Jeremy Birn is a Pixar lighting artist whose book "Digital Lighting and Rendering" has become a standard introductory text on the topic of CG lighting. I recommend it as a good first book for someone wondering what to do beyond the default lighting, texturing and rendering a 3D app like A:M presents.

I won't be reproducing his information here. The book is reasonably-priced and I've seen used copies go for as low as $25 on Amazon, so if you're interested in the content... get one. Make sure you get the second edition. The book is not about specific projects but, for me, it's been a great idea book for how to use all this stuff.

The good news is, it looks like A:M has a pretty full CG lighting toolset. However, since his discussion is mostly non-app specific there are some terms and procedures that may need translation to A:M terminology. Those notes begin below.

There's quite a bit to this subject so Birn can't cover each topic exhaustively, but I think after you've read this book and noted the translations for A:M you ought to be able to pick up any other CG or traditional lighting book and apply it to your A:M work.

Details of A:M features are typically covered in the A:M Technical Reference (TECH REF) and occasionally I will point to a page number in the Technical Reference where you may find further info.


Chapter 01 Fundamentals of Lighting Design

This is a very broad discussion of why you need to devise lighting to suit each scene. Not much needs translating here. The free sample files he mentions are at the link above. The "lighting challenge" scenes are all in OBJ format which is awkward for A:M users. I'm converting the "Fruit Bowl" scene to A:M format and I'll post that here when I finish.

Chapter 02 Lighting Basics and Good Practice

some translations (book term... A:M term)

background plate... similar to "rotoscope" image that you import to your camera frame.

point light... bulb light
spot light... klieg light

penumbra angle... width softness (A:M's term is really a better description of what he is talking about)

barn doors... A:M klieg lights do not have actual doors in front of them (i've never seen a program that did!) but you can precisely control the shape of a light with a gel (described in the A:M TECH REF pg. 100)

directional light... sun light

area lights... Birn says that scaling any of the basic lights above will not change the illumination. However, in A:M the light "Width" parameter can do that. A setting of 0 cm will mimic the basic functionality Birn describes, but if you set a higher value and choose multipass rendering A:M will jitter the light placement to simulate a light with real width. In regular rendering, setting ray-traced lights to have more than one "ray" will also do this.

true arbitrarily-shaped area lights are not part of A:M but in practice you can create them by constraining a light to repeatedly traverse a spline in the shape of your desired "area" on every frame in a multi-pass render. Birn warns that area lights are time-consuming to render. This is true of this technique also.

models serving as lights... similar to area lights but with a 3D shape. The same work around for A:M would apply.

Birn says a global illumination renderer can make any object a light. A:M has global ambiance rendering, which is not quite the same thing. An object surface set to 100% ambient intensity in A:M will not illuminate nearby objects when global ambiance is used. However, I believe this will work in a full-tilt radiosity render in A:M.

environment spheres... this is very much like A:M's global ambiance, although this will not create shadows in A:M. If you want shadows cast from one object to another you will need actual lights in your A:M scene.

soloing a light... the easiest way to turn lights off for testing purposes is to set their Active property to OFF. This does create a keyframe that you may want to delete later.

You can SHIFT-select multiple objects, including lights, at once and Active ON/OFF them together to save time.

If I don't plan to keyframe On/OFF in the final clip , I find it useful to RMB on the ON/OFF value in the props and choose "Constant" which forces A:M to just use one key at for the whole animation. Then I can flick Active ON/OFF anywhere and not leave keyframes littered around.

Decay... Attenuation in A:M. (TECH REF pg. 219) This is a percentage property. 0% = no decay, 50% is linear decay, 100% is the numerically accurate Quadratic Decay. A:M doesn't have an option for the extreme "Cubic" decay that Birn mentions. Quadratic decay (100% setting) is so severe, I don't think you'll need it.

Birn discusses but doesn't give a standard name for what we call "Fall-off" in A:M. This is the distance that a light gives its full illumination before decay reduces it. In my mind an object 22 cm from a light with 10 cm fall-off would be lit the same as an object 2 cm from a light with 0 cm fall-off but my tests indicate this is not the case. Lights in A:M with short fall-offs seem to have faster decays than lights with long fall-offs. Experimentation will be your best resource here.

Specular intensity... in A:M this is something we control at the object surface rather than as a light intensity parameter.

Light linking... "Light Lists" in A:M perform this purpose. (TECH REF pg 178)

Cookies... light gel in A:M


The Feedback Loop... aside from all the shortcuts Birn suggests to shorten your lighting development time, I'll add that A:M's on-screen "render lock" AKA "progressive render" (SHIFT-Q) is quite a boon. It allows you to get a quick sense of what the light is doing in any portion of your window you want without having to go thru the steps of a render to file or waiting for an full onscreen "Q" render to finish.


