This article is so long, it's been split up into 2 parts, you can find Part 2 here



Thank you to Nanaki-Murasaki for his efforts in helping write this!

---

Planning Stage

This is the most important phase of your production and often the most neglected. Most of the worst renders completely skip this step and dive straight into disaster.

You are lacking a purpose and intended audience
Before even coming up with an idea, it's good to think to yourself "Who is the story for? Why am I doing this? What do I really want to do?". A lot of people skip this step, thinking it's unimportant, this usually leads to nonsense like someone trying to test out a glass shader by making three characters, one of whom has a wine glass, or a person trying to pitch a B-Grade horror movie to preschoolers. Before you start anything, consider the who, what, where and why of your purpose:

Why am I making this?
Who is my intended audience?
Where am I planning to show this?
When does it have to be finished by?
What are the things I need to achieve?
How am I going to achieve them?

You wouldn't do something bright, cheerful and colourful if your audience was elderly antique collectors, similarly you wouldn't have fifty characters if your deadline was two weeks. Establishing these guidelines early will help your idea grow.

The idea is boring/unoriginal/too complicated/requires too much foreknowledge
A lot of good ideas can end up as bad renders. However, often you can start with a bad idea, and refuse to let go and give it a proper burial, which, by the time it's finally finished and produced, certainly smells like something dead that's been left in the sun for too long.

So how do you know if an idea is bad? Tell it to people who are part of your intended audience. It should be short, sweet, to the point and your listener should be someone with no vested interest in you (telling your mum won't help). If you can't tell it in less than the time it would take to share an elevator with someone, see if you can refine it down to something simpler. This is called an Elevator pitch. If people don't seem interested, then it's probably not captivating enough, too confusing, or requires your audience to know things that they don't. Again, talk to your audience. Ask what they liked, get them to explain what they understood back to you, this is a good way to make sure your audience is getting the intended message.

If it's not working, take a break, do something else and approach it from a different angle. A Mind Map is a great way of organising your thoughts and working out exactly what your idea should be about.

You haven't done any concepting or looked at any reference, if you're lucky, it looks like you've copied someone else...badly.
Concept art and reference is something that a lot of aspiring 3D artists feel they don't need to do. Planning is essential! You need to gather reference for your idea. Think of real world exampels and keywords similar to your idea and look for pictures. Combine lots of pictures together and pick the ones that are best. It's a good idea to steer clear of things other people have imagined. If you are making a werewolf, look for hairy men, muscular men, wolves, tigers, bears etc. Don't look for other people's impressions of werewolves.

Thumbnail sketches are a great way to explore and combine ideas from your reference without putting too much detail or work into something that may not fit. Do several designs of your characters and pick elements from the best to combine into a final concept.

Once you've got some reference and an idea of what the main elements in your scene will look like, establish the composition and lighting of your scene. Work out what elements are going to go where in the scene, what you want the focus to be, and how you are going to achieve it. This composition article gives you a good starting point for organising the point of interest and how to draw the viewer's eye through your scene.

Don't forget to plan out what your lighting will look like. Think about where your light is coming from, what the source of that light is and what it's doing. Don't leave this until after you finish modeling! Here's a great article outlining a variety of things to consider in relation to artful lighting and colour.

As a general rule, you should probably study how real-world cameras work. Knowledge of things like Focal Length, Depth of field and Aperture will come in just as handy in 3D as in the real world. Maybe more!

Planning is essential! You have to draw and explore your concepts, or you will be left in the dark, and that's an uncomfortable way to work. Draw it. Come to the story that allows you to conceptualise the scene you will then later create.

Modeling

Now here is a touchy subject. I will not go and tell you to model using this method or that method. Instead, we'll look at some common problems and mistakes that arise in your models.

You've either got way too much detail or not enough
This is a common problem that appears in people's work. You either have a rivet with 20,000 polygons that takes up approximately 1 pixel in your scene, or you have an apple taking up 3/4 of the scene in your foreground with only 32 faces. The first will make your render take absolutely forever, and the second will be faceted and look like you have no idea what you're doing.

Only create the level of complexity that is required. You should know exactly how close to the camera each element of your scene will be, and only make it accordingly detailed. Additionally, If you've got a chest full of gold coins and maps which never gets opened, only model one or two things hanging out the side, and nothing in the middle. There's no need to waste polygons on what you're never going to see.  


Here's an example of a badly subdivided model. While the organ pipes are all perfectly smooth despite being far from the camera and many of them hidden behind other things, the keys on the piano are rough and you can see the jagged edges because they don't have enough polgons. Similarly the seats and piano casing itself are made of a combination of straight lines rather than smooth curves. This ruins any render, no matter how good it is otherwise.


Your edges are sharp and fake-looking
No edge in the real world is perfectly sharp. Even a knife has a little bit of roundness and thickness to the edges on it. If nothing in the real world is perfectly sharp, neither should your 3D be.


In this image, you can see that even though it's all corners and straight lines, there is a Bevel on each corner to make sure nothing meets at a perfect 90 degree angle. Chamfering is a similar technique which provides this effect.

Everything's straight and painful to look at
Similarly, nothing in the real world is perfectly straight. Making it so in CG simply makes your work look fake. Be it fences, lines on houses, furniture, floor tiles, whatever, add a bit of warp in there to make them look realistic and never, ever have perfectly straight lines. The image in the previous example is guilty of having too-straight lines.

There are patterns where there shouldn't be.
One of a 3D artist's worst nightmares is Moire. If you have too many similar elements together and are viewing from too far away, the computer will try to sharpen too much and create strange patterns through your image instead. This is a problem you can get in 3D with small things and detailed features in the distance such as mesh clothing, fences, building structures with detailed lines, grills on vehicles, etc. It's unattractive and can be fixed by blurring and reducing detail, and fancy words like Mipmapping and Anisotropic Filtering.

Pinching!
The topology of your mesh is horrendous and it's causing everything to look like a sack of potatoes that's halfway to being mashed. If you do nothing else, check out This video about topology and smoothing. It will help explain why you are getting wierd pinching when you smooth. For even MORE information and no excuse for having it, read This thread which goes into detail about topology flow and edge loops, what they are, and why you need to think about them.

It's hideous
You haven't done any planning, you don't know exactly what you're trying to make, your topology is terrible and it looks like something only a mother could love, and even then she'd have to be half blind and hard of hearing. It happens to all of us, and it can be demoralising and upsetting, but the worst thing you can do is stick your fingers in your ears and yell "LA LA LA" at the top of your lungs. Look at reference, figure out where you went wrong, learn to look at other people's work that is just slightly not right and scrutinise what exactly the issue is. Good observations skills are key to improving your own work. Oh, and start again, don't try to salvage it, or you will end up with something like this: