Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
My experience has been that a depth pass will come out all black or all grey if the render camera can see "to infinity" - that is, there is a straight line from the camera to the horizon or off into space. I think it tries to calculate the distance between the furthest point and the closest and divides up the grey tones between them, but if the furthest point is infinitely far away... well, relatively speaking everything in the scene is at one distance compared to infinity so it all gets the same shade.
I've solved it by rendering the depth pass separately - after you render your other layers, turn off all your lights (just because the depth pass will render like lightning that way) and insert a plane perpendicular to the camera, large enough to fill the render and just beyond your most distant scene item. Then render just the depth pass and it usually works.
Good tip, but I think it's not entirely the solution. The depth pass for the space image worked great, even though there was nothing behind it. The depth pass for my War of the Worlds image came out all black, despite the fact that the terrain blocked the horizon.
My guess is that the War of the Worlds scene had hundreds of trees, each at a different distance (depth) relative to the camera. Perhaps that exceeded Carrara's 8 bit depth capacity?
The city sphere also has hundreds of replicants, but the depth is rather limited in 3D space.
Then again, I've done very simple images and had the solid black result.
I returned to that site today. It has been changed. Now the author wants $31.50 per chapter. I'm glad I visited that site, and downloaded the book, before it was changed.
This is a fantastic thread. Great job explaining the basics of compositing. Incredibly informative and very well written.
At first, based on your example of the producer changing his mind and the dog in the background, I was afraid you were taking a "fix-it in post" mentality. I was relieved when I saw you most certainly were not taking that approach at all, and you were using it as an example of "Shooting-For Post".
The difference between the two can sometimes be a difficult concept to wrap your head around, especially some actors. At least in my experience.
I'm really enjoying reading this and hope you post more.
Excellent thread! Thank you, Joe!!!
Marek
Joe, I'm just getting to the point where I need to start thinking about compositing and how it's done with Carrara and your tutorial is a great overview and gets me going in the right direction. Thanks for sharing this with us!
So I understand that you're trying to use audio as an example to explain compositing, but I'm a pro audio engineer, and the term "compositing" is never used in the pro audio world. No nobody records a song with one microphone (anymore anyways), most of the time a band will play together simultaneously recording onto multiple tracks, with multiple microphones. Once this first pass is done, individual players can either redo their parts, or add more harmonies or solos or whatever to the original track. This is called overdubbing. Then finally the process of bringing all the tracks together with eq and effects into the final product is called mixing.
Perhaps you can convince your peers to adopt the more appropriate and descriptive term, because it really is "compositing"... :) :) :)
In any case, I hope the analogy is still effective even if I was a little loose with the terminology. I figured that hrowing terms like "dubbing" in there might muddy the waters.
Thanks for the input.
Sorry, I was real busy lately and got diverted from the discussion.
Now, one other concept that might be useful to understand involves delving a little deeper into the grayscale render passes we discussed previously.
As you can clearly see after rendering out a bunch of different render passes from Carrara, grayscale images are HUGELY important in any compositing/post-work project. Each pixel in the image provides information to your compositing app which can be used to isolate or apply various effects to your images. So if your grayscale image has pixel values ranging, say, from 0 to 255, those numbers tell your app a specific value that each pixel represents.
Maybe it's a distance from the camera, ranging from 0 to 255. Or maybe an identifier that identifies individual objects in your scene. So a grayscale value of, say, 0 for a pixel tells your app that that underlying area of the image is "object 0" of maybe 256 individual objects in your render.
But more importantly, it can signify ANYTHING you want it to.
Keep in mind, it's just a simple grayscale image. And in fact, you don't necessarily need a render pass to generate the image. You can generate the grayscale image yourself very easily. Or you can modify an existing grayscale image to customize it to do what you want. And that is a VERY powerful concept.
