What is it that makes some scenes look almost lifelike, and others dull & flat?
cdemerit
Posts: 505
I've been looking at a lot of artwork lately, and it kind of struck me at how some is almost lifelike, and others are flat, even with a lot of detail. I know the Big three are Render Engine, Lighting, and Texture Quality, but is this really all the difference between a good looking scene and one that is almost cartoonie? Now I know 3Delight isn't the best render engine, but I've still seen some amazing work done in 3Delight, so I know it is possible.
But in truth, I'm just curious what your thoughts are on what factors can improve a scene's realism. (nor necessarily looking for Photo Real, just better than stick figures).
Post edited by cdemerit on
Comments
Any chance you have an example of what you consider "flat", perhaps along with another example of "not flat"?
One possible contributor: I see a number of human renders where the posing is very static. Many pre-made commercial pose packages use unrealstic and unnatural stances. When a photographer shoots a model, the model is almost always in motion. Made me think a great aniMate2 track might be of a model during a photoshoot. The poses would be much more fluid that way.
Evenly lit images, no matter how "realistic" are often considered "flat." I've played video games with visible polygons that were quite nice looking, and I've seen countless photographs that were uninteresting due to poor composition and lighting. 3Delight is capable, but the concepts are a bit more abstract than render engines like Iray and Luxrender. I certainly found it harder to get what I wanted out of 3Delight and often resorted to Blender Cycles in the past. Even then, the final render may lack the contrast needed to "pop" and require some post processing work in image editing software.
Greetings,
Just to throw a few random thoughts out there:
Honestly, I don't find the render engine (with any modern engines) to make a difference. You need to know how to light your scene given your engine, but it's the lighting that makes the difference, not the render engine.
-- Morgan
Repeating others, lighting is #1. It's amazing how much it can turn an image from flat and lifeless to intriguingly warm (or cold).
Posing. So many people think it's neat to texture emotionless dolls in some hinky posture.
I also like to tweak the models a bit to, at least, give maybe a few imperfections, age lines, asymmetric smiles, that sort of thing. Most real humans have asymmetries of facial expression, and it conveys a subtle sense of reality.
Well, a render engine can help, but really it is knowing how to use what you have. I have looked at some great renders by biased renderers and un-biased renderers. I have also seen some that are absolute crap by both types of renderers.
In my case, I want the scene to pop. I don't care if it's physically accurate. Truth be told, I like a certain realism, but I'm not a slave to it. I like to think of what I do as a stylized realism.
The thing that really helps a render are your shaders and lighting. They go hand in hand. Great lighting and lousy shaders look just as bad as lousy lighting and great shaders.
For me, the lighting needs to separate the primary subject from the other elements of the scene, while still being justified by the scene elements.I see so many renders with washed out lighting or flat lighting. The lighting, as I said, should provide separation, but it should also provide graduations in shadows, and texture details such as rough or smooth surfaces, specular effects, etc. This is where the un-biased renderers can really shine, but maybe require a little more work in a biased renderer.
The other factor in an image is the subject. I find Portraits generally kind dull. I respect the skill of the artist that can create a dynamically lit and posed portrait to be sure, but I much prefer something less mundane. That being said, sometimes I'll try a portrait render as a means of breaking a creative block.
Then you also have composition and other elements of the scene.
I noticed some problems when I happened to pick backdrops that were about the same brightness as the character or the character's clothing. Made for bad contrast.
Yep. Everything can contribute to the overall look of the scene. I think lighting and textures are mentioned most often because they are some of the biggest culprits for a lifeless looking scene.
I touched on it slightly above, but shadows are just as integral to lighting as the lights themselves. Dartanbeck, a friend of mine here, likes to call it, "painting with shadows."
Atmosphere also can create an incredible amount of life to an image by adding drama and depth- beyond what a simple DOF can add. That's one of the things I love about Carrara, the volumetric clouds. You can have minimal set and dramatically increase the perceived scale of the scene with some well thought out clouds.
Depth of field - focus on the subject
Distant haze- for those large open shots
Reflections & Highlights- from eyes to chrome
Shadows- probably the biggest thing for realistic. Ambient and direct
Camera position- looking down for large shots. Level with the horizion for the comfort shots. Or, looking up for the magestic shots.
Blur- If your subject is moving, give it a little motion blur or the background.
Everyone has mentioned lighting, but the one aspect of lighting that can make the entire scene jump out the page is the shadows,
Think about it. We're rendering 3d scenes onto a 2d medium - a .jpeg image. What gives an image the illusion of depth? Shadows, of course. 'Flat' lighting usually looks boring, but just chucking in dozens of lights won't help. Just one good strong light with well-defined shadows can do more for your scene than anything else. (Backlight is another nice effect, when used properly.)
