Announcement

**thomthom** · 21-08-2008, 01:47 AM

Re: Enviroment Mapping

Also, is one of the mapping types to be preferred over the other?

**dalomar_cg** · 21-08-2008, 07:29 AM

Re: Enviroment Mapping

I guess I'll do a whole explanation of what each is and the pros and cons of each. To start off, the challenge of HDRI and Environment photography is to capture the whole 360 degree environment. Thats pretty darn hard if you think about it, so all these different mapping types are different ways to solve the problem. As always, there are pros and cons and trade offs to each, so the pros of some mapping types lean on the production side, and others lean on the usage side. Basically we're looking at 4 categories when "classifying" a given mapping type; Usage, Creation, Distortion, and Efficiency. About the only one of those categories that need further explanation is effieciency, which in this sense is talking about the amount of the image that is actually used for the projection in relation to how much of the image is not used. Why? because it winds up effecting the amount of memory that is used for the map. Now on to the good stuff

Spherical
This is by far the easiest mapping type to understand, as its the one we're most familiar with. Think of the map of the world...this is actually a spherically mapped image. This basically wraps an image (that is generally the proportion 2:1) around a sphere, which keeps the horizontal lines parallel, but the verticals converge at the north and south poles.

Pros - These are very easy to understand from the point of a user. Also, every pixel in the file will be mapped to the environment, so its very efficient

Cons - They have quite a bit of distortion at the poles, but the actual visual appearance of that distortion depends on how well the image is prepared. These images are possibly one of the hardest types to prepare, since they require a number of different shots to prepare correctly.

Cross or Cubic
Cross mappings (vertical or horizontal) are another good mapping because they are pretty easy to understand. Basically unfold the 6 sides of a cube and you've basically got the cross format. The horizontal and vertical forms are just differences in how the cube is actually unfolded, but if you were at the center of the cube looking north, the face that you would be looking at is always located at the center of the cross. This means that the practical difference between horizontal and vertical cross mappings is how the face behind you is connected to the other faces. In the case of horizontal mapping, its connected to the face on your right, and in the case of vertical mappings, its connected to the face beneath you.

Pros - Again, very easy to use and with 2 secs of photoshopping you can go from one form to the other. The amount of distortion is significantly limited, which means that the image will be sharper. Also, each pixel are very close in size to each other, which contributes to the maps being sharper.

Cons - This format is very inefficient. Half of the pixels are not used for the projection, so this will take up a lot more memory.

I don't know to many details about how these maps are prepared, so I won't put those as positive or negative (although I don't imagine that preparing them is too hard)

Mirror Ball
These look like a circle on a black background and are kinda the "every man's" format. Why? they are deafly easy to prepare, require very little specialist equipment, and can be shot with fewer camera positions. The concept behind the mirror ball is that you have a mirror ball that is directly in front of (and level with) the camera. The ball will reflect all of the environment, except the portions that are directly blocked by the ball. This is true, but not all reflections are created equal. The reflections on the edge of the ball are significantly more distorted, and also described by fewer pixels than middle of the ball. And to boot, most of the scene is described along the edges of the ball. The last tripping point, is that with one picture you'll have a usable map, but you'll also have at least the camera (if not also the person) in the picture. To "remove" the camera and person, as well as add information to the distorted areas, you'll need to take shots from 2 angles and combine them with special software. Most amateurs don't take these steps leaving a person in your reflections.

Pros - They are very easy to prepare, and therefore you can get you're hands on a wide variety at next to no cost. They are not "perfectly" efficient like the spherical maps, but they are pretty good (~79% efficient) and this efficiency is with fewer pixels than the spherical maps (typically).

Cons - They have massive distortion, do not completely describe the scene, and are notoriously under prepared. In terms of usage, visualization of how the map describes the scene is "dicey", and the distortion may be so significant at points that the reflections are useless.

Angular
The last and least distributed format. Visually, these maps actually look almost identical to a mirror ball. Unfortunately, they are typically worse for a number of reasons. To start off, angular maps are just an image taken with a fisheye lense. Therefore, you'll only get 180 degrees of description and that cuts of 1/2 the information that you need. As with mirror balls, there is significant distortion with angular maps. The disappointing thing is that this distortion is typically right where you'll need it. A properly prepared angular map actually has the camera pointing vertical rather than horizontal like the mirror ball as to capture the 1/2 of the scene that is above the horizon line. Typically, the area that is about 20 degrees from the horizon line is what will be viewed in reflections the most, and thats where the most distortion is.

