Visual effects supervisor Eran Dinur on his new VFX text, including an exclusive excerpt and a special discount code for befores & afters readers.
Visual effects supervisor, visual effects trainer and author Eran Dinur’s follow-up to his fantastic The Filmmaker’s Guide to Visual Effects is The Complete Guide to Photorealism for Visual Effects, Visualization and Games. In it, Dinur presents a range of practical insights and perspectives on accomplishing phototreal 2D and 3D content.
The book is, as the author explains in an interview with befores & afters, not just for visual effects artists, but also for those working in games and visualization. Find out below what went into writing The Complete Guide to Photorealism, what Dinur thinks people might mean when they say something ‘looks too CG’, and also read an exclusive excerpt about the concept of ‘detail’ in CG content.
b&a: What made you decide to write this new book? What was the thing that made you think about a follow up to the previous work, and that it could focus on this photorealism side?
Eran Dinur: For the first one, I saw it as a book that was aimed at people who work with visual effects, but are not necessarily doing visual effects. So, directors, producers, editors, cinematographers, and a lot of the people who’ve read the book are from that world. But what surprised me is how many VFX artists actually read the book and in how many VFX courses in colleges and art schools it’s being used. This really wasn’t intended but I realized that it works very well as an overview of the whole visual effects process that also gives the perspective of the filmmaker.
So, if you are an artist, let’s say you are a rigger and you focus on doing rigs for animation. You may have a few notions of what else is going on, but you may be too busy just working in your own field to completely understand. So, the first book gives you that whole perspective.
I then realized that there are artists, not necessarily just in visual effects, but anyone who’s using CG or just digital tools to create imagery, where maybe I could delve deeper into the process and look at one big aspect that’s important across all levels, and that is, photorealism. And in visual effects, photorealism is crucial because what you do is, you put stuff in the frame and you pretend that it was there in front of the camera.
If it does not feel like it was shot in front of the camera like the rest of the footage, it looks fake, it doesn’t work. So for visual effects artists, the notion of photorealism is embedded into the work. I remember when I was just starting and I worked at a visual effects studio, one of the artists there used to stand behind people, stare at their screen and then suddenly say, ‘This doesn’t look real’, and walk away. And every time he did that to me, even though I knew it was a joke, a repeated joke, it still felt a bit hurtful. And to this day, a lot of time you get those notes from a client that says, ‘This looks CG,’ or, ‘This doesn’t look real.’
The essence of the work is to make things look photoreal. But then I also thought about visualization, especially architectural and product visualization. When you think of the fact that most of the IKEA catalogue is done in CG and you wonder, but they have the furniture, they can just set it up, right? Then you realize that doing a real photo session is going to cost them a lot more than doing it in CG and gives them a lot less flexibility. CG is a tool that’s being used to show products and in the architectural world, obviously to show things that do not exist yet. So, you want to show a client what the building would look like, what the apartment will look like.
Indeed, photorealism has become very important in architectural rendering. There’s a lot of disagreement about whether it’s good or bad for architecture, because many architects don’t like the idea of selling the client something that is real, but then eventually will look different because they didn’t have the exact materials and so forth. But the fact is that a lot of people who work in architectural visualization strive to create realistic renders. It’s part of their work.
b&a: And I suppose it simply is now possible since the renderers available just ‘do’ the work of rendering photoreal as a ground truth.
Eran Dinur: Yes, and now that’s also a thing in games. Obviously there are games that don’t try to be photoreal—there’s stylized games—but there is a branch of games that try to give a feeling of realism and these are getting better and better by the hour with all the technology.
So, I realized that writing about photorealism could be geared to a much wider audience than just visual effects. I mean, practically almost every high level renderer that we use nowadays is physically-based, which means that the renders are now as close as they ever were to actually simulating light behavior rather than faking it. There were times where computer power was just not strong enough to actually simulate light or what really happens to a light photon when it hits a surface and bounces. So, we faked it. We used point lights and spotlights and ambient light emitters. It’s actually amazing what quality of work was done before you had true global illumination and shaders at the level that we have now.
Nowadays it’s probably easier than ever to create a photoreal render. You put your stuff in there, your geometry, put a sky dome or put a directional light for sunlight, then press render. And if your textures are good and the shaders are set correctly, you’ll probably get something that looks pretty real. The problem is that the clients and the audience have also gotten used to it! If we look back at stuff that was done 20 years ago or 10 years ago and we look at it again today and for a lot of it, we feel like, ‘It’s not as good looking as we thought. Back then it was amazing.’
b&a: Let me pick up on something with that. A client might say, ‘Oh, that looks too CG.’ Or an audience member says it, or a film reviewer says it, or a person on Twitter says it. Now, in my view, there’s so many reasons why a shot or a character or a creature might not look right. It could be shot design, it could be the budget of the project, it could be the way it was filmed, so many things. I guess I wanted to explore with you, when someone says, ‘Oh, that looks too CG,’ what do you think they’re thinking or saying? I feel like people so easily say that these days, and I’m not sure they really know what they’re saying, to be honest.
