pouët.net

Dear pouet,

I am an engineer, and I don't know a thing about how to make stuff look and sound good. I want to change that.

Unfortunately, the usual route via >9000 hours of practice is unavailable. Demoscene and all this art stuff is just a hobby, and it is rather far from my full-time profession (all user-experiencable parts are made N layers of abstractions above me), so I cannot feed from that too.
There are probably around 50-100 free hours left available for demoscene activity until the end of the year, and I'm trying to find the best use for them.

To make things worse, I also don't know what things I don't know, and how these areas of expertise are even called.

Therefore, I have a few very dumb qustions:
1. Are there any introductions on how to make visuals appealing? How to avoid coder colors, ...
2. How to build scenes that induce emotional response?
3. What is the color of grass? Water? Sunset? I sure can derive them by mapping spectral absorbtion lines via CIE tables to RGB, but this looks like overkill.
4. How to compose and master music? How to avoid cofining oneself to random pentatonic notes fed through 3N-line delay and reverb?

Are there any introductory materials? Lectures? Maybe even online courses?

PLEASE HALP

Heyho,

same here...

for music... I dunnoo...this takes time... but you can learn some basics just by hanging out at a demoparty (I have the feeling, that EVERYONE can do at least a little bit music there...)

For the rest...
What I did is:
1 video of https://www.youtube.com/user/FZDSCHOOL every week.
This provides you with good basics... start with video 1 waaaay in the past...

And the other thing I really liked is the book "the skillfull huntsman". Which is like those Game Art books (e.g. The Art of the Evil Within) but without the game background.
The author describes how he would do a grafic novel to the fairy tale, how everything would look and why. This gives pretty good insights on the thinking you have to do, when composing...

good look and at first.. have fun!

I decided to leave coding and learn that other stuff, because it's a lot of fun for me...
hopefully you find what you like best... Coding is the coolest stuff. But for me it wasn't enough fun....
Aaaaand keep in mind: Demogroup means - you don't have to know everything ;)

Do the demoscene thing, and cheat as much as possible. Problems with colors? go grayscale. Problems with audio? get audio from someone who knows how.

There's no easy way to learn everything. It's a life-long thing. The most important thing is to make it fun for yourself. Also, making stuff, even if it's crap, is always better than not doing stuff.

Steal and hack! Ha! :D

Well, there sure would be the possibility to read some stuff on how "colors work" (regarding your "what color is grass?" question) and make things appealing. There are some basic guidelines and if you try to follow them, your stuff will most probably already look much cooler. Also you can use things like www.color.adobe.com to get a basic palette which you can use. That's at least better than the usual coder colors imho.

Other than that... if you don't have the feeling for visuals, it might be a bit hard. Best thing to do is, to check out stuff which you find "visual appealing" and just try to "copy" it (in the sense of color schemes, composition, etc.)

While this article mainly is for procedural graphics it might be interesting to take a look at.

Pixtur/Still held two seminars at Revision 2013 and 2014 respectively which should be perfect for you:
Concept, camera, composition and color
Rules of thumb for (slightly) better design

When it comes to music there's just "do more music". Ideas have to come from you, but when it comes to arrangement and the technical aspect of mixing audio, I'd recommend these two books:
Mixing engineers handbook by Bobby Owsinski
Mixing Audio (2nd edition) by Roey Izhaki
The former is better for beginners, the second is a incredibly well written book that systematically goes through the process of mixing audio, with comprehensive explanations on the various signal processors one uses for that (eq, compressors, delay, reverb etc.).

A good way to identify your areas of improvement for music production is to use reference tracks. Find a track that you like (in the same genre you are working in), and compare your track to that. You should be able to pick up quite fast where your track is lacking, and then you can work on improving issues as you go along.

And there's no shortage of musicians in the demoscene, so you might just opt to find someone to work with instead :)

See how others do it and don't expect to get everything right the first time.

Find someone who is already good at music or design and will give you honest feedback. Show them your work and listen to them when they say "that is shit". Getting criticism from a respected expert will help accelerate you in learning what is good and what is not.

It's actually hard finding someone who is patient enough to help you and is not a close friend who tells you everything you do is great. A good starting place might be other demosceners, Pixtur had a thread here where he offered to do paintovers. I am sure some kind musician here would also give you crits.

Like other people mentioned already, reference the work of people who excel at what you are trying to do. Watch closely your favourite demos (or any visual art medium) and see what is the common things that make something look good. Play your music back to back against your favourite musician. Hear what the difference is in composition and production/mixing.

All these things will still take time to learn but trust me, working in isolation by yourself will make it take much longer. Utilise the experience of others to accelerate your learning.

There's one crucial thing few of us pay any attention to that you didn't mention. The thing that most sceners sorely lack is an eye for composition (not the musical kind... or actually that one too). I find good colors to be relatively easy, trivial even, compared to even a decent composition. Especially in three dimensions most productions seem to have severe difficulties in it (and I don't claim to be an exception). Very little or even no attention is paid to the placement of elements within a scene, little attention to camera angles or focal lengths and perspectives.

An easy way out of course is just to use a center composition (a big faux pas in some fields), which has its uses and can sometimes be effective despite it's overuse. In the 2D and "oldskool" realm things in that regard are a bit simpler, since scenes are more likely be static, so you can carefully lay your points of interest to aesthetical positions. In modern 3D though there are very few good examples as far as demos go. It's much harder when you add another dimension to the mix and most likely start moving around both your camera and the elements. Honestly one of the very few that comes to mind are destop's works and to lesser extent pixtur's too (make sure to watch those seminars lug00ber mentioned). I also have to add Conspiracy's most recent work. Notice though how most of these tends to use relatively static scenes and shots (remember Pixtur's emphasis on slow camera movement).

How to overcome these issues and get better at composition? I recommend three things. First of all, get more familiar with painting and traditional arts, but in my opinion most importantly photography. Study some examples, start building a collection of works you find appealing and aesthetical, organize them, get inspired, don't just follow blogs and feeds that push miscellaneous design at our face, do your own research. I actually really recommend you to do this anyway; having a proper idea of what you actually find aesthetically pleasing helps you apply your opinions in practice. Also Read about it. No need to pretend to be an art school student, but get familiar with composition rules and guidelines. Even light articles on the subject can provide great insights. Additionally, I really recommend to actually apply your observations in practice by taking up either painting or photography. I'd say even this is an easy way to prototype your productions as you can later apply your discoveries in demomaking as well.

Choose your colors by looking at them rather than putting in arbitrary values. Be sure to have a setup where you can adjust the colors smoothly while your scene is being displayed. Use a monitor with decent color reproduction and minimal directional color distortion.

It is not that you can't generate your colors using a formula. You can do that, but you should verify that the result looks good, and experiment with the parameters of your formula (again: while your scene is being displayed) to find good values.


There is no such thing as specific RGB values for specific things, as it depends on lighting, exposure, white balance, surroundings etc. Also, it is not about which colors you use, but how you use them. The same RGB values can be horrible coder colors in one context but beautiful in another.

Always remember that the human eye very much likes the "golden ratio" φ.
Thus, it's a good idea to layout an effect screen in the Golden Rectangle.

Then there are several "rule of thumbs" on my regard:
- Water rarely appears as blue but rather greenish, grea-brown or white-clear
- Sky is *not* an uni-babyblue tapestry
- Treetrunks are rarely brown but rather grey-greenish
- Cobaltblue and Sulfuryellow together are *eye cancer*, turquoise and orange however very much work in harmony

Generally: in order to depict nature you'll have to experience and contemplate it ;)
Listen to your tummy more often instead of trying to overengineer it with your brain.

i think what is hard about this coming from a coder's point of view as i did (a relatively clearly specified world where results are measurable) is that it's not a case of right or wrong. its often not even clear if an improvement has been made when you change something. this requires a shift in mindset.

the first part to this process is to develop the eye: the opinion that something is good, or not good, or could be better. i dont think there's really a shortcut to this, it just takes time. try and look at as much material as possible and always ask questions of it: do you like it and why; do you like it more than another image of a similar subject. then find something you like and copy it. the tutorials, the "rules for design", they come in to play when you know "something is wrong" but you don't know why.

from the technical side i would say the major thing is to make it as easy as possible in your production environment to change things and experiment. the easiest way to find good colours is if you can change all of the colours you have in your scene easily in realtime with a colour picker. the easiest way to get a good composition is if you can move the camera and all the objects live.

it sounds obvious but if everything is bound to the implementation, if you have to quit/change/recompile/reload to see changes and you cant easily iterate and compare, the process is much harder. again, its not a case of "right or wrong" - just "better or worse" repeated many times over for many small changes.

