Face "Golden ratio"
Mendoman
Posts: 404
Hello all,
last weekend I tried to make a new model, and like every time before, I got something I thought was nice, but not quite the way I wanted. She looked ok ( well, it's really hard to make unattractive Victoria 7 spin off ), but not what I wanted. So I did what I usually do, and tried to render her from close and far with real lights, and then looked the pictures, not the render pop-up window. Also I tried different expressions, and took a break, and after few hours hours looked the pictures again. But no matter what, I just couldn't quite figure out what I didn't like. I just had that eerie feeling that there was something not quite right.
So finally I dived in the wonderful world of internet, and started looking for tips. I went through sites, and apparently we are programmed to recognice human shapes very easily, and since we use faces to regocognize each other, we are experts to see even the smallest details. Finally in my search for a perfect face I stumbled in a very fascinating site, that tried to solve the concept of beauty with mathematics ( http://www.intmath.com/numbers/math-of-beauty.php ). If I understood it correctly, it uses fibonacci numbers, and it's all about symmetry. There's even has been some university studies, and according to them, most people find symmetrical faces beatiful. Although the site noted, that not everyone that most people find beautiful follow these rules, I still decided to run some tests. I modified my character to follow those "magic" numbers, and here's the result:
The one on the left uses Victoria 7 ratios, and the one on the right uses "golden ratio". So what's your take, are those golden rules hit or a miss? Also, as a side question, what do you do, when you think that something is not right, and you just can't really figure out what?
PS: We really should have a stickied thread called "Help me, what's wrong with my model?" 
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
Aahh Fibonacci gets a bit too much credit here. The golden ratio has been around and has been used for far longer than that.
The ratio is more against symmetry than for it, as it divides a surface into a long and a short side. 2 parts of the ratio are never the same. You might have heard of the 'rule of thirds' that's used to teach the basics of anything artistic like photography, framing of a painting, use of negative vs. positive space etc. While the golden ratio is not exactly 1:3, the rule of thirds is a simplification of the ratio for the masses.
Symmetry is artistically on the boring side of things and if you make a symmetric face, you at least want to keep the freckless, moles etc asymmetric to break up the overall symmetry.
In your images, you would also have to either widen the face to match the width suggested by the mask or scale down the mask a little and rearrange face parts again. Every side of the golden ratio depends on another side, all measurements are in relation to another one from large to small, so if you miss one, especially one of the larger ratios, all the other ratios would have to be recalculated, or they're no longer in the golden ratio.
The most basic purpose of the rule is to teach people to avoid putting the main subject of a picture in the center, and while they're at it, they may aswell put it where it's proven to look best.
At the same time, you avoid symmetry. If you have a wall and need to build a door in it, the best spot for it is not the center of the wall making the left and right sides of the wall symmetric, but to one side, leaving one side of the wall longer than the other. Nature works in a similar way, while the golden ratio is more or less everywhere, there's no symmetry (unless you go down to some fractal patterns I guess).
Your whole body down to the fingers can be divided into the golden ratio as can a tree, a plant,a leaf... which leads to the flipside of it, which is... trying too hard to find the golden ratio everywhere, and trying to hard to apply it to everything, always: http://lolworthy.com/wp-content/uploads/2015/12/trump-golden-ratio-fibonacci.jpg
It's been used knowingly in architecture, painting, sculpting etc long before Fibonacci and is the reason why your screen is 16:9. Your credit card is also in the golden ratio, and if the designer had some education, he will have aligned the text on the card based on the golden ratio too, probably using the long side of the ratio as the distance from which the text starts from the left, and the short one for the distance from the top or bottom border of the card... or some variation of that.
Anyway, sorry for the lecture...
I think that if I was trying to create a pretty face, and by that I mean a face that *I* (with my tastes) perceive as beautiful, I'd probably start with looking through the average faces you can find on google... interesting way to see what ethnicity you like best, too :)
I also found averaged faces from models based on industries (food, fashion etc) here. All the faces are mostly neutral and facing front which makes it easy to see which type of lips, eye angles, face shape (round vs. angular) etc you like best, since those are things that the golden ratio mask won't help you with. Ideal proportions don't help as much if the eyes and lips chosen don't play well together and such.
I'd find the mask a bit too abstract to work with as beauty often comes from a collection of features like eyes, nose, freckles, lips, bone structure that are missed with the mask.
If you want to find what is off with your face, I'd take the averaged face from all those images that you like best, and overlay your face in Photoshop on top of it, that should tell you how far off from your ideal you are.
If you resize and squish your face to match the averaged photos better, you can hide your face and only show the lips or eyes of it on top of the averaged faces (hope that makes sense). If the results look off, that may help you determine which features of your face are being problematic.