Another good trick is making use of the camera lens. Most people use the standard Focal Length, but going for a dramatic wide-angle, like 20mm, can change a normal scene into something unique.
These are all basic photography techniques, and although it's not essential to be familiar with photographic jargon, studying photographic tutes is a great way to improve your eye and make your renders a lot better.
mac
While everybody is giving great thoughts, I think the key word in your post is, "realism" and that means materials, material, materials.. If you put garbage in, you'll get garbage out. It doesn't matter what lighting you use if the materials are one dimensional and cartoonish like many of the older materials availible. So if you are using older materials that lack details, it doesn't matter how much lights, shadows or depth of field you have.. it comes down to the materials and the amount of details it has.. Its the main reason that I consistantly argue for more detailed characters. The skin appearance should be exactly like the skin on our own body.. multi-colored and varied with hair, pores, nerves, veins, pimples, freckles... the better the materials, the better you can use all those other techniques to make that great looking material look better but if your starting off with lackluster materials don't expect much...
@ SnowPheonix, I can take a very high resolution photograph of the actual world, and yet it can look flat. I can take a screenshot of Skyrim that has depth despite running on an older game engine. I really don't see how materials have much to do with the reason that some images look flat, considering that one can't get better materials than actual objects in the world around us, and photos turn out boring more often than not. Yes, better materials can certainly enhance an image that already has great lighting and composition so long as that's what the artist intends to do. As an oil painter, I am also aware of how much depth can be conveyed using mere paint. Again, there are hyper-realistic paintings that feel flat, while some rather impressionistic works show amazing depth and atmosphere.
I think that the biggest problem with "realism" in CG is and will always be the uncanny valley. And renders that we think are awesome today will look dated and passe in the not too distant future.
I came across this yesterday. Uncanny valleys and PBR and polygons aside it all comes down to composition and lighting.
http://bensimonds.com/2010/06/03/lighting-tips-from-the-masters/
@Daveleitz You just hit the nail on the head... ""one can't get better materials than actual objects in the world around us"
Exactly the point I make when it comes to dealing with character skins... a point that is lost on many. You have to start with real looking skin from head to toe, including thumbprints, palm prints... every minute detail should be copied onto the characters skin so that what you start off with looks like a real person. Unless the details are in the original materials, you'll never be able to recreate realism with fake looking materials... notice the original author mentions "cartoonish" because cartoons lack details.. the reason this becomes an issue is because really good detailed materials take a long time to make and some people are skittish about using real life models but thats exactly what they should be doing.
They tried it back when the technology was still pretty far behind during V4 and that wasn't a boom apparently.. new engine, new character.. only a matter of time.... Have a great day :)
@ daveleitz, that 'flatness' that you describe sounds to me as lack of contrast (for the most part).
By far the easiest way to get rid of (some of) that is to adjust levels in a program like Photoshop.
How much effect that can have will of course depend on the level of contrast in the original image.
Depending on what your image editor is capable of you can add a levels-adjustment layer or work on the image layer directly.
It works best by far if your image has a 16-bit colourdepth for Red, Green and Blue, for JPEGs room for improvement will be limited.
Just move the left-most (darkness) slider a bit to the right, about 10 units for most images.
Don't overdo it, or the dark areas of the image become murky and solid black.
Then move the middle slider a little bit to the left, to about 1.05, to make the image as bright as it was before again.
Don't touch the right-hand slider, or you'll loose detail in the brightest areas of the image.
Very simple, usually very effective.
Cheers!
Erik
Yeah, I've noticed that even a tiny amount of Photoshop (or whatever) work can really help an image pop. I've started processing my images into a format to make it look a little line art/painted. Slightly.
A little more contrast and saturation can make a world of difference.
Here's an example from my webcomic. In the first case, yeah, I could spend a day or two trying to adjust the settings to get the saturation and color I really wanted. ooor spend a few minutes doing it in Paint.net.
I think expressions, poses, and eyes can make an image look flat and souless. Too many images have just dull-eyed females staring at the screen. Just does not seem real. Expressions and eye position are important to the composition of the scene.
You forgot the big one, "talent of the artist". You can give anyone a camera and they can take a snapshot of just about anything. But its just an image.
But a trained photographer will capture a moment in time, a work of art, a masterpiece. The difference is in the thousand small decisions that develop over time with training and practice. 3D art is the same. We get better by doing.
Please go easy on that saturation slider, unless your goal is a distinct 'artistic' look.