So, for example, let's say you have an image of stuff in the foreground, and stuff in the background. And you want to simulate a lens blur, to give your image a cool "depth of field" effect. How do you do it? Well, you make a grayscale image, and the grayscale value of each pixel of that image tells your compositing app how far away each underlying object in the image is from the camera. And generally, a "distance" grayscale image is darker for foreground objects, and lighter for background objects. Light pixels denote objects that are far away, and dark pixes denote objects that are close.
And if you look at a lot of distance grayscale images, you notice that they often look like a simple black-white gradient. And if you're at all familiar with any image manipulation apps like Photoshop, you know that it's real easy to generate a simple black-white gradient. And then maybe it occurs to you that you can very easily generate, and/or modify a simple grayscale distance image in Photoshop. Make a new layer, choose the gradient tool, and drag it across so that you generate a simple grayscale gradient. And voila, you have a "depth of field" grayscale image that you can bring into your compositing app (either as a layer or a new channel), and use that directly to generate a depth of field. And since that is so easy, you can then get fancy with different gradient shapes and generate an unlimited variety of depth of field effects.
Now, here's a challenge, since my last one was so widely received with such excitement... :) :) :)
In real-world photography, there is a very cool depth of field effect you can use to convert a clearly large scale image into what appears to be a photograph of a miniature. It's called "tilt-shift" photography. And you can very quickly generate the necessary DOF gradient using a simple Photoshop gradient, and apply that to your Carrara render if you want to simulate the same thing.
So the challenge is for someone to do a little research on this cool effect, grab a photo from the web, and manually generate the appropriate grayscale "depth" image to convert your Carrara render into something that looks like a photo of a miniature scene.
Bottom line, the purpose of this exercise is to get people thinking about how simple and powerful these grayscale images are, and how you can have ultimate flexibility in obtaining any result you want if you just use a little "outside the box" thinking.
And BTW, one other concept that this exercise illustrates is how simple camera effects, which only occur due to the specifics of the camera and lens you are using when you take an photograph, can drastically change your perception and interpretation of what you're seeing. It illustrates the power of very minor and simple changes in your images, and how those minor differences can greatly affect the viewers' perception
Okay, well, to get things moving on my challenge I did a quick search and came up with this photo of a stadium. As you can see, the shallow depth of field in this image converts a wide angle photograph of cars and people, taken at a distance, into what appears to be a tight closeup of some miniatures. Pretty cool, and just one example of how you can use a simple grayscale gradient to do postwork on an image and generate cool effects.
So if anyone wants to do a very quick challenge, all you need to do is find an appropriate photo on the internet and generate the necessary grayscale image and apply it to a depth blur in your favorite compositing/graphics app.
And since most of my challenges are greeted with something like "well, why don't you do it for us?", I spent all of 5 minutes grabbing a photo of a car show and applying a lens blur in PS so it looks a bit like a "tilt-shift" photograph.
Some real world stuff at the moment Joe, but I'm actually dedicated to doing something with your first challenge, I actually have files from a multipass render on my hard drive waiting for when I get a breather......this weekend I hope, I'm keen on keeping up with you because I use Fusion and I know there's a lot I can learn from you. Thanks for keeping it going.
And here's the simple grayscale gradient used to generate the depth of field blur in my last image. No, not too many beers, I was just in a hurry and it came out a little slanted... :) :) :)
As I mentioned before, there are umpteen gazillion things you can do with your images via compositing and other postwork. All it requires is that you first know what you want to do. After that it's pretty easy, once you've learned the basic skills.
And I also mentioned previously that one of the most common is modifying colors. Why would you want to modify the colors of your images? Well, again, there are umpteen gazillion reasons to do it. But basically the main reason is to make your audience respond the way you want them to. You might want them to feel like your image was like an old, turn-of-the-century photograph, so you tint the image a sepia (brown) in post, as well as applying other effects.
One extremely common example is tinting your images with a certain color scheme that the audience might respond to.
As an example....
If you've watched any feature films in the last 15 years or so, you've seen a very popular trend in color schemes. Many films, whether you've noticed or not, are colored with a blue and orange tint. You've seen it, guaranteed, but maybe it didn't stick in your mind. But it's EXTREMELY common. Why do they do it? Because the audience likes it. Blue and orange together are pleasing to the eye, and they evoke certain feelings of warmth and excitement, while at the same time being cool and relaxing. Go figure.
So the effects guys pass their images thru a color modification in their compositing apps. And if it's a 3D film, they can even access the main render lights and tint them either blue or orange.
Another easy challenge for those who are interested is to check some trailers for some blockbuster feature films in recent years and see if you can spot the color scheme. Shouldn't be hard to do. Start with something like "A Bug's Life", and you'll see a bunch of films just dripping with the teal/orange color scheme in recent years. And once you see the application of that color scheme, try to apply it to some of your images to "punch them up". Or whatever.
On the other hand, if you've seen so much teal/orange color scheme in recent films that you want to puke, then I'll understand if you sit this one out. :) :) :)
In any case, it's just one of a gazillion reasons for some sort of color correction in post.
Regarding the video color grading with orange and green, I find the films that do this seem to suffer from a visual sameness that seems uninspired, tiring and obvious. I've see it in the biggest blockbusters to smaller, more intimate films. Somewhere,a skilled cinematographer is curled up in a corner feeling depressed because all the work he did with exposure and lighting to achieve the director's vision on the day of filming, went right down the crapper when the guy with the computer got his dirty, color correcting mitts on the footage.
Don't get me wrong, sometimes the look needs to be adjusted for legitimate technical reasons. I think they're going overboard with it, and it makes it look as if 90% of the films are lit and filmed by the same people.
I'm having difficulty wrapping my head around the color correction idea, I mean I can totally see color correcting to make something more blue or more orange, but I'm having trouble picturing how you could make something both more blue and more orange at the same time. Probably just need to break out GIMP, load a photo and experiment so I can visualize what this would look like. Very interesting that films do this, I never noticed before.
Really appreciate the detailed posts, Joe. Very helpful. Will be looking into it more.
Often, what is done is to make the background a deep, saturated blue, and the foreground (generally the characters) an orange. If you watch many 3D films like the one I mentioned, they separate out the various lighting elements and modify colors in post. Keep in mind that some 3D renderers allow you to isolate the effects of each individual light on the scene, so you can get a pass for each one of your spotlights in the scene. So for example you might have a key light tinted orange, while the background lighting is tinted blue. Of course, there are many variations. But guaranteed, if you watch films or even TV (like the dime-a-dozen dramas about crime investigations, etc.), you'll see the constant use of deep blue background lighting, and orange-tinted foreground lighting.
Also, the on-set lighting when filming feature films, for example, might be tinted blue/orange, and the job of the color correcters is to punch up the saturation or balance or whatever they decide. It's very difficult to get perfect lighting on a set, so post work is mandatory for stuff like that.
And my apologies for even mentioning this. Once you recognize the effect, you'll probably see it everywhere, and from now on that's all you'll notice when you watch TV or movies... :) :) :)
If you ever watched Joss Whedon's Dollhouse, they used the trick with the gradient for the starting of the show. The exaggerated DOF, gives it a miniature look. It's interesting to see it motion.
Now, one other compositing/post-production application, which is extremely popular and useful for everything from feature films to TV shows to just about anything where 3D graphics is integrated into live action camera footage, is called "camera tracking" or "match moving".
Let's say you have some live action camera footage, shot from a camera that is moving (either on a tripod or handheld or using a camera stabilizer like a Steadicam), and you want to insert and integrate some 3D elements into that footage. For example, you have some live action footage you took with your cellphone of Martian droids marching down your street, and all you need is some Martian droids you rendered in Carrara to complete the production.
To do something like that, which is done all the time in the professional world, you have many challenges. One of the biggest challenges is that, since your video footage is in motion due to camera motion (in this case, very bouncy cuz you were holding your cellphone in your hand), you need to duplicate that motion EXACTLY in your 3D animation. You want the 3D elements line up exactly with their intended postion in the video footage.
