Do you try to avoid the Chinese Room?
polycounter lvl 11
There is a thought experiment done by a philosopher named John Searle to demonstrate the difference between a conscious being and a computer. He places a man in a room and has him follow a long list of instructions for converting the English language into Chinese. He can convert Chinese, but he doesn't truly understand what it is he is actually doing. He just followed a bunch of rules that relative to his understanding are 100% random. So, why is this important to a modeling forum?
More often than not I have noticed people both beginner and the highly talented alike do the same thing. And it appears to be universally accepted among 3D crowds. For example, you are supposed to use all quads when modeling, but very few people I know personally actually know why. Heck, I watched a guy stream last Tuesday and he intentionally left an n-gon in his model because it smoothed better. Yet he knew this was ok because he understood the logic behind why certain polygons are acceptable vs. not acceptable. They just know it's "just something you're supposed to do." Also, very few people know why you should triangulate the mesh before exporting to xNormal. They just know it's something you're "just supposed to do." I also see people deleting Maya preference folders when things go belly up, because that is "just what they are supposed to do."
You may be asking "If the end result is the same, why would it matter?" I have seen people who use hardcoded rules without understanding actually use them incorrectly. For example, people will triangulate the high poly for an xNormal OBJ export in addition to the low poly and cage. Sometimes they will triangulate the cage seperatly from the low poly and wonder why they get vertex errors. I sincerely believe an understanding of how triangle stripping and vertex re-indexing works can save the modeler some time. Also, several people were really excited about Crazybump because it uses a "magic algorithm" to convert photos to normal maps. This "It just works" mentality breaks down when you figure out how the "algorithm" actually works. Algorithms are ways of logically expressing certain behaviors, not some black box Harry Potter wizardry. For example, Crazybump will have distorted normal vectors if the photograph has some kind of biased light direction. And more often then not, photographed textures do because these photos capture the imperfections of nature.
I also believe it is a good idea to try and figure out what's actually causing the problem rather than just carelessly killing the preference list folder. Because more often than not I have noticed these problems are the result of bad practices that are repeated over and over again rather than "ooops, I guess their's a fluke in the Maya code. Gotta delete the prefs." It's been almost half a year since I have had to delete the preference list folder on my personal computer because the core understanding helps me avoid problems in the first place.
You can probably guess that I'm not a fan of Chinese Room thinking but I'm also not nearly as high of an authority as some of you guys on the subject of 3D. So what do you guys think? Is the Chinese room ok so long as it gets stuff done, or do you believe the Chinese room thinking is bad practice that could have been avoided?
More often than not I have noticed people both beginner and the highly talented alike do the same thing. And it appears to be universally accepted among 3D crowds. For example, you are supposed to use all quads when modeling, but very few people I know personally actually know why. Heck, I watched a guy stream last Tuesday and he intentionally left an n-gon in his model because it smoothed better. Yet he knew this was ok because he understood the logic behind why certain polygons are acceptable vs. not acceptable. They just know it's "just something you're supposed to do." Also, very few people know why you should triangulate the mesh before exporting to xNormal. They just know it's something you're "just supposed to do." I also see people deleting Maya preference folders when things go belly up, because that is "just what they are supposed to do."
You may be asking "If the end result is the same, why would it matter?" I have seen people who use hardcoded rules without understanding actually use them incorrectly. For example, people will triangulate the high poly for an xNormal OBJ export in addition to the low poly and cage. Sometimes they will triangulate the cage seperatly from the low poly and wonder why they get vertex errors. I sincerely believe an understanding of how triangle stripping and vertex re-indexing works can save the modeler some time. Also, several people were really excited about Crazybump because it uses a "magic algorithm" to convert photos to normal maps. This "It just works" mentality breaks down when you figure out how the "algorithm" actually works. Algorithms are ways of logically expressing certain behaviors, not some black box Harry Potter wizardry. For example, Crazybump will have distorted normal vectors if the photograph has some kind of biased light direction. And more often then not, photographed textures do because these photos capture the imperfections of nature.
I also believe it is a good idea to try and figure out what's actually causing the problem rather than just carelessly killing the preference list folder. Because more often than not I have noticed these problems are the result of bad practices that are repeated over and over again rather than "ooops, I guess their's a fluke in the Maya code. Gotta delete the prefs." It's been almost half a year since I have had to delete the preference list folder on my personal computer because the core understanding helps me avoid problems in the first place.
You can probably guess that I'm not a fan of Chinese Room thinking but I'm also not nearly as high of an authority as some of you guys on the subject of 3D. So what do you guys think? Is the Chinese room ok so long as it gets stuff done, or do you believe the Chinese room thinking is bad practice that could have been avoided?
Replies
On the art side, I like to know what's happening so I'll mess around and see what breaks and what doesn't. Especially with game art, to understand what's happening with things like tangent bases or CrazyBump takes a certain understanding of how programs work and what capabilities they have, of vector maths and how the results are used. I didn't start with any of that knowledge but I've gained some by looking into how things work (there's a lot of things that are still way beyond my comprehension and I'll take the Chinese Room approach to that until I do).
There are a fair few artists who don't have that understanding and who balk at the idea of even trying, it's daunting; code and complex maths is hard stuff to grok. That doesn't make them bad artists. Although it's likely that taking the time to understand that stuff would make them better artists - but that's a fair undertaking for someone who's not so technically minded.
Canning the preferences is the easy fix. You'd be wasting potentially hours trying to figure out what it all means and what's going wrong just to surgically fix it. It's probably a net gain in time spent over time to just wipe it and start afresh every so often. There are diminishing returns for time spent trying to understand something and that time could be better used doing the thing you'd set out to do in the first place.
And, honestly, a lot of that isn't the responsibility of an artist, especially in a production environment where things often have to be "good enough" rather than "perfect".
Basically "I don't know — that's how things are done around here"
I personally encourage everyone to stop and ask a lot of probing questions whenever they run into that mentality. That's not popular in Chinese culture and its a big stumbling block to creativity and ingenuity. I personally thank the cultural revolution for that but... that's another thread all together...
I don't like calling them rules, they are more like helpful suggestions from people who beat their heads on their keyboards and figured out some general ideas that MIGHT help you avoid all of that head on keyboard time. But they can just as easily cause a monumental amount of HOK time if you don't understand why.
To be successful it is almost mandatory that you not just learn the "rules" but that you learn why the rules are created. Until you understand why, you will run into many road blocks and make a lot of work for yourself.
Learn everything, ask questions, find out why and do whatever works. You learn why, so you know when the rules can be broken.
Forgetting why or not passing along the reasoning behind the rules, is one way societies end up with some pretty stupid rules that actually end up hurting them more than helping.
Just to make sure I'm doing my part...
Quads
Using mostly quads are generally easier to work with, but there are times when being dogmatic and following that as if it was a universal law will be a wild exercise in futility. It's fine to use tris and ngons if it works.
Triangulation
Triangulate models, because invisible edges can be flipped when importing/exporting which changes the angle of the normals and causes seams and shading errors. If you make all edges visible (triangulate) you physically define the edge and the application can't flip it. Search for "tri stripping" if you're interested in learning more.
Triangulation
Is doubly important when baking normal maps or exporting a mesh that is intended to be normal mapped. The algorithms for generating tangent bases used to bake and render normal maps (almost universally) work based on triangles. Different triangulation will give different results, breaking your normal maps.
It only really matters if aiming to become a tech artist / programmer, or if want to one day be in charge of other artists as a senior/lead.
Point: Knowing more than you need to about a process will take precious time from actually doing things. You'd get the knowledge anyways if you persevere and learn on the job, so getting in depth on technical points can divert your attention from more important things like artistry.
Counterpoint: Muzz is nuts deep in color theory and it allowed him to make Color Constructor; a really unique and useful tool. So knowing your shit is never bad, and can open doors that you wouldn't know existed.
This alone is a problematic statement. Different modeling tasks and end products require different approaches.
Overall, I think it's pretty obvious that understanding the technicalities behind this field is crucial. Some artists might be able to go by without being curious or lacking a certain critical eye allowing them to spot issues, but sooner or later it eventually backfires, especially in a team environment. As a matter of fact, pretty much all of the best modelers out there do care about all these technical details very much, because they know from experience that it can affect the quality of their work.
So I think that's a bit of a moot point. Don't waste your time overthinking whether or not one should care about these details, but rather, spend your energy on investigating them precisely, either on your own or by asking specific questions here that everyone will benefit from
I bring this up because it's something that has held back a lot of beginner users, yet somehow more advanced users can get away with it. So weather or nor Chinese room thinking is relevant is a discussion that can benefit everything. The Chinese room experiment is a different way of thinking rather than just a "How do I do a specific function in specific software?"
Have I just been lucky?
It happened mutliples times that I would do something without understanding the foundation and purpose of my action, solely going for the ''end result'' and depending on how important that knowledge will help me progress, I'll then take the time to think If I have to learn more about it or not.
If we were always thinking on why we do certains things a way or another, or how everything works around us and why, we'd be all philosphers-scientists...
I'm not sure what you're getting at to be honest. The examples you used as something "a lot" of people don't know about are pretty obvious and have a direct visual impact on what you just did. Are you talking mathematical forumulas? Because that's kind of outside the scope of a focused artist, more of a technical artist. And further from the technical artist is a programmer. And then the programmer has a lead programmer and so on. And all of that shit is based on years of work and research done by a lot of people creating libraries. So yeah, you can go super deep in knowing how something works but you'll probably need a few lifetimes to know EVERYTHING.
Not 100% sure if this is true as it's been a few years, but I think one of my old professors had said that when he wrote a book on using Maya that he interviewed and asked questions from people who worked directly on Maya, and not one person was able to know everything the software does and how because it's just so many years of people and work.
You could spend your entire life reading wikipedia and die not getting even close to reading it all. It's all a matter of what is important to know to you. But yeah questioning and learning is also good and I'm a 100% for so don't think this post is about just doing something because it works. That's why fundamentals are important to understanding and getting better
My 2 cents anyways.