Terrible Advice/Myth Thread.
mod
Alright, I think we've all seen them time and time again. Advice that gets recycled and overused so much, and without explaining why, that it loses all meaning. I want to start this thread to look back on these cliches and actually explain why, in some cases, they have merit, or to debunk them entirely. To participate in the thread, simply state a cliche, and do you best to explain in a realistic situation, how it applies.
I'll start.
NEVER USE TRIANGLES/NGONS!!!!!
This is pretty bad advice in almost every context, however it gets repeated probably more than any other single myth. Here are a few points that should cover most situations in which you should/shouldn't use triangles/ngons:
1. When it comes to lowpoly, modeling only in triangles tends to be a bad idea. Keeping a clean mesh with nice flowing quads can be essential to having an easy to edit mesh. However, many types of topology are best terminated with triangles, and forcing yourself to never use them is a wasteful and silly practice.
2. When it comes to highpoly sub-d work, it is true that it is best to avoid triangles or ngons in high detail areas where shapes are twisting and turning, on convex or concave surfaces. However, it is perfectly fine to use them on flat surfaces, or unimportant areas that are likely never to be noticed. Ngons can actually resolve better in some special cases than quads as well, so experimentation is key here. Again keeping all quads for the sake of doing it is just silly.
3. When it comes to creating cages for sub-division sculpting (zbrush, mud, etc) this is where this "Myth" actually has merit. Triangles can cause weird issues and bugs in sculpting apps, and sculpting detail on triangles can result in bad visual artifacts as well. When it comes to sculpting, its best to use sensibly flowing, even shaped quads, wherever possible.
4. Ngons can be give unpredictable triangulation, so using them in excess, and not cleaning them up before the model is finalized can give you some problems. However, being anal retentive about forcing quads while you're modeling is only going to slow you down.
I'll start.
NEVER USE TRIANGLES/NGONS!!!!!
This is pretty bad advice in almost every context, however it gets repeated probably more than any other single myth. Here are a few points that should cover most situations in which you should/shouldn't use triangles/ngons:
1. When it comes to lowpoly, modeling only in triangles tends to be a bad idea. Keeping a clean mesh with nice flowing quads can be essential to having an easy to edit mesh. However, many types of topology are best terminated with triangles, and forcing yourself to never use them is a wasteful and silly practice.
2. When it comes to highpoly sub-d work, it is true that it is best to avoid triangles or ngons in high detail areas where shapes are twisting and turning, on convex or concave surfaces. However, it is perfectly fine to use them on flat surfaces, or unimportant areas that are likely never to be noticed. Ngons can actually resolve better in some special cases than quads as well, so experimentation is key here. Again keeping all quads for the sake of doing it is just silly.
3. When it comes to creating cages for sub-division sculpting (zbrush, mud, etc) this is where this "Myth" actually has merit. Triangles can cause weird issues and bugs in sculpting apps, and sculpting detail on triangles can result in bad visual artifacts as well. When it comes to sculpting, its best to use sensibly flowing, even shaped quads, wherever possible.
4. Ngons can be give unpredictable triangulation, so using them in excess, and not cleaning them up before the model is finalized can give you some problems. However, being anal retentive about forcing quads while you're modeling is only going to slow you down.
Replies
BUT ISNT THAT TOO MUCH PADDING?
You can never have too much padding. Its an impossibility, what are you worried about, that your textures will mip too well? When you're doing a bake, the padding will stop when it hits another pixel that is being "padded", so your padding will simply stop inbetween two uv islands at the averaged point. For this reason, you can never have too much. AO, Normals, Diffuse, it doesnt matter, you can simply never have too much.
Sure, you could do something entirely excessive like setting 300 pixel padding on a 256x256 map, that would just be silly. But in all reasonable situations, you can never have too much and the goal should be to have very little/no, black/empty space on your image after the the bake.
1. I've seen this all too much, and normally, I would agree. But sometimes it just doesn't matter at all.
The polycount is just one part of an asset that you want to keep under control to help efficiency and usability, but triangles are very inexpensive compared to the alternatives. There are many factors to be considered when optimizing a lowpoly mesh outside of simply affecting the silhouette. Will adding some extra verts allow you mirror portions of a texture and possibly drop the size of the texture? Vertices can also be evaluated for their aid in lighting, collision, or deformation if the model is to be animated. Overall though, often details on a model or bevels on an edge can help the model much more than watching your tri count drop by 20 ever could, especially when it comes to normal-map driven assets.
While i generally agree with you here, lets try to add some discussion/reasons when bringing up a point. I've written about this specific point before, and what it comes down is contrast. You shouldn't put a very light object on a white background, and you shouldn't put a very dark object on a black background, similarly, you wouldn't put an object that has little contrast and medium value on a gray background.
Just saying, lets try to avoid "This advice sucks, because it sucks" and expand on why it sucks a little more.
And doesnt it it wreck the levels or something, you percieve colours in relatives, so its going to make the colours pop out more, look more saturated than they really are.
A lot of successful game art is hinged on how strong/interesting/recognisable a silhouette is. If any part of that silhouette ends up being in shadow, or textured darkly, and you composite it to a black background, you've lost your silhouette.
You don't want a white background, because throwing all that light at the viewer makes the low contrast/darker/subtler parts very hard to see.
I think the black background thing still stands under most circumstances, because people tend not to light their work well enough to be able to sit on black. IMO, that's the crux of it.
1. EQ can attest to that I'm sure but I still run into plenty of people that don't understand this. Unless you have the extra geometry to support the shape your trying to bake its actually a good idea to use hard edges so the degree of deviation between your low poly vertex normals and the high poly geometry isn't some insane offset.
It's true you get baking problems but you can just overcome this by using a cage. So really its not an issue.
Booleans Are the Devil.
1. Once again this is just a bit of bad outdated advice. Yes Boolean create a mess, and yes it takes time to clean up said mess, but it all too often takes much more time to try and get that detail any other way. Especially when dealing with complex curved surfaces.
I mean TECHNICALLY this one is true, but when people say it they mean 3d artists don't need art fundamentals and that's fucking bullshit. I mean if you really want to go the time consuming route and master color theory and anatomy and gesture through incredibly elaborate renders, bakes, and models, i guess that's cool, but you're going to learn how to draw/paint through the process anyway. The logical method is just to pick up a god damned pencil and become a skilled artist.
Yes but at the end of the day the advice you are giving is to understand silhouette, contrast, and lighting, which is very very different than "dont use XXXX color". Thats the whole point here, that spouting a canned line like "Don't use a black background" isn't actually helpful, unless you understand why. So while there may be plenty of occasions to not use a black background, that advice on its own isnt worthwhile.
Not really going to disagree that being a better traditional artist will help you overall, however i feel the day has past that someone could focus solely on traditional art and old school texture painting and things like this and make it in what is a highly technical and highly specialized job field. Painting/drawing skills alone are far from the only thing that makes a good artist good, when it comes to game art. Even someone with the luxury of spending all day doing "artistic" things like sculpting, needs to be highly technical to be proficient at his job.
Understating the importance of the technical aspects(proper lowpoly construction, proper baking techniques) embedded in modern day art production, I would say is misguided.
So wrong from my experience yet I bounced against a wall with so many people when talking about. Professors and tutors from university, schools before and even other students tried to tell me that I eventually have to decide on one thing anyway. But looking now back neither of those people ever had real experience in the new world of digital media nor did they really compare to what I like to do.
So from my experience: The more grounds you can place your feet's on the better as key skills that intersect with areas are in the end what many companies and studios really need.
