Daz 3D is part of 
Connect
DAZ Productions, Inc.
224 S 200 W, Suite #250
Salt Lake City, UT 84101
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.Daz 3D is part of
Connect
DAZ Productions, Inc.
224 S 200 W, Suite #250
Salt Lake City, UT 84101
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
I would immediately think the Diffuse could be used as a 'base' pass on which to put other passes, and the Specular pass of course could be made stronger or weaker in opacity to make the highlights stronger/weaker, and do some color changes to the spec so it isn't pure white but maybe blends more with the background colors of the scene to mimic reflective highlights? However I'm betting that's all too obvious and maybe there's some other even more interesting stuff that can be done with these two passes?
I was watching some Gimp tutorials on how to use Levels and Curves to enhance an image and it's frankly kind of amazing how much you can change an image to make it 'pop' better using these tools alone. Doing some playing around with some Carrara renders I find it surprising how much 'hidden' info and detail is in my Carrara render that with some of the tools can be 'brought out' into detail. I don't mean to wander off topic from ways to use the various passes, but I'm just a little shocked by how much enhancement even just Gimp can do to a straight out render. Factoring in various passes into the mix, and I guess there's a ton of stuff that postwork can do that I never knew it could do (perhaps the most shocking thing in this thread was the description of the normals pass and the fact you can actually 'relight' your scene post render; I'm still trying to grasp the possibilities inherent in that)
Tutorial describing concepts behind various render passes. Maya based, but the concepts should apply.
http://www.3dtotal.com/index_tutorial_detailed.php?id=1507#.VS3uMzpFCUk
Compositing concepts and guidelines in Photoshop - based on stock images but concepts are general (hour long video)
https://www.youtube.com/watch?v=38Ku7XA3LKI
Diffuse/Diffuse Material Color/Diffuse no Shadow
These passes only render the diffuse component, giving a very matte look. Diffuse Material Color is an unlit version that can be multiplied with lighting. However some problems occur with rendering some of the mental ray shaders. The Architectural Mia Shader and any SSS shader will appear black in the Diffuse Material Color and Diffuse no Shadow passes.
There is a workaround for this by placing an ambient light with a value of 1 in your scene, but the Diffuse pass renders correctly and so this might be enough in most cases (Fig.02 - 04).
Specular/Specular no Shadow/Reflection
The specular and reflection components can also be rendered out separately, just like the diffuse component. The way to add these in post is by using either Screen, Color Dodge or Linear Dodge. However some pretty dramatic reflections can be created by using Overlay, Soft Light or Hard Light (Fig.05 - 07).
Shadow/ShadowRaw
The Shadow passes need to be used with Multiply, Color Burn or Linear Burn. The shadowRaw pass serves as a mask, but when inverted it can also serve as a Shadow pass when using the same blending modes as with the shadow pass (Fig.08 - 09).
DirectIrradiance/DirectIrradiance RGB Masks/Indirect
These passes are the lighting passes; they are great for tweaking the balance between the direct and indirect illumination, and the intensity of specific lights. The directIrradiance and Indirect pass are created straight via the render passes. The RGB Masks variation is a custom-made pass. The three most important lights are picked and given red, green and blue as color respectively. These can then be extracted via the individual color channel to get a grayscale map. The only reason why red, green and blue are used is because it allows you to extract the information of three lights at ones. You could also render each light individually with a white color, but that would be less efficient (Fig.10 - 12).
Depth
The Depth pass is a real workhorse. It takes just seconds to render and can fake an effect that is one of the most expensive to render: depth of field. Next to that you can also use it to create fog (Fig.13 - 15).
Ambient Occlusion/Ambient Occlusion Max Distance/Reflective Occlusion
Ambient Occlusion is a great way to quickly add realism to your render. It is lot faster to render then Global Illumination or Final Gather. The only drawback is that it doesn't create color bleeding since it is a grayscale effect. The maximum distance has a huge impact on the look of the effect so it can be practical to render more than one Ambient Occlusion pass.
The Ambient Occlusion pass uses an infinite max distance and the ambient occlusion max distance pass uses a finite max distance. The fact that these passes are rendered at 16 bit helps a lot with compositing. It even means that you could fake a max distance render by blowing out a lot of detail on the Ambient Occlusion pass.