Pros - only the efficiency really.

Cons - Distortion and usage are about the same if not worse than mirror balls. The production is significantly more prohibitive because it requires a fish eye lens, which are very expensive (and not to useful really).

So hopefully that explains some more about the mappings, what they are and how they work. I kinda painted a bad picture for mirror ball mappings, but they really aren't that bad. Angular mappings you're not very likely to come across, so don't worry about them too much.

Personally, I like using spherical maps because they are easy. If I had the choice between all four mappings, I would probably try to use the cross version if I had the memory to spare. If memory started getting thin, then going from cross to spherical would be the first thing I'd do.

HTH

**thomthom** · 21-08-2008, 12:31 PM

Re: Enviroment Mapping

Wow! That was a great explanation!

This makes thing much more clearer.

Btw, is it normal that with physical camera that I have to bump the intensity by the ten's of thousands? With the HDRIs I got I usually have to set the background map to 20000~25000 and the GI to 30000~50000. Or does my HDR simply not have the light range?

**dalomar_cg** · 21-08-2008, 01:40 PM

Re: Enviroment Mapping

No problem...this has been one of those things I've been meaning to get down on paper, so consider it a first draft.

When talking about hdri and the physical camera, you have to understand HDRIs are not prepared to correspond to physically based intensities. HDRIs are still primarily visual, and even though the do have significantly more information than LDR images, they are still limited. Technically an HDR "could" have physically accurate levels, but the image would be white and utterly useless to humans. Thus, you're going to have to manually adjust the intensity of the image to better match the illumination levels that are needed when using the physical camera.

Keep in mind that the amount needed to increase the map also depends on the exposure level of the scene. By default, and when using the physical sun, the exposure level is low because there is so much light. In cases like that you'll have to make more of an increase than if you had a scene with significantly less light. I've had instances where I've increased the map even beyond 50,000 but I've also had instances where I've been able to use 20-30. It really does depend significantly on your exposure.

**thomthom** · 22-08-2008, 02:31 AM

Re: Enviroment Mapping

Interesting.
Do you adjust the exposure with the camera settings? Or the Exposure controls in the VFB? My personal experience with the VFB exposure has not been that great.
I will do some tests with the camera controls and compare.

**dalomar_cg** · 22-08-2008, 06:32 AM

Re: Enviroment Mapping

I always try to get as close as possible with the camera settings. This is because V-Ray will figure out the sampling based on that illumination level and if its off, it can either oversample or undersample the image and neither one of those is good.

Proper exposure is very important, and if you would like to see what I mean, go through the following test....Set up a scene that has a good range of tone values (good deep shadows, and some fairly bright spots), figure out the proper exposure level, and save the image for comparison. Now change the camera exposure so the image comes out pretty dark. With the VfB Exposure adjust the image so its close to the first one with proper exposure. Now look at the darkest point in the scene. There's most likely some artifacts or more grain in that area because V-ray will cut off sampling when things get real dark. Lets try the opposite, adjust the camera exposure level so that the image will be very bright (over exposed). Render, but keep a note of the render time. This time, you won't have any artifacts, but the image should take much longer to render. This is because very bright areas will get sampled more by V-ray, thus taking longer for information that is basically too bright for us to see. So through these two tests, you can see that its important for both speed and quality to have the proper exposure set before you click render.

I will use the vfb exposure to get a good idea of how much to tweek the camera settings, so I know approximately how much to decrease/increase the exposure. When not using the physical camera, I will use the vfb exposure, but only if I'm decreasing the exposure.

**stefanq** · 22-08-2008, 06:54 AM

Re: Enviroment Mapping

This it's very smart! Thanks!

**thomthom** · 22-08-2008, 07:02 AM

Re: Enviroment Mapping

Thank you very much dalomar for all this information. It's been very useful. I have a much better understanding of how it works now.

**nomeradona_cg** · 22-08-2008, 08:25 AM

Re: Enviroment Mapping

damien, what a joy to read your explanation.

Announcement

Enviroment Mapping

Enviroment Mapping

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment