How Do You Know What Dimensions To Use In Pixel Art?

[HDthunder]

I don't know how big of a canvas I need for pixel art... I usually just work in a 64x64 area and make the best of it and try to do something but I feel limited when I have big ideas and I don't what dimensions to use. how do I get a feel what dimension to use when making a creative piece of pixel art?

[0xDB]

Just make more art. Developing any kind of feeling for anything just requires experience. As a quick fix, just start with a big canvas, start pixelling anywhere and crop it later.

append: Also most tools allow resizing the canvas without scaling the pixels, so you could also just start with whatever dimensions you think will suffice and then later change the canvas if you need more space.

[HDthunder]

Just make more art. Developing any kind of feeling for anything just requires experience. As a quick fix, just start with a big canvas, start pixelling anywhere and crop it later.

append: Also most tools allow resizing the canvas without scaling the pixels, so you could also just start with whatever dimensions you think will suffice and then later change the canvas if you need more space.

Thank you this is actually what I needed.

[Cyangmou]

Look at resolutions from old consoles, which are usually good to get a feeling what pixel artstyles you can reproduce on what canvas size.
Like a specific game's look? Check out the dimensions it was made for.

I listed some resolutions in point 4
http://cyangmou.deviantart.com/art/5-Tips-for-the-Aspiring-Pixel-Artist-457660938

[Ai]

I feel like 'just make more art' is a little too general -- it will improve your feel for how big is small enough to work fast (practical workflow consideration), but not necessarily how small is still big enough (optimization/design problem). What's improved my sense of 'feature granularity' is methodically improving my ability to make accurate measurements (eyeballing things and judging their proportions; measuring the real distance and gradually reducing the amount of error in judgements); secondarily, translating these measurements accurately by eye into differently-scaled areas.

IMO practice of this gross(coarse) ability is what develops into the ability to readily pack and align features into an area, distributing the differently scaled features harmoniously.
Which possibly sounds excessively abstract; it's just intended to express a view that:

* an object is a system of proportion relationships, where each set of dimensions can be classified into one of 3-7 'bins' of 'level of coarseness'
* This is useful because the coarseness of the 'finest' bin is the determining factor for how small you can render the object before details you want must necessarily be omitted.
* You also need more or less leg room -- extra resolution beyond that minimum -- in accordance with exactly how accurately you want to maintain coarser proportional relationships. It's particularly important here to remember that spaces and outlines have visually significant proportions too.

Ultimately it boils down to a huge optimization (ie. design) process. Cyang's tip #5 and suggestion of reverse-engineering their designs are helpful here.

Finally I'd also suggest just taking some time to mess around with combinations of features at different sizes, and think about your results to try to abstract out some principles from them.

(EDIT: clarified to hopefully convey that I'm not talking about rules, I'm talking about actually learning how things work in detail so that you exploit the mechanics intuitively. Breaking down things thoroughly enough and practicing them enough that you internalize the principles and also learn how to break them.)

(EDIT2: list of example principles removed entirely.)

[0xDB]

I feel like 'just make more art' is a little too general

And I feel like a long bullet point list of many specific "rules" or facts is counterproductive to getting into any kind of flow experience. It may be useful for production rules and polishing later down the road but beginners who seek to develop a feeling for things should not bug themselves with memorizing rules too much and instead gain more hands on experience through their own experimentation and observation.

That being said, what's most useful depends on each owns personal learning type. Yours seems to be very verbal and fact based... it feels overly scientific to me even though I myself do have a bit of a fetish for diagrams and facts (struggling to get away from that myself as it often feels like an autism amplifier to me).

[Ai]

I feel like 'just make more art' is a little too general

And I feel like a long bullet point list of many specific "rules" or facts is counterproductive to getting into any kind of flow experience. It may be useful for production rules and polishing later down the road but beginners who seek to develop a feeling for things should not bug themselves with memorizing rules too much and instead gain more hands on experience through their own experimentation and observation.

Uh.. why do you think that I'm promoting rules?
I described a.. pretty general.. process (learn to measure stuff accurately and pack it accurately). Then I used bullet points to try to break down 'what is the view that I'm trying to describe here'. Later I commented on the value of learning how pixel topology works through experience, giving some few examples of principles I learnt. (EDIT: I can understand why you might have interpreted that one as 'memorize this' though. I edited it to clarify.)
I have no bullet points of stuff I'm recommending.. I'm pretty sure. I reread a few times. Feel free to give some specific examples and reasons why you think they promote rules.

The point of my first comment was simply that you are IMO basically saying 'do more artworks' -- and the world of art (even just pixel art) is huge. Being too narrow in your approach can hurt, but so can not having a frame of reference.