Reflective Occlusion is a special pass that is meant to be used to occlude specular and reflection passes. Just as the Ambient Occlusion pass is actually meant to be used only to occlude diffuse lighting. When compositing, the specular component can be dulled out by the Occlusion passes. If that happens, you need to add the specular after occlusion (Fig.16 - 18).
IncidenceCN/IncidenceCNMat
Incidence is a mask for a Fresnel effect. Meaning that it visualizes the facing ratio of the geometry. Black is directly facing the camera; white is facing 90 degrees away from the camera.
CN and CNMat are different variations of this effect. CN does not take the normal map into account but renders mental ray materials with normal maps. CNMat does take the normal maps into consideration, but is unable to render the special mental ray materials such as the architectural MIA shader.
You could also use the blue channel of the NormalCam pass; when inverted it gives an incidence pass that does render both types of materials, but only the mental ray materials with normal maps (Fig.19 - 20).
NormalCam/Normal World
The Normal map passes generate normal maps depending on the coordinate space: camera space for the Cam variation and world space for the World variation. The NormalCam pass can be particularly useful since it contains an incidence pass in the blue channel and a vertical and horizontal light pass in the other two channels, which can give nice effects in Overlay mode.
Some compositing packages can use the NormalCam pass to fake lighting. This way you can light your image interactively in post-production (Fig.21 - 22).
Incandescence
The Incandescence pass separates all the self-illuminating parts from the rest of the image. This is great if you want to add a glow around light sources during compositing. In this example you can only see the light dome in the distance, since the model does not contain any light sources (Fig.23).
RGB Masks
The RGB Masks pass is a custom pass in which you can define up to three different material types that you want adjust separately from each other in post. Each color channel contains one mask that can be used by the compositing package. By rendering out more RGB masks you could potentially create a mask for each object (Fig.24).
Before and After
Below is an example of what compositing can do to an image. Compared to the composited image, the original render looks a bit dry and dull. Compositing is a good place to tweak the colors and contrast, and can either make or prevent your image from looking CG (Fig.25 - 26).
Compositing
In compositing we take all the layers created during the rendering phase and combine them into the final image. At this point all the decisions have already been made about the look of the final image, but compositing is the stage where the image either remains CG-looking or becomes something striking and believable.
Besides the technical considerations and the way you need to blend all the specific layers (which is explained in detail above) most of the compositing has to do with evoking a feeling in the viewer and playing around with colors. There are no real guidelines for this because it varies from image to image. The color corrections done for a portrait render, for example, do not apply to a full body render (Fig.27 - 28).
Okay, well let's discuss what Carrara calls the "Velocity" pass.
What is a velocity pass? Well, first of all, "velocity" indicates the speed of something. But it also indicates the direction (X and Y coordinate) that the object is travelling. It is described by what is called a "vector" which you can basically visualize by just an arrow. The direction the arrow is pointing indicates the direction the object is moving, and the length of the arrow indicates the speed it's moving.
So the renderer tells you that information by making a Velocity pass, which provides that info in two of the three RGB channels. Typically, the Red channel contains the X info, and the Green channel the Y info. The blue channel is typically no used. And you can see that if you bring your velocity pass into your compositing program and look at the RGB channels. The "vector" data is imbedded into the R and G channels.
In the scene I've been working with, I animated the sphere to the left and up, and below is the resulting velocity pass.
And below are the individual Red and Green channels which provide the X and Y velocity vector information
Try it with a bunch of objects like an exploding wall of bricks in an animation. it's not as useful as you would expect
Now, as I mentioned before, the velocity pass can be used, for example, to generate motion blur in your images. And what exactly is motion blur? Well, the image below is a decent example. It's a blurred image of a car in motion.
But what causes it? It may seem obvious that stuff in motion has motion blur sometimes, but why?
Well, as with many visual effects that are applied to images, motion blur is actually due to the characteristics of the cameras and film that capture most of the images we see. And we see them so often, we tend to associate them with "reality", and the way things ought to be. Even though they are synthetic side effects of the way cameras work.
Some of those effects include depth of field, lens flares, and so on.
With motion blur, it's primarily due to the fact that when you click the shutter of a camera, exposing your film to light for a brief moment, stuff that is in motion during that time can expose more film, and cause a blurring effect. For example, let's say that a car is moving past your camera at 20 feet per second. And let's say you leave the shutter open for 1 second. In that time, the car moves 20 feet.