What I usually do is the opposite, I decrease saturation a bit, especially after contrast enhancement, because my default render settings (Octane) tend to exaggerate things from the start.
I mention saturation because a lot of contrast adjustments will end up washing the image out. As always, have to have a good eye.
Tips from one of the best arch viz artist around:
http://bertrand-benoit.com/blog/the-photographic-look/
Linear workflow (LWF), plausible lights and materials, and range.
Great link wowie, thanks!
It's interesting that depth of field and motion blur have been mentioned in context with "real and lifelike". As a long time photographer I believe in their use but they are artificial constraints of the photographic process. I agree that lighting is a main contributor but I personally believe the overall composition plays a bigger part. The whole scene has to be believable, even if fantasy. By this i mean internally consistent. Shaders make a big difference because we inherently know what cloth, leather, metal, glass etc look like in different lighting. This is a big part of why I like Iray. The other pieces are hair and skin. Too many of the hair products look like plastic or some strange helmet. I recognize the difficulty in creating the products.
In case nobody's mentioned it yet. Shadows and contrast are highest on my list. Contrast without shadows is uninteresting. Shadows without contrast are flat. Your shadows should be noticeable but not intrusive, and the image should have areas of dark and light and an over-all full range of contrast. But unless you're on the moon, your shadows shouldn't be black!
People who remember good black & white photography (probably those of us over 60) know about contrast making the image.
Excellent discussion!
@SnowPhoenix, I totally get what you're saying. Back before DAZ Studio had this wonderful new renderer, I spent many, many hours importing DAZ scenes into Blender and then creating my own materials using nodes. I experimented quite a lot trying to get SSS to work right when it was a new feature in Cycles. I think I got burnt out on it after a while, since it was rather time consuming and tedious. Of course, that "realism" bug is still stuck in my head, but at least Iray promises to make the image creation process more straightforward and fun now.
@Erik Leeman, Image editing is a fun process. I have a Wacom Intuos tablet and have spent many hours adjusting images in various programs. Even so, there are some images that just don't merit the time to spend correcting them. I strive to get it right in the camera as much as possible before attempting post work. Using RAW files from a good DSLR certainly helps the process. I can usually spot a botched photo by excessive post work used to "correct" it.
One last thing about flat images: The monitor that you use may not show the image in the same way to another viewer on another computer. Either calibrate your monitor(s) or view your images on multiple screens. What "pops" on one, may not look so great on another.
While motion blur is an artificial constraint of the photographic process, it is also an artificial constraint of the human eye. What we see in person we consider "real" to some extent (or at least some of us do). So when you see a prop airplane with it's engine running, all we see is a blur, much like a camera with a slower shutter speed. An aircraft flying without any motion blur to the prop would therefore look fake.
Flaws or effects associated with photography might also make a render seem more real, simply because in the past people tended to associate a photo with reality somewhat more than a render which we know is not reality. Of course photos aren't necessarily real, but they are real more often than a render. So there is a psychological aspect there, even if the flaw/effect (lens flare. depth of field, etc.) isn't "real", it is sometimes associated with the attempt to capture reality.
My belief is that to understand "realism" don't look at 3D renders. You're bound to get caught up in the technical aspects.
Instead, look at the painting masters of the Renaissance and Baroque, for starters.
One of my favorites is Caravaggio, a much-maligned but extremely influential Italian artist who was known for his realism in texture, light, tone, and pose.
Then there's Canaletto, who's early cityscape works especially show amazing detail in the poses of people in everyday Venice life. Even though the images of the people are small, it looks "real" because they are in active, dynamic pose. A famous example is https://en.wikipedia.org/wiki/Canaletto#/media/File:Canaletto_-_Bucentaur's_return_to_the_pier_by_the_Palazzo_Ducale_-_Google_Art_Project.jpg
The masters were masters because of their understanding of light. Another Caneletto masterpiece is this wonderful tone study: https://en.wikipedia.org/wiki/Canaletto#/media/File:Giovanni_Antonio_Canal,_il_Canaletto_-_The_Stonemason's_Yard_-_WGA03867.jpg
These examples, and others by the masters, showcase the many good suggestions here, but in a way where the qualities of good "realistic" art are more plain to see.
My advice to everyone: go to a museum! Study the works of those who came before us. They already figured all this stuff out.
Yep...practice, practice, practice. And then, practice some more.
Unlike when I was starting to dabble in photography, you don't need to spend huge sums of money on tools (cameras, lenses, etc) or supplies (film, processing/developing supplies). You can render until you fill your hard drive...then erase all the junk and start over.
Two words...Ansel Adams