And just because... I made some quick Rule of thirds and Fibo spiral meshes that people should be able to place in front or parent to their cameras in DAZ to help with the composition of their renders. If you type rule of thirds into google, you'll get the basics of how it works, can just look at the images.
I'm still learning Daz Studio, but you should be able to place the spiral or grid on a viewport the way you want to render (or parent to camera I think ?), and use another viewport to move the pieces around until they're aligned to the grid or spiral in your render viewport. The meshes come with a few morph maps to control thickness and the outher frame.
Mesh in FBX is here.
I don't mind the lecture as long as it's educational
I'm still learning this whole artistic stuff, and every piece of information is welcomed. I have experience from 3D modeling, but I have like zero knowledge/experience of photography or anything even remotely artistic, so that rule of thirds mesh is a really nice helper. I actually already took a quick peek from Google how it works, and hopefully I'll get to try it in practice soon.
And when it comes to faces, there might be some magic in those golden rules, but there's still lots to be done. It certainly doesn't help, that we people like different things. For example some like bigger and some smaller noses, so that "perfect" face might be impossible to create
Still, your idea to use those average faces to help finding problem spots was really interesting, and I really need to test that next. Thank you for all your help.
Also, it's important to keep in mind the lens your looking through and the way the face is lit, as both of these can drastically change both the actual shape of the face in the image (in the case of the lens used), or our perception of the shape of the face (in the case of the lighting).
- Greg
The "average faces" that strangefate_asdf mentions has been studied for a while in psychology:
http://www.faceresearch.org/students/averageness
I really found the Daily Mail link strangefate provided to be interesting. It's not quite the same as the average face, because it shows that corporations have a certain "look" in mind for their spokespersons, but it's applicable. I wonder if some enterprizing DAZ employee has done the same thing with, say, an 'average' Victoria 7 face based on a large number of characters sold at the store. :)
I think you can use golden thirds for general composition, but I'm not sure it has much value in making art of "beautiful" people. There are aspects of both before and after images that I'd consider elements of "beauty," but neither is of a beautiful face. There are conflicting elements to the image, some of which have nothing to do with proportion. For example, there is a very stark shadow to the right of the model's nose that is distracting and gives false visual cues. It makes her nose look far fatter than it really is. Your tests should really be done using flat lighting so that tonality is left out of the equation.
Here eyes might be considered quite large for the width of her face. The ears are very pronounced, with an "elfism" that detracts from her overall appearance, and make her look "funny" or unappealing. Many art books would consider the level of the eyes too loo in comparison with the top of the ears. Generally, they promote a straight line from top of the eye to top of the ear.
Consider that while symmetrical faces might be a benefit for models and actors -- Tom Cruise is said to have an almost perfect symmetrical face -- that's far different from beauty, or even good art. One of the most regarded pieces of art, the Mona Lisa, uses an asymmetrical mouth, and it's her mouth that is often considered the cornerstone of that painting.
Excellent overall summary in your post, but just a note about this comment: 16:9 monitors are actually not golden rectangles. The original 16:10 monitors were (or very close to it). Golden ratio is about 1.618, whereas 6:9 is 1.78; 6:10 is exactly 1.6.
Reasons for changing from 16:10 to 16:9 have more to do with manufacturing costs, and is also considered a better compromise between the standard aspect ratios of original filmed content, which ranges from 4:3 to (about) 2.35:1. (The caveat here is that widescreen movies have seldom been at this aspect ratio; over the years it's changed, due to modifications in the so-called "sidewall" content that accompanies many films, such as the DTS audio timecode track. Today it's really 2.39:1.)
It's curious that there has never been a standard film aspect ratio based on the golden rectangle. Europe's 1.66:1, also found on some now-defunct formats like VistaVision, is about the closest we get. For decades, most films shot in North America, or by Hollywood production companies, have used either 1.85:1, for matted "flat" projection, or so-called 2.35:1, for anamorphic projection.
I think all of this points up that we don't need, or at least no one's ever really missed, using the golden ratio for watching movies and TV.
Mighty Mysterio - That link as a nice average face generator, thanks!
Tobor - You're correct. The 16:10 ended up as 16:9 for all the reasons you name, just didn't want to read up on the reasons and elaborate, was a long post. A lot of things only come close to the ratio but are off for one technical compromise or another, and in some cases, people don't think about the ratio and end up with similar proportions because they tend to feel better.
And btw Mendoman, that feeling that something is not quite right, is probably the 'uncanny valley'. You can google up and read up on that too, I think it may make you feel better about your achievements, since it's still very hard to avoid unless you go for a cartoony or clearly CG look.
Some things that Tobor pointed out would probably help with the issue. If you like big eyes and you want them angled to look more elfish, you have to find the sweet spot that's not too large and too angled to not look really human anymore. It's very easy to start tweaking features and end up with something that doesn't feel quite right anymore.
The Uncanny Valley is a pretty interesting theory. I don't have any data to support this, but I don't think all of us are equally susceptible of the uncanny valley. It might be somewhat similar to how some people are just unconvinced by classic stop-motion animation while other people used to watch Ray Harryhausen films and were amazed at how lifelike the monsters were. Another analogue would be the believability (or not) of high-end wax figures at Madame Tussaud's.
so you just use mathematics to please everyone ?
as you say "most people find...."
I would second what algovincian says about the lensing used.
I'm seeing a noticeable fish eye effect which is throwing all your proportions off.
Check the focal length on the camera, put it to about 110 to 'flatten out' the image.
James Gurney did an interesting series of essays on the Golden Mean. You may find it useful.
http://gurneyjourney.blogspot.com/2013/01/mythbusting-golden-mean-part-1.html
Wow, lots of responses and sorry for a late reply, but I've been busy studying rule of thirds
Well, I might have worded it poorly in my opening post, but my idea was to talk about if these "golden rules" really work. Like is there any magic in those mathematical numbers like fibonacci. Still of course I'd love to hear constructive feedback, since I'm still not entirely happy with that example character, so thank you Tobor and strangefate_asdf. The ear height probably is too high like you mentioned. I think I'll try to change her ears back to human form, and then position them, and then finally stretch them to elf length again. Otherwise I think character's profile would look quite weird if ear tops would be positioned at the same height as eyes. Either eyes would be too high, or ears too low because elf ears are longer than human ears ( heh, at least in my fanfiction ).
Oh, and thank you for that "Uncanny Valley" information. I've never heard of that before, but it's good to know I'm not there alone
.
Also I'm sorry about the lighting, it was just thrown there for a quick example renders for those "golden ratio" pictures, but now that you mention it, I should have used flat lightning to see differences better. Unfortunately lighting is still in my todo-list of things to learn, so didn't really think about that, but I'll get there....eventually.
I'm not quite sure what you mean? English is not my native language, but I wrote it as well as I could that according to university studies most people.... And well, the whole point of this thread was mathematics and golden rules in faces, and like I said in my second post, I think there might be something, but it's not the whole truth.
Interesting read, and quite funny that even professors/experts disagree how these golden rules work. I also found an interesting read https://www.sciencedaily.com/releases/2008/04/080404122139.htm. Apparently computers can make our pretty models soon, so all I have to do is wait few more years
The concept of Uncanny Valley is largely aimed at presentations of entirely or mostly human characters. Harryhausen effects or other stop motion wouldn't be a part of it, just like the heavy use of CG characters in an otherwise live-action film doesn't seem to bother most people. The new Jungle Book is testament to that. The human character is real; the animals and environments came from computers. Who cares -- it's made close to a billion dollars.
The results have been polarized for some completely "realistic" CG films. Polar Express or Adventures of Tintin are two examples. In the case of Polar Express, the artistic style of the movie followed that of the books, so it made some sense. In Tintin, the art style of the film was completely different from the original books; they just thought they'd push the envelope. That said, both of these films were box office successes, so at for the producers, no one is complaining. Reviews from critics and some movie-goers mentioned the uncanny valley, but since these films were of high technical quality, whatever the hesitation was, it seemed to have been quickly forgotten by the audience.
I wonder if we sometimes limit ourselves as to what is realistic in art in contradiction to the actual diversity of human forms that we observe when we engage the world, or watch sports, or go to the movies over a significant period of time. The actual ratios of human limbs or facial features are quite diverse, even if the overwhelming number of people have ratios within a narrow band. Consider, for example, that the leg inseam of the winner of an olympic gold medal swimmer was the same as the leg inseam of a gold medal distance runner, yet the swimmer was about 7 inches taller (one specific example of two particular people). Thus, the ratio of leg length to torso length was very different between the two. Their specialness also made them more likely to be seen by large audiences. You are more likely to have seen them on TV than to have seen me on the street, yet my ratios may be more realistic, as defined by artists. Similarly, the ratio of arm length to height of professional basketball players is not generally consistent with Da Vinci's circle, and again they are seen by large audiences on a regular basis. Are they unrealistic? It is one thing to critique an art piece on grounds of what artists are taught and expect, and quite something else to assert that the anatomy ratios are not realistic - if unrealistic means containing limb or facial ratios that someone would not observe within a few months or a year of walking around a densely populated city, attending public events, watching TV, etc.
Here is a TED talk that often sparks continued conversation.
https://www.ted.com/talks/david_epstein_are_athletes_really_getting_faster_better_stronger?language=en