Eran Dinur: Well, that is exactly what the book is about. And of course there’s no single answer. Also, there is no absolute truth. People may look at stuff and one person will say, ‘This looks realistic to me’. And the other person will say, ‘No’. And the problem is that most of what we do in visual effects and also in games, and to some extent in visualization, is the accumulated work of different people working in different tasks. If you take a typical CG shot in visual effects, you have a modeler, you have a texture artist, you have a shading and lighting artist, You have a compositor, you may have a matte painter, all these people join together to create something. Each one of them in each one of these steps can do something that will either improve the photo quality or kill it or reduce it.
So, it’s very hard to put your finger on what exactly is wrong. And that is the challenge we’re dealing with every day. Me, as a visual effects supervisor looking at shots, and basically every artist, every CG, and comp supervisor, we all try to put the finger on what needs to be changed for it to look more photoreal. So the question is, can you have more tools or knowledge or thought processes to help you get there? And I think that you can.
I’ll give you an example of defocus. First of all, in the first part of the book, I look at general concepts, and one of them is the difference between how we see things and what the camera sees. There is a reason why we say photoreal and not just ‘real’. Because we’re not trying to emulate how we see the world. We’re trying to emulate how a camera sees the world. And the question is why? And the answer is that we can only see what’s in front of us. So, we can only see another person’s vision or a place that we are not in right now through a lens. We’ve gotten used to that for every kind of imported reality. Everything we see that is not through our own eyes is on a screen, on a TV or a monitor or on paper. So, we’ve got used to the fact that all of it is through the lens. So when we see something that looks like it was shot through a camera, through a lens, we say this looks real. Our eyes see differently.
You see, our vision gets blurry towards the edges. If you put your finger to the side of your head, you’re going to notice it, but it’s going to be very blurry. Cameras don’t see like that. And if we try to emulate this with tilt shift, people look at it and say, ‘Hmm, that looks weird.’
I’ll give you another example; dynamic range. So we, in our eyes, we do HDR photography. We lower the highlights, we raise the shadows. Our brain does that automatically. But when you take a camera inside an apartment during daytime, the windows are completely blown out (or the interior is too dark) because cameras can only expose to a defined level.
Now, people came up with HDR photography where you shoot different exposures and you combine them. And when we look at it we usually say, ‘Hmm, that doesn’t look real. That looks fake. This looks weird.’ Because, we are expecting the camera look. So, I think that understanding what the camera sees and how we see is the first step towards photorealism. The best artists in visual effects, especially compositors, are the ones that have experience in photography.
b&a: I’m not a visual effects artist, but I was thinking, as a visual effect supervisor, it’s kind of intuitive when you’re working on a sequence or a shot, about how you approach it. How tricky was it for you, for the book, to come up with these topics and sort of break down all the ‘topics’ that relate to why something is photoreal?
Eran Dinur: Very, very tricky. In fact, the book has four parts and the second part does not talk about visual effects or 3D or 2D, it only talks about the real world. And that part, even though a lot of it we kind of know intuitively, that part is really more about physics and optics, which for most of us, me included, is very, very hard to understand, unless you’re into physics. I see three numbers and I get dizzy. So how would I understand this? Physicists look at light, they know how light works. They know what color means, but they know it through a scientific mindset, which is numbers and equations.
I was trying to find a way to explain all these things to people like me who understand visually, but don’t want to read an equation. And so that part was very hard. Putting the different subjects, trying to hit the different subjects and also describing them. The third and fourth parts were a bit easier because they are actually about working in CG and about working in 2D. These are areas that are part of my daily work and it was easier to organize them.
b&a: I imagine you had to find this delicate balance of being accessible, but also still talking to a technical audience.
Eran Dinur: Yes, exactly. So, the first book is really talking to people who may be missing on the most basic terminology. In this book I had in my mind that an artist—a CG artist in games or in VFX or visualization, already working with Maya or Max or Nuke or After Effects—who is looking to delve a little deeper in, but without going technical, would like this. First of all, I tried to be software agnostic because I think that’s besides the point. You want to learn software, you don’t need my book for that. What I try to focus on is the ideas in general. So it is maybe a bit more technical or a bit deeper than the previous book, but still, it’s fun to try and explain things without talking numbers. There are people who can do that better than me . I want to talk as an artist to artists.
b&a: There are a lot of images in the book. Do they come from projects you’ve worked on or are they from bespoke imagery that you and any collaborators created?