You are only degrading yourself and limiting your future if you stick to 1 thing, except that is your thing and you are not that interested in other things. But in the end I think that a job that involves multiple areas is so much more interesting as just one focused key skill.
oh wait...
well, the people, they don't like filters and fx in photoshop, but it can do the trick and save some time, used with a bit of subtility (not the halo though)
Well thats kinda funny isn't it ? We agree that "BLACK BACKGROUND IS BAD!" is a myth to debunk, yet here you're just repeating it with "white".
My point is, a white background can be extremely useful too. Check out some 2D art on CGHub, and more generally, 3D game art coming from Asia overall. They are very often presented on a bright BG. I started using that recently and it is a tremendous help. It forces the artist to push color qualities much further, and thats a great thing if the art style of the game asks for it. Put a SF4 character on a white backgrond, I guarantee you it'll look fantastic.
Anyways, great thread. I would like to start a similar one, focused on debunking Object Space myths. I am really getting tired of them, and I feel that many, many headaches could be avoided if there was a better understanding of this tech amongst artists and programmers alike. What do you thing EQ ?
I thought Earthquake countered my point pretty well, which I agree with, hence no further rebuttal from me.
Also, calm down. Me offering a different perspective isn't the end of the world.
Yay for more knowledge!
Object space myths:
# can't animate:
we have animated characters with just vertex normals. vertex normals are stored in "object space". Object space normalmap is simply storing values per pixels. The same transformation, that allowed our vertex normals to look correct when animated, can be applied to our per-pixel normalmap.
# can't mirror:
as presented in the quality mode for object-space in 3pointshader, mirroring is just as possible. It only requires that we store the mirror "plane normal" (orientation of a mirror) per uv-chunk (in the end per-vertex). Once we know the reflection vector, we can correct the stored object-space normal map value and apply the reflection (ie. mirror) to get the right one at the mirrored parts. And because those parts were mirrored, there must be a mirror plane
---
it's all the same vector math, and the hardware is capable of the same arithmetic operations in both vertex and pixel level, therefore nothing stops you from doing it at one stage or the other. It's purely a matter of overall engine architecture, performance and hardware resources, where you decide to do your stuff.
in the end for lighting you need the normal vector and the light vector in the same space (coordinate system) however it's mathematically easy to get a vector from one to another coordinate system. And the same stuff that "deforms/changed" your vertices, can also deform/change your normals. The only difference is that normals are not "moved" but only "rotated" around, that actually makes it even easier for them.
Polycount has other threads that are in a similar spirit to this one, that might do well to be mentioned here.
Why HDR images can be uneccesary
Why Max displays normal maps badly in the viewport
There was a thread about extreme tri/poly reduction being unneccesary (and sometimes to a detriment) too but, as much as I've tried, I can't find the bugger.
more stuff can be found in archives
http://www.polycount.com/forum/forumdisplay.php?f=52
I think if someone is missing the silhouette on a black background that person is failing to back/rim light their object properly. I agree that black can rob and white can wash out. With any kind of art you need to take into account surrounding colors, values, the silhouette, the pose, the action lines bla bla bla. Most of the time when someone fails to light a object properly, its because they aren't thinking about a bunch of other things they should be. The color of the background isn't going to change that much...
About padding.
You're in danger of replacing one extreme with another.
Properly pad color groups, don't go nuts padding every shell within that group. Don't detach the hot pink lips, scale them up bigger than the torso and surround them with brown body parts then wonder why pink is bleeding into your brown. Group the colors accordingly and pad the groups appropriately. If brown bleeds into more brown no one cares, especially when the whole object at the lowest mip is 10px on the screen.
At some point you're limited to the texture size you can use, if you go too far with the padding at some point you start effecting the size of your UV shells. Players won't be examining the lowest levels of mip up close like we would. As long as it mips well in the game go with what works. Which you should probably be testing out when your blocking in your main color groups.
"JUST LEARN THE RULE(S) DON'T BOTHER TO LEARN THE LOGIC"
I'm not really sure why people hold to and pass on absolutes without knowing why. As artists, we need to learn the tricks, add tools to our arsenal and deploy them as needed. But its just as important to be able to create your own rules, not just memorize rules handed to you. The logic and reasoning behind the rules is way more important than just knowing them.
IF YOU ADJUST YOUR RENDERS/SCREEN CAPTURES IN PHOTOSHOP YOU'RE CHEATING BECAUSE YOU CAN'T DO THAT IN A GAME.
1. Every engine I've worked with all the way back to the ps2 has had post processing, so use it to your advantage for still shots just like you would in a game engine.
2. Bloom, vignette, chromatic aberration, noise, HSV, brightness contrast, colour remap, glow, these all make your final image look better when used in moderation and are standard in most rendering engines.
3. The first thing I like to do when achieving a look/theme is find reference of a photo, game, or movie and break down what part of the look is 2d post work and which part is 3d and how that relates to game rendering.
i actually got this from someone a long time ago, and for the longest time, i agreed. because let's face it, french people are.... french!
but nowadays i view pior the same way i view elephants and colour. all elephants are gray, but not all gray things are elephants.
pior is french, sure... but not all french people are pior!
no french people were harmed during the making of this post.... or elephants... or pior
Black and White Background:
Like already said you can do amazing things with them, sometimes i'm suprised of how much depth people can bring to something with a plain and simple white background
Michael Kutsche for example has two paints that illustrate the point so well
on white on black
Looking at reference is cheating!:
Most of the artist here on PC will probably know this all to well but this is a mindset that a lot of people starting out have.
On Mechanical work people won't be saying these things but with characters and creatures people sometimes feel like they are cheating.
You can't draw/paint/sculpt or model what you don't know! and you'll only learn how to create your own designs if you know how things in real life are constructed.
People need something recognizable to hold on to when they see a creature for example, so you'll need to look at how animals and humans are constructed.
Underlying structures are really important when it comes to humans, animals or creatures so learn your anatomy by using reference and artbooks like hogarth and loomis
you are so talented i wish i was that talented:
People aren't born with talent! People are born with a passion , it's the pursuit of that passion that makes people talented
Get off your lazy ass and start working :poly121:
I totally agree, but i think overstating them is misguided too. You have way more art experience than me so at the end of the day i'll cede to your viewpoint, but i would say there's nobody here making badass art who would be producing it if they didn't have very solid art fundamentals.
Some of you guys learned it while doing 3d, but the amount of arrogant ignorance of the color and anatomy and rendering knowledge from a lot of guys posting noisy zbrush and photoshop patterns is astonishing. Even semi-technical stuff like how to use a spec map properly to communicate the surface (seen gloriously wrong in metal textures in p&p all the time) is fundamental 'traditional' art knowledge.
Obviously if you're learning all of that doing 3d it's going to be just as good as learning it painting, but when i see people tell newbies outright not to practice drawing if they want to learn anatomy, or to just use filters to get the stylized look they're shooting for, i cringe.
I hate how often I get the question "so do you work on a mac?"
You could not have put this better. I've faked screenshots for past shipped projects in Maya and Photoshop touchups when our engine wasn't up to screenshot quality at the time. As it applies to personal portfolios whatever makes your stuff look great.
I think some people thing of adjusting something in 2d as cheating. While doing it in 3d (game engine, UDK, Marmoset, etc) is looked at as ok. Seems like a double standard if that's the opinion with this.
A lot of the time people ask for break down shots when you apply for things past the modeling dept. It just isn't possible to do incredible paint overs for every frame of a film or in a game so if there's a huge disconnect in your breakdowns between what you can do "in post" and what you can do in photoshop there are probably going to be issues. If they pretty much sync up and you just used photoshop because its faster, who cares.
I think if you're doing a shot for the sake of doing 3D go crazy, hold nothing back.
More so than just the black or white backgrounds, it's about anything that obscures your model being a bad thing if you're trying to show it off for feedback.
I also think that's where the 'rule' comes from. I've never heard anyone I even remotely respect say 'don't make art with black or white backgrounds', I don't think. it just sounds like stellarly bad advice.
This simply isn't a trick that always works. I'm not really sure when you should use it to be honest (I think for skin maybe).
I've just seen people make this comment too many times and then claim it would make things look more realistic.
Any examples of this working well? it seems totally bogus, for spec to replicate ambient you'd need to make it increadably blurry. Blurry spec = diffuse shading.
So why not just bake it into the Diffuse and use the spec for what its intended for.
As a matter of fact, "blue spec is wrong, real life doesn't work that way, don't use it!" would indeed be a myth to debunk :P
One thing I do disagree with, is justifying it because "it's the color of the sky". I just don't see it that way - spec highlights reflect point light sources aimed at them, not the overall surrounding tint. To fake the influence of the sky on a character asset one would be better off using some shader tinting blue whatever points up (world dependent. Thats different from spec, because spec is view dependent.) Or a blurry cubemap /spherical solution. Or even more basic : a blue ambient color in the shadows.
I do use blue spec on skin, always. It might be a stylistic choice, or justified by nature, it doesn't really matter. It being physically inaccurate shouldn't prevent one from experimenting with it!!
So far awesome tips!
You gotta understand why man. It's a bit difficult if you're not into the math behind it, so here's an illustration I whipped up:
The example highlights on the right are really burnt out, but with softer spec values it's just way more apparent, you'd get yellow-pinkish spec all over without using blue.
Well since a lot of game tend to be outdoors, the reflection would mostly be coming from the sky, showing the subtle blue colour in reflections. Even indoors were a lot of ambient light is entering is still blueish on the first few bounces where the light hasn't picked up other colours from it's surroundings.
Thanks for bringing this up EQ..
I'm not a character guy so I have to say for hard surface stuff, this myth is generally false. I've seen people do this without regard for materials and often it looks terrible. Instead, you should be using your specular color to separate materials, represent coatings, and inversing the color for white highlights (see why so blue article.) For hard surface, specular IS the most important map, so don't overlook it by making it all blue!
Now THIS is a myth...
You only need grayscale spec!
I hear this from programmers all the time. Color is essential for defining many materials, getting good variation and in general just keeping things in motion. Without colored spec, things like these shells wouldn't be possible:
About talent:
There is some kind of talent. It's true nobody is born with talent. I am not sure though, there are many theories. But what we call talent is (psychologically) an infant development and a development that takes place at childhood. It depends on what you did there. If you played a lot with the piano at your early years you'll be a talented piano player then or something that's similar to what you've learned by playing it. This is only a very rough example there are many more factors that have effect upon us. Education and other happenings that seem completely unrelated to the talent itself at first glance.
Ofcourse everybody can learn everything, but talent is what makes peoples learning and catching a meaning faster and easier. If somebody's over 20 (or 30 which is psychologically the age where quick learning fades away), trying to learn something new, he wont progress as fast as somebody with what we call talent.
This is my opinion. (:
And there are many different opinions.
For example to get back to "born with talent" hereditary factors that say people are born with talent.
Peace
Edit: And I think passion is the most important thing!
Well i was over-generalising with the outdoors, but then aren't most games set outside?
Overcast and midnight skies generally still give of different shades of blue light, probably even more prominent and saturated at night due to the sun setting.
I see what your saying about the different environment settings, but i still think blue is the most prominent colour due the the sky being one big ambient light.
But isn't the Sun a small point in space compared to the ambient light from the sky?
This is so crazy - the sky is larger than the sun - the sun is thousands and thousands of times brighter than the sky.
-'specular highlight' is a computer graphics thing - its a hack - it means nothing to physically accurate shading models - you have to keep that in mind when you are comparing what your shaders are doing to real life.
Well like you say, specular is a cheap form of reflection, so shouldn't specular and reflection colour be the same, since there are the same thing?