Chapter 03 Shadows and Occlusion

A good introduction to the advantages, peculiarities and difficulties of CG shadows. Only a few broad terms need translating for A:M:

pg. 51 Shadow color... A:M Kleig lights with Z-Buffer shadows do have a shadow color setting. While Birn says that setting a shadow color to white would be the same as turning shadows OFF, in A:M it will get you... an actual white shadow!

Click to view attachment

I'm sure there's a use for that.


pg. 55 Depth Map Shadows... known as Z-Buffered shadows in A:M. An option on Kleig lights. The alternative is ray-traced shadows.

pg. 57 Depth Map framing... Birn briefly mentions drawbacks of depth-mapped shadow with sun lights and bulb lights. A:M's sun and bulb lights have raytraced shadows only, however. It is possible to simulate a depth-mapped bulb light with an array of kleig lights and a gel on each one to prevent them from overlapping. For all I know, that may be how other software does it too.

pg. 61 Depth-mapped Shadow Transparency Support ... Conventional Depth-mapped shadows do not support partial shadows from semi-transparent objects, but A:M performs a basic approximation: 0-25% transparency casts a full shadow, 26% to 75% casts a half shadow and 76% to 100% transparency casts no shadow. A:M depth-mapped shadows do not recognize transparency created by a transparency map on an object. Use ray-traced lights if you need that.

pg. 64 Trace depth... A:M's renderer doesn't use a trace depth setting, the number of surfaces a ray can hit before it stops. It seems to avoid trouble on its own without you needing to set a limit. Reflection does have a "level" setting to avoid needless ray bouncing.

Figure 3.19... I don't see much difference between those two pictures. Do you?

pg.74 Occlusion in Global Illumination... "Global Illumination" as Birn uses the term seems to most resemble radiosity renders in A:M


pg. 76 Final Gathering... I only know of this term in relation to radiosity parameters. Apparently it can be a sort of one-bounce radiosity technique by itself, but in A:M it is always a subset of your radiosity process, not a separate rendering method.

pg.77 Image Based Illumination... This seems most like A:M's Global Ambiance when a image map is chosen rather than a single color. I'll note that while A:M's Global ambiance does not produce shadows, the older work-around to simulate image based lighting - the spinning light rig - does. So, spinning rigs may still have a life in A:M lighting.

pg.78. Shadow Only Light... in A:M turn off both Diffuse and Specular and turn on Shadows. This seems not to work on ray-traced lights. This shadow-only light in A:M seems to create total darkness where its shadow lands even if another light might be shining there. Don't know if that's a bug or a feature. An alternate technique Birn describes of pairing a positive and negative light also works in A:M, with both Z-buffered and raytraced lights and produces shadows that interact with other lights in the expected fashion.

pg. 80 Shadow Objects and Baking Lighting... Don't think either of these exist in A:M yet.


... to be continued

Chapter 04 Lighting Environments and Architecture

This is a great chapter where he explains the fundamental workflows of lighting interiors and exteriors. Not much needs translating here.

He makes quite a bit of use of simple kleig lights and I notice that A:M Kleig lights seem to have a more sudden drop off at the edge than what he shows in the book, even when they are set to 100% edge softness. It's possible this is an artifact of mis-set gamma on my monitor. But you can make a gel for A:M Kleig lights to create any drop off you want.

Near the end of the chapter he revisits more complicated lighting devices. As far as I can tell the terms translate like this...


pg. 112 "Conventional Radiosity"... is similar to what A:M did around V10 where it actually stored an image map of bounced light on the surfaces. Birn notes this technique has been largely superceded by...

pg. 113 "Photon Mapping"... which is more like what A:M now calls Radiosity

pg. 115 "Final Gathering"... in A:M is an option of the radiosity process, not a separate technique. Maybe someone can figure out away to Final gather without doing much Radiosity?

pg. 116 "Caustics"... likewise in A:M, Caustics is a subset of Radiosity, not a separate rendering process.

pg. 120 "Ambient Occlusion"... In A:M AO is occlusion from the infinitely distant environment sphere. An AO image in an enclosed room would be black since everything in the room is completely cut off from the environment sphere. Birn's AO is occlusion from any surroundings which works also for interiors.

update: Nancy reminds me that it is possible to turn off occlusion casting for objects such as walls in a scene to allow more exposure to the environment sphere by other objects. This is a useful option although still not quite the same as the general AO described in the book. See discussion here.


Chapter 05 "Lighting Creatures, Characters and Animations"

This chapter is probably most like what you would read in a traditional studio photography lighting book. After you read this chapter those sort of books would probably be good further study.

Most of this should make sense to A:M users who have gotten this far.

pg. 131 Three point lighting... If you open a default chor in A:M you will indeed get three lights named Key, Fill and Rim. Remember that those are just names that have been given to two Sun lights and a Kleig light, those are not types of lights in A:M. The "Key" and "Fill" light roughly correspond to placement Birn discusses in his section of three-point lighting, but the "Rim" light is not in a location to serve as a rim light. The default chor was probably made the way it was so new users could plop a model in the chor and get something presentable fast. But don't take it as an example classic three-point lighting described in the book. Do what the book says and start from scratch with one light.

pg. 144 Facing Ratio... he describes a Maya parameter that varies the surface qualities of an object based on how much it faces the camera. In A:M this is done with a "gradient combiner" in a material. Gradient combiner has a start and end parameter for making colored bands, but if you leave both at 0,0,0 and set "Edge gradient" to something other than 100, 0 or -100 (+- 50 is typical) the material will transition from one of the child attributes to the other depending on the angle of the surface as seen by the camera. An example is velvet materials that have an edge sheen.

pg. 152 Mapping variation... A:M doesn't have a map type to vary Subsurface Scattering properties. I can imagine a compositing solution but that would need some R&D.

pg. 153 Faking Subsurface Scattering... aside from his suggestions, I'll note that A:M has the plugin diffuse "Skin" and specular "Pharr Skin" shaders (meant to be used together instead of, not with, SSS) that give fast results, but have not been much explored by A:M users.

pg. 153 ramp shader... in A:M this is a plugin shader>Gradient or Gradient Light. Like a toon shader but without the toon lines. Or, you could use the Toon shader with lines set to 0 width.


Chapter 06 "Cameras and Exposure"

This chapter gives you some background on the functionality of real cameras and how that enters into our attempts produce more realistic renders.

pg. 160 f-stop... Birn doesn't really define it so I will... an f-stop number is the distance from the lens to the film plane (focal length) divided by the width of the internal aperture opening. So a 100 mm focal length lens with an aperture sized to 50mm would be said to be set at f2. The same lens with an aperture of 25mm would be set to f4. In actual practice modern lenses have so many pieces of glass in them that the stated numbers may not reflect physical measurements, but that's the optical theory behind it.

I mention this only so it makes more sense as to why a large f number really means a small opening in the lens. You can think of it as the bottom of a fraction. f8 means the aperture is 1/8 the focal length.

Birn mentions f-stops as he talks of...

pg. 160 Depth of Field... A:M doesn't use f-stop settings with it's Depth of Field effect. A:M DOF has settings for an actual focus distance in front of the camera and "near" and "far" distances that are distances from that central focus point that you want to remain in focus also. All three numbers shoudl be set to positive values. The A:M Tech Ref covers DOF use on pg 223.

If you have a compositing program that can use a depth map in conjunction with a blur filter, Skipping DOF in A:M and doing it in post is probably the most flexible way to go. A:M can render depth maps in an OpenEXR file render.

A:M creates DOF effects two ways. In regular renders it applies a blur to parts of the image. It's a rather slight effect, but it is very fast. In multi-pass renders DOF is created by jittering the camera position slightly. This creates more accurate blurs but many passes are needed to smooth large blurs.

p. 165 Motion Blur... A:M motion blur is set as a percent of the frame duration rather than a shutter angle. 25% would match a 90° shutter opening. 50% would match a 180° shutter.

Like DOF, A:M motion blur can be had in two ways. Regular render does it as an added blur which is fast and works well for straight line motion. Multipass combines images from slightly different points in time during a frame. More accurate but many passes may be needed for large blurs.

pg 170 Video fields... turning on Field Render in your render options will create an interlaced image such as Birn shows in Fig 6.14.

Chapter 07 "Composition and Staging"

Whole books and university level courses have been built around the topic of composing an image within a frame. Consider this a Cliff-Notes introduction to the aesthetics and technical complications of working within that rectangle, intended to get you started.

No translations needed.

pg. 191 The Line of Action... I've always heard people refer to this as the "180° Rule".

Layout and storyboard Guy Mark Kennedy has more elaborate explanations of it here and later here.

Mark Kennedy's blog, in general, is a fine place to read about composition issues from someone who has been in the thick of it. I recommend checking every post from the beginning.



Chapter 08 "The Art and Science of Color"

pg 214 Color Mixing... A:M uses a 0-255 RGB scale to represent color levels rather than 0-1

pg. 240 8 bit color... Most of A:M's image formats are "8-bit". Targa and OpenEXR support an alpha channel. PNG and BMP may also but maybe not in the normal sense. see more discussion below

If you wanna be sure... use targas.

pg. 241 HDRI... OpenEXR is A:M's High Dynamic Range Image format. Many online CG discussions confuse the term HDRI with Global Illumination or other advanced rendering schemes. HDRI may be useful in those but is not a rendering process.

Birn doesn't discuss Quicktime, but I'll note that most Quicktime codecs (with names like MPEG-4, H.264, Sorenson, Cinepac...) are 8-bit color formats. Most of them have settings to vary the amount of compresssion they apply to an image.

"Animation" codec is 8-bit and is the only Quicktime codec that can support an alpha channel and is the only Codec that can be truly uncompressed or lossless (if you set its quality to 100%). "Animation" makes huge files; you would not want to post a clip on a web forum in Animation codec.

Most A:M users who need an alpha channel in their work render to a targa series, however.

Avoid the "Video" codec, it is an antique 5-bit format. I think A:M will give an error if you select it.

"Photo-JPEG" is unlike most codecs in that every frame is a "keyframe". While this makes for somewhat larger files, it also makes scrubbing a quicktime movie frame-by-frame in the QT player very responsive.

To set QT compression settings, choose "Quicktime Movie" as your image format, then expand the Format triangle and click on "set"

A second screen, "Compression Settings", will appear where you can choose a codec from the "Compression Type" drop-down list. Most codec have the parameters shown in this screen shot but they may vary.

Click to view attachment

"Frames per second" is ignored by A:M. A:M uses the frame rate of your project.

"Key frame every" tells QT how often to store a frame that is not dependent of neighboring frames for part of its image. If you leave this unset QT may try to make every frame a Key frame, resulting in large files. Set this to a large number, such as 48, for smaller files.

"Limit data rate to" instructs QT to use enough compression to keep the size of your file under a certain amount per second of footage. Higher numbers make larger files but better quality. With some codecs useful data rates can be as low as 100. If you leave this option unchecked QT will use a general data rate target consistent with the "Quality" setting you choose on the slider below.

For more control of compression parameters, render to Targas and use a utility such as Quicktime Pro to compress the image sequence. Quicktime Pro also gives you access to audio compression, which A:M does not.


Chapter 09 "Shaders and Rendering Algortihms"

pg.248Shaders... "Shader" in A:M has approximately the same meaning as Birn ascribes to it. A:M's default shading behavior is what the regular surface parameters control. The other shading behaviors are chosen in the drop down lists at the bottom of the surface properties.
Click to view attachment

They either override, make special use of or add to the paramters you seen in the regular surface props.

They are not all well-documented, although many have industry standard CG names by which info can be searched for on the web. Comments by ypoissant (who coded most of them for A:M) on this forum are also out there.

Some work in pairs, for example "MatCap" has both a Diffuse and Ambiance shader

pg. 250 glossiness... A:M creates the glossy effect in the middle of Fig 9.3 with "Soft" reflections, set in the render Options. The strength of the effect is controlled on a per object basis by the specularity parameter. See TECHREF pg. 234.

(There is a "Glossy" specular Shader, but that is for specular highlights. A:M has a reflection "Falloff" parameter but that only diminishes the strength of the reflection without blurring it.)



See post 13 below for a multi-pass approximation of glossy reflection.


pg. 250 a diffuse parameter... A:M surfaces don't have a "diffuse" percentage but A:M does have "diffuse maps". Much like a diffuse parameter they can darken a surface. Rarely used, but there. Putting scuff marks on a wood floor that was textured by a material would be a use for a diffuse map in A:M.

Diffuse Falloff... Birn doesn't mention this but it's an important parameter in A:M right below Diffuse Color. TECHREF hits it on pg 16. Most objects will have a diffuse falloff that is higher or lower than the default 100.

pg. 254 the fresnel effect... like "facing ratio" above, use an edge gradient material to give edges and faces of objects different reflectivity. The only disadvantage to the A:M way is it's awkward to add this as an after-thought to your material. If you're doing a reflective surface it might be good practice to always make that first node a gradient combiner.

pg. 257 Anisotropic highlights... the Ward and Westin specular shaders and possibly others have this option.

One caution... the Ward shader has a long drop down list of presets. Unfortunately none of them change the parameters at all. You'll need to experiment with parameter son your own.

pg. 258 BDRF... I wish he had shown a picture of what this does! The Ward shader actually has a parameter called BDRF Normalization. If you figure out what it does, let me know.

pg. 259 Anti-Aliasing, oversampling, under sampling, adaptive oversampling ... A:M does anti-aliasing two ways:

Multi-pass renders "over-sample" or "under-sample" depending on how many passes you choose. 16 passes samples each pixel 16 times. Technically that is "oversampling" although the anti-aliasing you get from 16 passes is what we consider normal for most purposes.

