt this second quote slightly in conflct with the first? Maybe I misunderstand. Wikipedia (and i do apologise for quoting from wiki) simply documents specularity as: "Specularity is the quantity used in 3D rendering which represents the amount of specular reflectivity a surface has."I read that to mean it is the weight of the specular component of any surface that has properties that are either mostly specular or mostly diffuse. Are you saying that this doesnt apply to an overall model? 5854 and 5882 demonstrate what they call heiligenschein, dont they? I thought heiligenschein was something slightly different. **edit: now that i look, many of the AS11-40 images seem to be taken down-sun (judging by the shadows). Ps: We may have crossed wires regarding the moons apparent brightness as seen from the earth. Id like stick to only the effects at surface level if thats ok. Pps: Its very late! Ill post anyway so that my response is timely, but please forgive me if Ive missed something silly. edit:"You're trying to use words that mean specific things to radiometrists but conflating your invented meaning with their conventional meanings."accepted. im just a casual observer. all this stuff is very interesting - it seems its all a lot more complicated than i first thought. so on reflection (pun intended : ), i dont want to frustrate you any further with my abridged wikipedia inspired interpretations and im happy to differ to your expertise. (thats almost punny : ) theres no need elaborate any further. youve already been more than thorough.
|
|
|
Post by gwiz on Apr 26, 2009 5:43:38 GMT -4
5854 and 5882 demonstrate what they call heiligenschein, dont they? I thought heiligenschein was something slightly different. Heiligenschein's the part of the effect due to the glass spheres I mentioned earlier. By coincidence, the Wiki page on the subject includes an Apollo image as an example: en.wikipedia.org/wiki/Heiligenschein
|
|
|
Post by Kiwi on Apr 26, 2009 5:56:14 GMT -4
Can you give me a concise definition of a specular highlight? Im starting to suspect that my wikipediesque internet sources are a little off. I gave an old one (off the top of my head) at the end of post 13. Specular highlights in photographs before CG always meant washed-out, overexposed areas produced by light sources or their barely-diminished reflections from surfaces such as glass, chrome, or water. It sounds, from what you say, that CG people are misusing the term, which seems a bit silly if they are. The dictionary definition of specular is, in this context: "of, relating to, or having the properties of a mirror: specular reflection." -- Collins (1979)
|
|