And one way to accomplish that is to analyze the motion of individual pixels in your camera footage and try to figure out, using some extremely complicated math, what the camera motion was during that time to make the video look that way. And if you get that, it's fairly simple to send that XYZ motion tracking information into your 3D renderer so your Carrara camera motion will be exactly the same as your cellphone.
You need to look at various clearly defined "landmarks" on the video image and track their motion. The more points you can track, the better accuracy you can obtain in guessing the camera motion which caused that motion. And once you have a lot of points tracked exactly, the software can calculate what the XYZ camera motion was during that time, and you can keyframe all of that tracking info.
Of course, the motion information for the 3D camera is just part of the problem. You also need to simulate the lens characteristics of your cellphone camera, adjust color, clarity, focus, DOF, etc., so that your 3D render looks like it came from a cellphone. But once you have your tracking data, you can use your compositing app to bring all of those elements together and make them appear to be a seamless video of Martian droids marching down the street.
So I have another challenge for everyone: do exactly that. Take your cellphone or whatever video camera you can get your hands on, make a short video of whatever you want, and use one of the free camera tracking apps to generate tracking info. Then use one of the free compositing apps to bring that all together. I believe Blender has a free match moving feature which does a nice job.
And while I'm thinking of it....
One more EXTREMELY popular and overused technique in recent years is autofocus.
Professionals realized long ago that if you want people to really really believe that something actually happened, the way you do that is to simulate someone shooting a live, handheld video, camera shake and all, even though your scene is totally computer generated 3D. And we've all seen it in a zillion TV shows and feature films. If you have a completely implausible scenario, the sure fire way to make it seem real is to have someone record it on their cellphone. :) :) :)
So, aside from duplicating the camera motion using camera tracking, a sure sign of a video recorded by a cheap consumer video camera is the autofocus.
In professional shoots, they hire a cameraman's assistant to make sure the camera is always in focus. Because out-of-focus images are generally not desirable. The guy who does this is called a "focus puller". So in professional shoots you never see a fuzzy picture, unless they want you to.
However, if an amateur uses his cellphone or a consumer video camera, it has a motorized autofocus which shoots a beam of light to measure distance to the subject and automatically adjust the camera focus. But it takes time for it to figure out the correct focus, especially if the camera is moving. So often with handheld shots you'll see the camera autofocus searching for the correct focus, and the image briefly goes in and out of focus. And viewers associate that autofocus action as a real-life video. "Oh, it was shot with a handheld cellphone or something? Yeah, dude, it actually happened, no question". Go figure. :) :) :)
So the job of the compositor might be to simulate that autofocus action in the CG render. Or they may choose to convert the live action video that was shot professionally on a rigid tripod with perfect focus. If they want to stress the "real life" nature of the video, the compositor can turn a well focused, fixed camera, into a autofocused handheld camera, thereby giving the video a real world validity in viewers' minds.
By the way, if you want to see an example of the "motion tracking/camera shake/it has to be real" concept, the first film that comes to my mind is "District 9". There are a million examples of this technique, but for some reason that sticks out in my head. Some really well done compositing in that film when it comes to "it has to be real". :) :)
That one is great example of that. Great compositing of the ship as well, in that it actually looks as if it is enveloped in the humid and smoggy atmosphere.
Cloverfield really leveraged the handheld look.
And BTW, if anyone has noticed....
A common theme running thru most of the effects I've mentioned that are performed by compositors and VFX guys revolve around the perceptions, expectations, and likes/dislikes of their viewers. They like blue and orange tint. They associate a handheld camera look with reality. They associate different camera lens effects with different perceptions about the subject. And on and on.
It's something I've mentioned here a thousand times, and something that isn't intuitively obvious to those who have never produced images for others. But a large part of understanding compositing and postwork and just about any type of art is understanding what the viewers want and need and expect. If you want to tell them a story or make them feel a certain emotion or believe what you're telling them, you need to speak their language. And if they'll tend to enjoy your production more if it's tinted blue and orange, then there's nothing wrong with tinting it blue and orange. Or if they associate a handheld-looking video with reality, and your goal is to make viewers believe what you're showing them is reality, then there's nothing wrong with "faking" a handheld camera look.