More passes = more samples = finer anti-aliasing = longer time

Fewer passes= fewer samples = less antialiasing = shorter time

If you want the ultimate in under-sampling, choose multi-pass with one pass.

In a Regular Final render (multi-pass off) A:M does the "adaptive over-sampling" that Birn describes. This is much faster and produces anti-aliasing about as well as 16 passes in a multi-pass render.

pg. 262 Adjusting the contrast threshold... A:M's anti-aliasing doesn't have a parameter for this. I've never missed it.

pg. 263 filter... Likewise, there isn't an optional anti-aliasing "filter".

pg. 271 Index of refraction... A:M's IOR goes from 1 to 2. I guess you're our of luck on diamonds (2.42)

pg. 273. Colored refraction... you could duplicate Birn's RGB trick with multipass by sweeping your light color and IOR setting during the course of a frame.

pg. 274. Reyes Algorithms... based on comments Martin and others have made about A:M's renderer, it is probably very much a Reyes-style thing as described by Birn.

pg. 276 Raytracing, Z-buffer Rendering, Scanline Rendering... Martin has said A:M is a hybrid render; it does what ever method will get the desired result the fastest. If part of the image won't benefit from raytracing, it does scanline rendering there. If part needs raytracing to be done right, the renderer goes into raytracing. All automatically, you don't need to make these decisions for it.

pg. 278 hardware Rendering. as Birn defines this, this is the Shaded render and real-time display you see on your screen. Shaded render is extremely useful for animators who want to quickly see how their stuff is moving at full frame rates and aren't worrying about shadow or lighting details yet.

pg 278 GPU acceleration... Birn describes a number of reasons why this promising technique is not widely adopted yet but has hope for the future.

Martin has said that they tried implementing the "Gelato" renderer in A:M several years ago but were not able to get useful replies to their technical questions. Perhaps other GPU schemes will present themselves more favorably.


Pg. 279 Interactive Previewing... A:M doesn't have literal interactive previewing but the progressive Shift-Q onscreen render is pretty close. It's great for nearly real-time feedback on things like shifting material placements or adjusting lighting parameters.

In addition, an OpenEXR render with lighting buffers can be viewed in an A:M Composite and you can actually adjust individual lighting levels after your render is done.

Chapter 10 "Designing and Assigning Textures"

An overview of all the ways that surface detail is created for models without actually modeling it. Another huge topic that is necessarily getting condensed treatment. There are many good books devoted to the topic in CG and although they are typically written as app-specific, the workflow is 100% transferable to A:M.

There is a book specifically on painting textures in this same "[digital]" series. I haven't read it yet, but it's on my Amazon wish list.

pg. 286 Specular Mapping... A:M has two types of specular map, intensity and size. Remember not to leave the specular color unset in your surface parameters or specular maps may appear not to work.

pg. 286 Figure 10.4... looks like a checkerboard to me!

pg. 287 Incandescence Mapping... known as Ambience Intensity map in A:M

pg. 288 Transparency Mapping... A:M's Transparency maps work opposite from Birns's description. In A:M black creates transparency and white creates opacity.

A:M's Spec, Ambience and Transparency maps are gray-scale only. You can apply a color map but they operate as if they were averaged out to gray-scale.

pg. 290 Displacement Mapping... A:M uses middle gray as the level for zero displacement. That means you only get 128 levels above or below that in 8-bit grey, but OpenEXR format maps can give you more. Displacement mapping can behave a bit unpredictably at high percentage levels especially on the parts of the model seen on edge.

pg. 295 Other mapping techniques... in A:M these include the Diffuse map, and several maps for controlling the appearance of hair, such as density, length, and direction.

pg. 310 Decals... in A:M "Decal" is the term for all image maps, whatever their scope, and not just ones applied to a small area. The TECHREF covers decaling beginning on pg. 59


pg. 315 Figure 10.38... I really think he has those reversed. The left one looks like a spherical projection and the right looks like a cylindrical projection.

BTW, I think cylindrical and spherical mapping are under appreciated for character mapping. Used right they can greatly simplify the "unwrapping" process, because the projection does most of the wrapping for you. One example of the process This mapping technique works well with 3DPainter.

pg. 316 Camera Projections... I've also heard this called "Camera mapping". In A:M you can do this in any view in the chor. Position your object, then select it in the Objects folder in the PWS and RMB>New>Decal>choose decal. the map will appear for positioning. The sizing controls are very twitchy, be patient.