So if it starts 5 feet to your right, and ends up 15 feet to your left in the time the shutter is open, then in the fraction of a second it passes within the view of your camera lens it continually exposes the film. And the result is that all you see is a blur.
But this depends totally on the exposure time and the speed of the moving object.
Not sure if you're responding to me or someone else, but I don't understand what you're talking about. Sorry.
Now, unfortunately motion blur can get quite complicated. Yeah, the concept is pretty simple, but in practice it can get quite difficult.
For example, if the moving object is casting a shadow, should you blur the shadow? And if you want to blur the shadow, how do you do it? How would the car image look if the car and background were blurred as they are, but the car's shadow was sharp and crisp? Kind of strange, right? So you have to figure out how to make it right.
And as you can see from the photo of the car I posted, the camera was moving (and shaking a bit) during the time the shutter was open, which caused the background to be completely blurred. But the foreground car was also blurred, but not as much, because the speed of the rotation of the camera wasn't fast enough to keep the car apparently motionless during that time. So how do you simulate that?
And if you look at this image, you can see an absolutely gorgeous image where the camera is in a fixed position, the shutter is left open for a long time in order to expose the background sky, and only the water is moving and is highly motion blurred. Something like that is incredibly difficult to duplicate.
So with motion blur, you really need to think about what you want to portray in your images, and what the moving object was doing, what the camera was doing, what the shutter was doing, etc., if you want it to look "right".
Because most viewers, while they can't explain what is happening in a motion blurred image, they do know what looks "right".
Many CG renders rely upon a simple, fixed camera, and a moving object that has a simple blur applied. Which is fine, but as you can see if you search for "motion blur images" or something, there are many, many ways of making very interesting motion-blurred images. And in fact, you'll probably realize that it's more common for a motion blur to occur in real life when a camera is hand-held, and both the foreground and background are blurred. But simulating that takes some planning and thought.
Otherwise, if you just want a straight motion blur with a fixed camera, you can take your velocity pass and use your compositor's motion blur tool and blur your image. However, the procedure for doing so can be quite different depending upon the application, and I really don't have the energy to outline the steps in different applications.
Actually, the point, and the title of the thread were to discuss compositing and post work in general. But since I tend to get angry comments when I don't discuss Carrara specifically, I figured I'd frame the discussion around Carrara render passes. Of course, I still get angry comments for other reasons, but anyway... :) :) :)
And more importantly to the discussion of "how", let's spend a moment discussing "why".
Why would you produce an image with motion blur? Well, as you can see when you look at the blurred image of the car, you can FEEL the motion. If, on the other hand, the foto was taken with a high speed camera, and both the car and background were perfectly clear and crisp, would you get the same sensation? No.
Again, this is another general tendency (no, not a rule...) of human perception. When you and I and everyone else see a motion blurred image, we FEEL the motion. Now, if you want your viewers to feel the motion it's great. But if you don't care, then why bother? :) :)
Congratulations, you win the bet !! :) :)
And one other important aspect of motion blur that is often missed in the discussion....
Keep in mind that most of the films you've seen over the years were filmed at a frame rate of 24 frames per second. And that frame rate was designed to make sure that people didn't notice a flickering in the sound and visuals. At significantly slower rates you and I start to notice flickering. And flickering is annoying.
So the industry is trying to go to higher frame rates because it appears more real and pleasing. In fact, they have realized that significantly higher resolutions than are being used now don't really add much to the realism and enjoyment, because we can't really see much more at resolutions higher than 4k or so.
And what else appears pleasing is the slight motion blur of each frame to make the motion appear smoother. So even with low levels of motion blur in animations can be more pleasing to viewers.
That is, if you care about stuff like that.
Okay, well, next up I suppose is the "ambient occlusion" pass. What is ambient occlusion?
Well, as with many things in this world, a lot can be learned from its name. "Ambient" refers to "ambient light", which is the diffuse, non-direct light that bounces around our world and makes diffuse, soft shadows. And it makes stuff look "real".
"Occlusion" means "blocking", or a least something like that. It refers to how objects in the world around you "block" ambient light and cause shadows in corners and cracks and crevices.
And if you look at an ambient occlusion pass such as the one below (grabbed off the internet), you can see that it looks like an extremely diffusely lit scene (no direct light just a general bounce light, like an extremely cloudy and foggy day in Scotland).