The more you understand your audience, the better you can be at telling them a story and imparting an emotion.
And also BTW....
You and I and just about everyone in this forum, and anyone who watches feature films and TV shows and music videos, are pretty much the same. We *tend* to have similar expectations and perceptions and likes and dislikes. Sure, there are exceptions. But while it's easy to dismiss the stuff that filmmakers and compositors do for their audience, keep in mind that *WE* are also the audience. And most of us, whether you realize it or not, tend to be no different from the majority of the audience.
We associate a handheld camera look with reality and immediacy. We tend to find blue and orange tinted images to be pleasing. And on and on. No, these aren't rules (heaven forbid... :) :) ), but they are general tendencies which seem to work for a large segment of the audience. Including you and me.
So before anyone dismisses these techniques as "oh, it's for the feature film audience, I don't need to worry about it", keep in mind that if you want to feel better about the stuff you produce for yourself, you might want to consider employing some of these effects. It might turn a render that you think is "kinda cool" into something that is "wow, that's awesome!!"
But first you need to understand yourself, and your expectations and perceptions and likes and dislikes... :) :)
One more concept which might be of interest regarding compositing is that of "composition". While the words sound alike, they are two different things.
Any artist will tell you that a key element of any artwork they produce is composition. In other words, how will they arrange their image to best serve the intended purpose.
In deciding how to "compose" your images, you need to consider how the viewer "sees" the image. For example, when a viewer looks at an image, they tend to expect a focal point. They expect the image to lead them to the important point in the image. If the image is too cluttered, with no clear focal point, they might get annoyed and distracted, and lose interest in your image. Your image won't have the desired impact. If everything in the image is clear and in focus and well lit, the viewer can't find a focal point, and their eyes wander all over the image looking for what's important.
Very often I see renders with no clear focal point. Many hobbyists tend to throw a bunch of stuff into their scenes, and light it so that you see all their cool objects, and as a result the viewer gets confused about what they are looking at, and what is important. Most viewers, including you and I, are that way. We want to see a clear focal point.
So the artist needs to "lead" the viewers' eyes using various techniques. How you arrange the objects in your image. How you light it. There are many, many techniques to lead your viewers' eyes to satisfy their needs.
One technique used very often by compositors is the use of blur and fog. Blur and fog are used to make certain parts of the image less likely to grab the viewers' attention, so they will instead focus on the important stuff. Even a slight amount of blur or fog tells the viewer "Move along, nothing happening here", and the viewers' eyes move to the brighter, more in-focus areas of the image. And especially in moving images/video, it's very important that the viewers' eyes don't wander around the image too long, and possibly miss the important stuff.
There are many compositing tools to help with this. People tend to look for the brightest and clearest part of any image. You can drop the brightness of an area you don't want to focus on, or decrease the saturation of the colors, or place areas of the image out of focus, or add a slight layer of fog. But the challenge is to also make your effect not stand out or be too noticeable, or else the viewer might spend his time scratching his head wondering why the image looks "off", instead of focusing on the important stuff.
Again, it comes down to understanding the needs and expectations of the viewer. And as with so many other things in the CG world, the really important stuff isn't something you'll learn about in a software manual or video.
Joe, this is really interesting and informative. What I'd be particularly interested in is how to relate to Carrara and the various passes; what they do, how to use them to improve a render, how they interact and what blending modes are used.
I know this is a very big ask, but I'm sure you and the other experts can give good advice on this :)
I think that was one of my challenges....figure out what the passes are in Carrara, and how to generate a final image from combining the various passes.
I think that with a little thought it should be pretty easy to figure it out. For example, you know you need shadows, and specular, and diffuse, etc....just start out with the most simple passes (no reflections or GI or stuff like that) and figure out how to combine them to get the rendered image.
Which reminds me, I need to confirm that the shadow pass now works for me.