This also works for applying a map in a "pose" window where you have distorted a mesh (to flatten it , for example).


pg. 318 UV Co-ordinates... A:M uses the "Explicit UV Co-ordinates" Birn describes, since A:M's meshes are free form and not resticted to grids like NURBs are.

pg. 320 Multiple UV set... easy in A:M, every decal has it's own UV arrangement. If you should need two maps to align exactly (a color map and bump map for example), just stamp one then drop the matching maps into the Decals>Images folder that is created. Remember to set them to their intended type.

Click to view attachment

pg. 323 3D paint Programs... There is a 3D Paint program made especially for A:M, Pixosaur 3DPainter


pg. 324 Procedural Textures... these are A:M's "materials". A:M has a number of pattern and noise combiners to create complex materials. see pg. 14 and after in the TECHREF

If you are itching for additional combiners the Enhance A:M set has quite a few. And works with V15 even though it says V13.

Another powerful procedural material authoring solution is Darktree which creates textures to be used by the Darksim plugin included with A:M.

pg. 330 organic modeling tools... programs like ZBrush have limited use for A:M modeling because they do not produce good topologies (spline arrangements) for an A:M model. Models that have been created with a strictly "Sub-D" process come closest to A:M topologies but will still need editing because "Sub-D" is not aware of A:M features like 5-point patches and hooks.

pg. 331 Baking Geometry into Displacement maps... I recommend rendering your map to OpenEXR for the best gray-scale resolution.


Chapter 11 "Rendering Passes and Compositing"

pg. 336 layers... A "layer" in A:M is a flat plane that an image is applied to. Birn's use of "layer" varies. In some software a "layer" is a group of models in a scene that can be turned on or off together. A:M doesn't have such grouping, but you can organize objects in your chor into folders, making it easy to Shift-select a whole bunch at once to turn them on or off.

Note that turning the Active property ON/OFF (circled in blue) affects whether an object will appear in a render. Setting the Object Render Mode icon (circled in green) affects whether an object will appear in real-time view but not in render.

Click to view attachment

In the above screen the lights in the scene have been set to not be visible in the real time view but they will still shine in a render.


pg. 337 Background Layers... Now he's using "layer" more like an A:M layer... an image on a plane in your render.

pg. 338 Matte Objects... A:M doesn't have a Matte Object shader (someone could write one) but you could quickly create a pass such as shown in the right of Figure 11.5 by turning off all the lights, setting the grass to white with 100% ambiance intensity and the egg set to black. In your compositing program (which you have because if you didn't have one you wouldn't need a matte object anyway) substitute that pass for the original alpha channel that was created by the render in Figure 11.4. After Effects is great at such swaps.

pg. 340 Why Bother with Layers?... He actually framed this, it's so important. The last bullet point should be reassuring to A:M users who, if they find a bug in A:M, wonder if they've made the wrong choice and that other software wouldn't have any bugs or limitations.

pg. 342 Alpha Channel Issues... By default the alpha channels that A:M creates are "premultiplied". Compositing software such as After Effects understands both sorts, but sometimes footage you import to compositing programs needs to be explicitly set to the correct kind.

A:M has an "unPremultiply" post-effect, however at this writing it has a bug. I'll submit an A:MReport on that.

pg. 361 Global Illumination Passes... I'm not sure how you'd produce something like the image in figure 11.27. Any ideas? Perhaps by subtracting a regular render from a Radiosity render?

pg. 366 Rendering Many Passes at Once... A:M does this via the OpenEXR format render. A:M can render these passes as "buffers" within an OpenEXR render:

Depth
Normal
Diffuse
Ambiance
Occlusion
Reflection

and separate Diffuse, Specular, and Shadow buffers for each light or "Light Object" (a collection of lights acting together for one purpose)

and maybe others I haven't stumbled upon.

Occasionally the OpenEXR setting won't make all these buffers or will make a black render. Try switching your Multipass setting and try again.

A:M's compositing ability can make use of many of these buffers. Forum member "Fuchur" has done a very good introductory tutorial on using OpenEXR and A:M Composite. English Version. Auf Deutsch.

For more elaborate manipulation you may want to use a dedicated compositing program. I believe After Effects still requires a 3rd Party plugin to be fully OpenEXR aware.

This screen shows a PWS listing of all the buffers rendered in one pass, and below that a "Composite" built from those buffers

Click to view attachment


Chapter 12 Professional Pipelines and Professional Practices

This last chapter spends quite a few pages explaining the professional production process and how the lighters' and renderers' work fits into it. This probably a more detailed look at studio production than animator-centric texts ever do.

No translations needed here.

This is very interesting Robert ,Im terrible at lighting so am learning as I read

After reading Robert's comments, I decided to purchase the book from Amazon.
I hope I can understand enough to make it useful along with Robert's conversion of the terms.
Thanks for all the help translating it to AM.