Now, with that being said, an important aspect to consider with ambient occlusion is this: it doesn't exist. There's no such physical phenomenon in the real world called ambient occlusion. It's merely a render trick designed to take advantage of our perceptions of how we see the world, and what we consider "real".
Consider corners and crevices in the real world. Typically, they gather dirt and junk and therefore are, or appear, darker. A concrete wall outside on the street typically has dirt and junk at the bottom where forms a corner with the sidewalk, because rain washed the dirt down the wall and into the corner.
So ambient occlusion isn't really a calculation of actual direct or indirect lights in a scene, it's a render trick. For example, if two surfaces meet at right angles (ie, a corner) the calculation "assumes" it will be darker, and therefore darkens that corner in the AO pass. It assumes that the ambient will be occluded by the wall and floor at the corner, so it makes it darker.
Now, what can you do with an AO pass? Well, since it gives the "appearance" of ambient shadows, you can use the AO pass to give your renders the appearance of ambient, diffuse light causing dark shadows. And as we did with the shadow pass, you just take the AO pass and multiply it, and you'll get the AO effect. And you can change the levels of the AO pass to increase or decrease the effect.
And this can be especially useful for those with the dreaded "scene ambient brightness" disease, characterized by cranking up the scene ambient to give a cheesy appearance of ambient light, without providing any corresponding shadows. Which is why it's a disease :) :) :)
And FWIW, here's an AO pass from the simple sphere scene I've been using. And to prove that it's not a "real" pass (ie, not rendered using real scene lights) I first deleted all scene lights before rendering.
BTW, note how nasty it looks. :) :) ...specifically the area under the sphere
And below that is the AO pass from the car and girl scene...nasty... :) :)
Might not be something you want to use without some cleanup... :)
EDIT: I updated the AO pass of the car after cranking up the accuracy/quality settings...
Oh, I forgot....Carrara has some accuracy/quality settings which improve the AO pass, so you can play around with those to make things look a lot nicer. :) :)
And of course, as you crank up the all-important "Occlusion Radius" the effect of the fake shadows expands.
Here is an AO pass with the quality/accuracy settings cranked up, as well as the radius setting cranked to 0.5 ft. from the default 0.08 feet.
Also, notice how cranking up the radius also increases the apparent "coolness" factor... :) :)
And here's the super cool AO pass we all know and love...with a 10ft radius. :) :)
So, can AO passes be used for anything useful? Of course !! Anything can be useful if you just use your head. :) :)
As you can see, depending upon the AO pass settings, you can get a wide range of results that look like pretty cool shadows. Now, does that mean you need to use the entire AO pass to make shadows or something? Of course not. ISOLATION !!!
Keep in mind that you can even isolate sections of your AO pass image and use that as a way to generate shadows in just a section of your image. And you can change the radius to get a very wide range of shadow types. There is no limit to the results you can obtain.
But as with all things, unless you have an inspiration that gives you some idea of an effect you want, you may have difficulty finding uses. Which is why I suggest that CG artists spend some time looking at the real world, or even photographs, to try to understand the way the world is. You can get some wonderful ideas from nature. :) :)
As a simple and easy experiment, you might start by trying to mix sections of your shadow pass with an AO pass using various settings or masks, and see what you can come up with.
Now I could have sworn I did something on UV passes, but I guess not. Looks like I just mentioned what you can do with them, but never did an example.
So what is a UV pass? It's basically, again, what the name says. UV information for an object. You use grayscale values in the R and G channels of your image, and those values define the U and V coordinates of the surface of the object. And once you have that info, it's pretty much the same as having the UV map in your 3D program.
Now, as is often the case when you are doing 2D post work on images rendered in 3D software, using UV passes might not be as good as doing it directly in the 3D app. Because typically all you're doing in 2D is using a texturing tool which replaces the texture with a different one. However, keep in mind that all it does is change the texture, but you still have to worry about shadows, colors, bump, etc. Which means you're going to also need other render passes to make it all seamlessly integrated. But still it is a very useful tool.
Next I'll put together a render pass/composite example using a compositing app like Fusion or Nuke, since it's a lot easier in one of those apps to do stuff like this. For me, at least.. :) :)
But the basic process involves taking a UV pass, a beauty pass, and maybe some Object ID masks, and bring them together into a Texture tool, which uses the UV data and the mask to apply a new texture on that surface.
