Photo-realism, what really makes a photorealistic thing, photorealistic?

I’m talking about RealTime renderers now. I’ve seen few games, which have perfect lighting and perfect materials, but still, they lack, photorealism, whereas few games they just look too photorealistic.

I’ll show you one by one and ask one by one.

First, as we all know, Unity has PBR (Physically Based Renderer), still, few things don’t look realistic at all like -

[ATTACH=CONFIG]514432[/ATTACH]
Also, another one -
[ATTACH=CONFIG]514433[/ATTACH]
(rendering from Unity Asset Store from A post Apocalyptic Asset)

As we can see, those images, look quite Beautiful. They have Perfect Lightning and Perfect Materials, but why they fail to look Photorealistic?

On the other hand, few screenshots of a game -
[ATTACH=CONFIG]514434[/ATTACH]
It’s from Steam, of Half-Life 2: Downfall.

It’s too looking photorealistic. Too has Perfect Lightning and Perfect Materials, but why it doesn’t look photorealistic to us?

On the other hand, there are games which look super cool and realistic, for example -



(Adam - Unity)

Doesn’t that look too Hyper-Realistic?

So, what really makes a photo-realistic rendering, really photo-realistic?

Uhh, I think it’s quite realistic

Physically plausible energy conservative materials, accurate direct & indirect lighting. Models and textures to scale. This isn’t rocket science, rather geometry and photometry. And above all else, plausible motion and physics. Doesn’t matter how good those still images look if things fall apart once temporal action is introduced the observers mind notices immediately.

1 Like

the most important thing is lighting.for example a hdri lighting helps alot, vs a simple sun or point lap (but you can archive good results too, if you know what you do).but for simplisity you cant get faster a “real light” as with HDRis.
then PBR shader,textures from Substance painter or quixel megascans ect.they look very realistic.
and as meantioned before,accurate models with details of course.
and some camera settings can help,for example depth of field,motion blur,lens imperfections,les flares ect.if you go for a cinematic look ,then color grading like orange peal could help.all this things in the sum, can help to archive a look ,what you want to archive.

but the most important things are lighting,materials,detailed models

Something ‘too clean’ will make it unreal.

One thing evryone forgets. Proper shadows and proper colors. Those two are super tricky to get right. Even with all technologies out there to make life easier you wont get machine to do it. You must fake it so it feels realistic to our mind.

I’d add: proportions that match a real world expectation, and the proper application of physics. Hence, buildings that would be functional in the real world, with correct door and window size and placement, solid supports for upper levels and decks, with no implausible-looking weaknesses in the structure. Normal wear and tear certainly adds to the realism, unless it’s a brand new structure. Perhaps a design flaw that needed patching.

One thing that often gives away an image as CGI is inconsistent detail. If one texture is limited by the resolution of the mapping rather than the render and another is not makes it look weird. For example, in the image above it looks like the concrete in the wall is HR but the street is not.

Of course, “there’s no such thing as ‘photo-realistic,’” because a photo is not at all realistic.

The term might be better put as, “it looks like a photo (in a magazine).”

And so, you should grab a copy of Vanity Fair magazine and carefully analyze all of the photos that occur before the table-of-contents. Every one of them appears “realistic,” do they not? And yet, none of them are.

In particular, you need to pay very close attention to the physical characteristics of the medium: in the case of Vanity Fair, film and the printing-press. For video, the video monitor or projector. The extremely compressed tonal-range, the effects of the gamma-curve (and the reasons for it), and what the image looks like on a thoroughly un- calibrated monitor. (Didja notice that the picture in every one of the TV’s on sale at WalMart is different, while every display in the Apple Store is precisely the same, no matter what kind of device it is?)

(Yes, your image is probably doomed to be viewed on the visual equivalent of “a cheap set of AM Radio car-speakers.” Or, gawd help us all, “earbuds.”)

Speaking to the film world, photographer Ansel Adams reduced the complex subject of densitometry to an easily-digested “zone system” that has influenced photographers ever since. It applies to video, too. Ansel wrote a series of books, and you should read them all, even if you never make a single image using light-sensitive pieces of plastic. Every medium that we use to express our work has physical characteristics that must be understood, and that can be creatively exploited.

A “photo-realistic image” is not “realistic,” but it does “look like a professional photo.”

getting a realistic cg is very hard it needs time, experience and lots and lots of trial and error and/or a really eye for detail,

for me this is my list to make photo-realistic or at least

1.lighting is the primarily main thing to make it real.
2. material and texture
3. the model
4. camera

even you use this vray, corona, maxwell, redshift, octane, and other engine or cycle you can’t make a photo-realistic if you really can’t its just depend on your skills and eyes and beside there is no magic button to make a box looks more like a box with candy inside it takes time and courage.