This is really great, thank you Robert.

It is possible to set "Cast Occlusion" = OFF for the enclosed room model (under options for shortcut to model in chor). So one can do occlusion renders "indoors" in A:M, but probably better to just have an open ended room

Thanks, Nancy. I hadn't really paid attention to that thing.

Turning cast OFF for the walls does permit more occlusion in an enclosed space but still not quite the AO look in the book.

Here's a comparison...

the top is with the walls casting occlusion, so the only real exposure for anything is via the small window and the missing wall behind the camera. This is an interesting look, none-the-less.

The middle is with walls set to not cast occlusion. This lets more ambiance in, but surfaces of objects facing the walls show no occlusion

The bottom is my approximation of general AO with a traveling light and multipass. This is closest to the AO effect shown in the book. The nearer two surfaces are, the less exposed they are to their surroundings and the darker they are. It's as if every point in the volume of the room was a shadow casting light. It's not perfect, the post in the middle of the room is much darker than it ought to be for being so exposed to its surroundings.

Click to view attachment

I like the bottom, except for the dark post - and darkness in from the window - But makes me wonder where this light (klieg? pseudo cube klieg?)is traveling? do you have it staying outside the room, projecting in from the window and "jittering" on some path? or is multipass doing the jittering? or both? What size is the light? Shadow softness?

looks as if the light is behind the post and in front of wall inside the room?

The light is my omnidirectional Kleig light traveling a zig-zaggy path just inside of each wall, ceiling and ground. Making a truly random path is a challenge. The zig-zag still has too many samples in straight lines when there are 256 passes.

I'm thinking that to do this better the light will need to also travel the interior volume of the room and not just the sides.

The kleig light ball uses about 20% softness to try to cover up the stepping of the shadows. I also use the lowest shadow map res since higher is slower and not needed.

I'm trying to figure out how to make one big cube of randomly oriented lights that could blast this effect out in one pass.

I'll be interested in your reaction to it. For me there have been a lot of "a-ha" moments when I realize I could be doing something in A:M like what he's talking about. On the other hand because he's being non-app-specific he can never really take you through a specific project from start to finish so you have to build your own experiments from scratch.

A:M doesn't have a "glossiness" surface parameter, which makes surfaces somewhat reflective but increasingly blurry as distances increase.

This approximation set surface "roughness" to about 10 and sweeps "roughness scale" from .01 to 1% over a multipass frame. To de-grain it I rendered it at double size and shrunk it back in Photoshop.


Click to view attachment


This was 9 passes and took about 5 minutes.

Click to view attachment


The same setup with regular reflection:

Click to view attachment

This looks like it might be accomplished with "Soft Reflection" in AM, but your solution probably renders a lot faster.
Camera Properties > Render Options > Reflections > Soft Reflections

I tried soft reflection ON, but i couldn't detect any change at all. Is there an example project that shows it?

Maybe it's just not working and needs an AMReport. But I tried V13 too and didnt' see anything different.

it stopped working

As I recall Birn's was a very good book but I actually got more out of Gallardo's 3D Lighting: History, Concepts and Techniques. It's not an easy read and it is more like a text book however, beyond a complete dress down of the subject, it has a lot of specific information on settings for different lighting as well as tables of temperatures, refraction data and the like which I don't recall seeing in Birn's and in fact Gallardo's book is one of the ten books that sit on my desk within easy reach. Mind you I read Gallardo as well as Birn's (and also Demer's [digital] Texturing and Painting -- a good portion is on lighting) many years ago -- around 2003 I think -- and so admittedly I'm going on memory as far as the coverage of the subject goes (the reference value of Galardo's is definite). In fact I now see that you're talking about a newer edition of Birn's that's come out since I read it, still these are my experiences. I'll have to see how much new material is really in Birn's book... perhaps pick up the new edition.

Cheers,
Rusty

Here is an example project that has a chrome material applied to a table. One image rendered with soft reflections on and one without. It's a standard Choreography with standard light setup, both models and chrome material were on my v13 disc. For the soft reflection image: Reflections "on", Levels "2", Soft "on", Quality "40%"...both rendered in v15.0i.

Hope that helps.

Thanks. Your PRJ did render soft for me. 30 minutes! but much smoother. I guess there's no way to vary the softness effect.

So i wonder why soft didn't work on my test? Maybe the refection has to be set in a material rather than in a group?

I notice in your test, the soft version correctly captures the reflection of the sky color while the not soft render does not. That is also interesting.

I'd recommend the Birn book because it seems to be mostly accessible for a typical CG program user, but it's certainly not the last word. I'd recommend it as a first book for someone who knows not of lighting.