Which reminds me...I have a question for the experts :) :)
If you look thru the list of render passes in Carrara, you might notice there isn't a "Bump" pass. Interesting don't you think ? :) :)
Just about all of the other surface attributes have separate render passes, but Bump isn't there.
Can anyone explain or guess why? :) :)
If you get the correct answer you win the Carrara Forums Scrumtrulescence Award. :) :)
HI Joe :)
Sorry for the delay in getting back.
Sometimes there's more important issues
To re-cap where I left off,. and to clarify.
We were discussing the fragment passes,. and you asked for an example of how I'd used it,. so...
Rather than say (I've used it to adjust the edges of an anti-aliased object when compositing) ..Which is what was mentioned as it's intended purpose.
I offered an example of an alternative (and admittedly strange) use of that pass,
that was obviously confusing, so forgive me.
I'm well aware of the Velocity pass, and what velocity means.
I presumed (wrongly) that you would know that by this stage, ...My bad, again.
The reason I didn't use the velocity pass for that specific scene, was quite simply that it wasn't usable, it didn't give the desired results.
It seems like there's some Cycling of the colour values in the pass when there is a lot of objects travelling at the same time towards the camera.
I again presumed (wrongly) that you'd see that in the four frames I posted.
So there we are,.
Hopefully that makes it a bit clearer,. perhaps not,. I'm presuming again.
Why no Bump pass ....
Bump is a lighting effect,. ....It's not really there.
The information to rebuild the dark and light areas is contained in other passes.
The Shadow and Specular passes should contain the dark and light information to rebuild the bump effect.
Hope that makes sense
:)
Congratulations !!! You're on the right track with that answer. You can't really isolate the effect of a bump or displacement effect in a single render pass. It can affect the normals, the diffuse (by visually displacing the color/diffuse information when simulating a 'bump'), the shadows, the reflections, etc. So it becomes more of a challenge to do postwork on the bump/displacement information.
Okay, back to the UV pass and how you can use it in your compositing process....
Below is an image from a compositing program showing how you can use the UV pass from your render, combined with a new texture image you want to apply to the surface of one of the objects in the scene.
The image below shows the "flow" of the compositing nodes. Basically, looking at the flow from left to right, you take the UV pass, which has the UV information stored in the R and G channels, and combine that with the new texture map, using what's called a "Texture" tool. What's shown in this flow is the first stage of the process. You basically apply the new image (in this case a "white plaster" image) to the entire scene. As you can see, the output of the texture tool shows the new image on ALL surfaces. Not quite what you want... :) :)
Which is where the Object ID pass comes into play. That pass is shown in the upper left of the image.
One important side issue I want to mention....
It is VERY important, when doing compositing, that you know and understand what is in all of the channels of your images. And also it's important you understand what channels each tool needs to do its job. And often you need to re-configure your input images in order to make it all work.
In this case, the Texture tool requires that you move the UV information, which is residing in the R and G channels of the UV pass, into some auxiliary U and V channels before going into the texture tool. In this case you do that with what's called a "channel boolean" tool.
Confusing, but extremely useful.
Next what I did is to apply the new texture ONLY to the big LED screen on the wall.
You can do that a couple of ways, but it generally requires some sort of matte which isolates, in this case, the screen so the Texture tool will only apply the new texture to the screen.
Some apps allow you to take an Object ID pass, which includes multiple objects (like the Carrara pass), and automatically matte any object you want. But in this case I took the Object ID pass and selected only the screen object and made a matte manually.
As you can see, the result is that the new texture is applied only to the screen.
And once you have brought in the final beauty pass and merged it with the results of the re-texture, you get the starting point for your new image.
Keep in mind, as I mentioned before, you now need to consider how you will integrate this new texture into the scene. You need to consider colors, and shadows, and reflections, etc. But once you have this simple flow done, you can easily step through various texture images and decide in real time which you want. And that is extremely useful.
And another concept to consider: REUSABILITY. Once you have developed a flow like this one, you can save it and reuse it in the future, only changing the input images as needed.
And by merely changing the Object ID matte, you can apply the new texture to, say, just the girl's dress.
In this case I also multiplied the AO pass, using a matte for the dress, to give the new texture some shadows.
The possibilities are endless....
And for anyone interested, this image shows the composite flow along with the output and AO pass images