Photos tend to contain colors that match in brightness and saturation, so instead of trying to handcode everything downscale a good-looking photo and pick the tones from there.

I'd rather decide what scenes to make or colors to use (even if they will be coder colors) rather than use some guidelines.

Or you could semi-randomly generate palettes and ask your friends, or like genetic algorithms of "beauty".

Idea: A demo that improves itself, genetic algorithms with preselection from users.
Idea 2: AI that reads Pouet comments, about colors sucks, that guitar riff sucks, and adapts demo.

Optimus, that's just asinine. Beauty is something that people roughly can agree on. Yes there's differences due to culture and upbringing and what have you. But known good guidelines get results that lead to a wanted direction and it's very like it's leads to direction you want to achieve too. I strongly wonder if either of your ideas are feasible in the near future.

I'm following this thread very keenly.

Hmm.. an interesting topic for coders.

I tend to use ugly coder colors, unfortunately, so I am more like thinking out loud here, to also get some feedback.

I think coders tend to underestimate the emotional meaning of colors and palettes.
Usually coders pick the most remote colors to see different parts of the image better like in syntax highlighting or when you debug your algorithm visually - and in their use case it makes perfect sense.
But obviously, colors/palettes do have meaning - cheerful (saturated bright colors) vs toned down, high contrast to show tension etc... There are a lot of resource to read about it.

So, my recent recipe is to start with grayscale or even clean b/w sketch of what I try to do and keep it in this stage as long as possible to only add texture/colors later when I fully understand what kind of emotional response I want to stimulate (with easy to use UI). And I guess, one should be very careful here, as even small details might change this emotional response in the sensitive/non-technical audience.
Therefore, as a rule of thumb, better to add less if you are not sure, then more.

Step 1: be talented.

A few very basic "tricks" that I use:

• Use only a few (one or two) dominant colors per scene (or even per demo, for a more consistent look).
  
• Generally stay away from RGB primary colors. Prefer colors that have some contribution in all channels, as they tend to behave more like physical materials.
  
• Do the overall design of the demo over a very short amount of time. This helps making the design consistent.
  
• Use only a few different fonts. Again, this improves consistency.
  
• Either do dead-centered design, or stay far away from it. Close to dead-center designs are just disturbing to look at.
  
• Try to avoid image features that almost align, but not quite (this is a obviously a variation of the point above). Push features either clearly together, clearly apart or exactly aligned.
  
• Avoid modelling human features as long as you can't get it *really* correct. Our human eyes are masterminds at spotting errors in human features.
  
• No fucking Neuropol. Ever.
  

This list could probably be cleaned up and generalized, but I'll leave that task to someone else ;)

Step 2: Don't use brown, yellow and green together, unless you're showing poop.

I'm very much on the n00b side of this, but on the coder-colours front: I believe it helps to get into the habit of working in HSL rather than RGB. For example, if you've got a colour that you want to make lighter, don't just guess and bump up the RGB hex values in your code by an amount that sounds about right... open up the colour picker in your favourite image editor and adjust the saturation/luminance sliders appropriately.

If you want (for example) grass colours, then grab them directly from a photo - that's what many digital painters do as well. But of course grass isn't just "one colour" .. depending on the lighting conditions, ambient colour, colour bleeding, shadows, light direction, and so on..

Appropriate colours and lighting make up a huge part of a picture's "believability", so getting them from real-world sources can make your life a bit easier

Mhh nope.
I learned in the last year, that everybody can do this! Ragnarok showed me some books and made be believe this... being talented just means to make some things without knowing it... but you can learn everything...

never think something like "nah, I got no talent, I quit it"!

First of all, work on getting more time available for this hobby, you'll need that little time already for doing a thread on pouet. Also you run into danger to end up like all those lazy bums in scene that don't get their assets and projects finished within years. Also stick to a single discipline like code, music or gfx first. However it is never wrong to have skills in as many disciplines as possible.

I can agree with Gargaj, but only when it comes to top-notch designs, and we are talking basic "therapy" here (I believe).