It's got a few tables now regarding color temperature and light settings.

I think that after reading the Birn book and translating it to A:M terms, someone could pick up a more detailed lighting book, even one written for traditional photographers, and make sense of it for CG purposes.

That surprises me. It only took about eleven minutes to render on my laptop...and mine tends to render slower than most. The softness should be adjustable using the "Quality" setting...I think.

I haven't used it much, so I don't know all of the ins and outs.

Here's a render with the setting at 10 (the lowest possible)

Click to view attachment


I'd say the blur effect is about the same dimension, but it is grainier. Only 8 minutes to render however!

The softness is related to Specular Size. Softness, glossyness, roughness, they are all basically the same thing.

Thanks, Yves! I should have checked the TECHREF. It's in there on pg. 234.


Here's my test of variable soft reflection via multipass:

Click to view attachment

That works quite well Robcat. Did you animate the cup's specular size to get the animated soft reflection?

No that's by varying the roughness.

I did that before Yves check in and explained the proper use of the soft reflections.

I've finished my translating and annotating. See the first four posts above.

If I've made any mistakes, let me know!

I believe png, QT mov, openexr, bmp (that dooesn't seem right?) supports alpha channels - I haven't tried, but the option exists when rendering to any of those other formats. It doesn't show up for jpg, avi.

I believe it's ok to leave the color unset (will use the model's color, spec color) - I believe it's the specular size and/or intensity that should be set in surface properties in order to get the specular maps to work.

Nancy writes...



Well, I suppose I should actually try them...

I'm testing these in Photoshop 6...

BMP Photoshop couldn't read the file that was rendered with Alpha ON A:M recognizes that the BMP is 32 bits but when I load it as a rotoscope the whole thing is semi transparent, not just the no-object areas. That's odd.

EXR had an alpha channel

PNG loads into PS without an alpha channel but as a Photoshop "layer" with transparent areas. You can see the default PS checkerboard behind it. I guess there's some transparency information in there but not accessible for manipulation in PS.

There's no standard AVI codec I know of that supports alpha channels.

In Quicktime the "Animation" codec can support alpha.

My specular test map had no effect until i clicked that color chip to white.

Another way would be to leave all the models as they are and use the following settings (no need to change any models):

hmmm...okey-doke...maybe it's version related ? but I will take your word for it - it's been awhile since I've done that, and as they say "it couldn't hurt"

EDIT: just tried - YUP - had to set the spec color as well

EDIT2: NOPE - I take it back - I tried again and got it to work without setting spec color...hmmm...

Birn's Matte Object is something else not involving shadows. It's an object that will leave a hole in the alpha channel. So it doesn't appear in the render, but it's different from a transparent object in that anything behind it is disappeared also. We had someone ranting about this on the forum a while back I recall.

You can read sample pages from his book on Amazon and if you search "Matte Objects" you can read about it on pg. 338-339.

I think it's fairly exotic to need it but I think it would be simple for a plugin writer to make.

I just turned off shadows for the Klieg light and added some more models to same chor as above - I believe this could be used in AE somehow - but I should go read (I have the first edition - got a long time ago)

It depends on what's in front of what in that scene. You need it in combination with the regular render.

Here's an approximation of Birn's scene:
Click to view attachment


Here's the alpha it would normally have:
Click to view attachment


Here's the alpha he wants if the Egg is a "Matte Object"
Click to view attachment


My approximation resembles the correct result but i don't know if it's pixel perfect in the transitions from black to white.

My approximation was made by setting the Egg to the Null Shader and the grass and the ground to the Ambient shader set to white.

I swapped that result with the original alpha channel in Photoshop causing this appearance...
Click to view attachment


If I composite that in AfterEffects with a new egg like this:
Click to view attachment


I get this:
Click to view attachment


I don't see any fringing so I think that is working properly.

okey-dokey - now I getcha (I think).

I turned off all lights in the chor, made the sky black (should have rendered with alpha ON instead), the ambiance shader for the ground = white. Made ambiance shader for 1 instance of Tom = black (the matte object), and the other instance of Tom's ambiance shader (or any other models that are not matte objects) = white

Don't necessarily like this approach as would have to monkey with the model properties - can't just set ambiance shader in the chor if it hasn't been set in the model. But I think I understand the objective now.

Now if we could get one of those things to make itself an alpha channel without having to do the copy and paste in Photoshop.

One way in A:M would be to add this image as a transparency map to a flat patch that had the original RGB image, and rerendering that, but that requires you to try to align your 4x3 patch exactly with your 4x3 camera to avoid resampling degradation. It probably can't be done perfectly.


There are some tantalizing commands and merging functions in A:M Composite but I don't know enough about them to scratch it.