Eran Dinur: The bespoke ones, first of all, are the illustrations. I realized that there was a lot of, especially in the second part that talks about light and light and surface interaction and tries to explain the physics behind it in a visual way, that obviously illustrations help with. And I realized that I want them to first be in the same kind of style, and be very clear and concise and just illustrate the point. So that’s why I approach my friend and colleague, Ben Zylberman who’s a matte painter, and we worked on it quite a lot.
Then there are photographs. There’s two types. There are the ones I took myself to illustrate a certain point. So I just went around with my camera, and shot certain things that I wanted to show and people probably thought I looked a bit weird going down and shooting my own shadow or shooting into the bushes. But this was to show things like direct shadows or ambient occlusion. Or, taking my DSLR and a phone and shooting into the sun to show how the same lens flare looks very different through different lenses. Some other photographs were just things that I’ve shot in the past that I thought could illustrate a point. Like going through photographs and saying , ‘Oh, this one is perfect to show aerial perspective.’
Sometimes I would take these photographs and gain down on them to show things that you wouldn’t see normally. Say, if you dehaze a shot that has a lot of fog or haze in it, you see the buildup of the water and dust particles or the combination of them when you increase the contrast or do some artificial dehazing.
And then the last type of images is artwork by different artists. I just went around the internet and looked for things that impressed me as very photoreal and then approached the artist and asked them if they were willing to contribute some artwork. And so there’s some really interesting work there from very different kinds of areas. There’s architectural renderings, there’s stuff from VFX, personal projects, visualizations, and others.
b&a: I’m curious as you were writing this, were you writing about some topic and then actually it came up as part of a visual effects shot you were working on? Did you come across a shot that wasn’t quite working and you went, ‘Oh, I just wrote about that last week. I’ll try and implement that’? Or did anything like that happen as you were writing it?
Eran Dinur: Yeah. Actually, well, I wrote it in a little over two years. If you ask my friends, colleagues at Brainstorm, they might say that sometimes I became really annoying because in the middle of a review I would start going off on a tangent explaining and they’re like, ‘Okay, can we get on with the review?’ Once you think about these subjects , you start seeing them everywhere of course in real life, but also in the work you do. Sometimes it went both ways. Sometimes I was looking at a shot and suddenly it came to my mind that I need to write about this or maybe I wrote something incorrectly.
I’d start looking into shadows, and some artists might say, ’These are just shadows. Do we really need to talk about it?.’ But I think it’s interesting to look at shadows and try to understand, for example, what affects the softness of the shadow. Most people would probably say, okay, the scale of the light source, but there are actually other factors, too. Like, how far the light source is from us, how far the object is from the surface that it cast shadows on. Just trying to write about this also helped me understand a lot of things that I just felt intuitively or wasn’t quite sure about.
b&a: Do you think you learnt anything new about CG or compositing techniques that you hadn’t necessarily come across before, while writing the book?
Eran Dinur: The truth is I learn all the time in my line of work. I now work as the head of the 2D department FuseFX New York. Just working with comp supervisors, almost every day I learn something that I didn’t know, or didn’t think there was another way of doing it. We do so many things that we are kind of used to do in a certain way, and when you put them into words and try to explain them in any organized way, you have to learn them deeper. So that example of the shadows that I gave you, it was something that I never quite thought of and when I went into it, it was a headache just trying to figure it out precisely. I literally did experiments with the spotlight on my cell phone and moving objects around.
The ‘physics’ part of the book was the one I needed the most help and for that I’m really grateful that I had a friend who’s a scientist in Paris and he introduced me to a friend of his who’s a physicist who works mostly on lasers and light in general. I sent that person a bunch of questions. It’s interesting how we talk about things like ambient light, or direct light, or photons, as part of our work, but not always have a clear understanding of them. What is absorption? What is reflection exactly? What’s the difference between reflection on a nonmetallic surface and a metal? And then when you try to understand it a little bit differently—why do metals not really have diffuse color or diffuse light, for instance?
It helps you understand why when you pull your slider in Arnold from the electric to metallic, then the diffuse color suddenly grays out. There is a reason for it, but then you have to understand how the electrons in the molecule work and how they react to photons. So I got this help and I learned a lot from it.
befores & afters readers can use code BEF21 to get 20% off the book via Routledge.com, until 31 January 2022.
Join the VFX community by becoming a b&a Patreon...and get bonus content!