That being said, what's most useful depends on each owns personal learning type. Yours seems to be very verbal and fact based.

Not sure about that -- It's plausible (programming can really make you think that way, and I don't have any special insight into my learning process), but I don't communicate how I think AFAICS. Most of my communication style is because I've repeatedly learnt that listeners don't understand generalities as the same thing as the speaker does -- Chinese whispers problem.  So I try to really nail things down, so that replies can be equally specific, and it's harder to assert you understand without actually understanding (or disagree without understanding).
That tends to add : 1. verbosity and 2. logical structure (in the syllogistic A + B therefore C sense)

I can sympathize with worry about getting obsessively lost in stuff. It's been a major concern for me, and actually why my approach has mostly been .. 'do more art'. But, I've come to the conclusion that obsessiveness is really about imbalance in your life, and it's only really fixed by having other serious and varied demands on your attention (ie. not changing the tendencies you have, but simply arranging things so that it's hard for them to take over). I'm not interested in having my learning process be driven by my insecurities.

EDIT: Sorry if this post comes across a bit combative. I feel like my post was pretty comprehensively misunderstood.

[0xDB]

Disclaimer: Do not read this post with an angry voice or while upset. Assume a monotone and read the whole post first and don't start fragmenting and responding to every fragment out of context. I'm suggesting the latter because you might just end up unnecessarily writing yourself into a rage or something.

You assume that I'm thinking that you're promoting hard rules. I am not.

Ai, I feel swamped. The obsession I feel in all of that is one to explain yourself (presumably you might be thinking that I said "you're wrong" when I did not?) and I'm getting some hostile vibe(or combative as you call it yourself) from it and frankly that does annoy me somewhat (feeling misunderstood simply irks me a lot... just as it presumably irks you?).

I'd just walk away but since you went through all the trouble of explaining yourself I feel you deserve an explaination from me too, so we may understand each other better.

I did not just say something generic as "just make more art". It was merely the first sentence and then I proceeded to explain the reasoning and motivation behind it, qualifying those words further.

Let me qualify them even more, sharing my thought process that lead to what I wrote first and what I wrote in response to your list of "rules" (note how I had already put rules in quotes in my first response to signal it is not meant to be taken literally but in a more relaxed interpretation of the word).

So I absolutely did not call you something like a "rule obsessed pixel nazi" which seems to be what you had in mind judging by the overly defensive stance you assumed (which I'm reflecting in this post, defending my view even if I could just walk away and let it go but then we'd just misunderstand us again in the future).

But now the thought process:
The original question started with "how do I get a feel...". Emphasis on FEEL.

So the question was about getting a feeling for something and not asking for technicalities or specifics. That's why I think a bullet point list of specific "rules" or facts (or someone elses (seemingly random) observations) is not the most useful answer to the broad/generic question of how to get a feeling for something.

Like I said though, there is a use for rules, I just don't think it's useful right from the start because it takes away the possibility for someone to develop their own feelings/observations about a thing. It is, imo, more likely to cause someone to slavishly follow the list of rules, feverishly trying to adhere/obsess about all of them and giving away the freedom of cheerful play and experimentation which is important to keep up the motivation in the beginning.

The problem I see with swamping beginners with "rules" and too many specifics(e.g. counting pixels to make round things) is that if they take them as such(as real actual rules with potential punishment involved upon breaking them or something equally ridiculous), they'll just be programmed robots applying items from a list of rules but do not develop a "feel for something".

Again, I'm not saying "you are wrong" or "your views/approach is wrong" and I really do not think of this as an argument/combat but rather as an exchange of our thought processes, so we can understand why we write the things we write.

In the end it's not important and nothing to worry/obsess about. It's just pixels. Please do not feel misunderstood.

(Also... the original question did not appear to be about "feature fidelity/detail granularity/sampling resolution" to me. I refused to interprete more into it than what it literally said. It seemed about canvas size and developing a feeling for how big to chose it from the start before really knowing what to pixel.)

So well. Communication: It difficult. *grunt*

[yrizoud]

It is implied that such guidelines are rules of thumb.
For example, for game sprites of humanoid characters, the figures I keep in mind are :
32 pixels : Stylized and SD characters only.
64 pixels : Anatomically correct character. Arms and legs can be detailed, but you can forget about showing individual fingers. If the character is not SD/anime/cartoony, you can't show the eyes (whites) yet.
96 pixels and above : Fighter sprite (ex: Street fighter). Subtle animations (breathing) and facial expressions become possible.

[Ai]

Disclaimer: Do not read this post with an angry voice or while upset. Assume a monotone and read the whole post first and don't start fragmenting and responding to every fragment out of context. I'm suggesting the latter because you might just end up unnecessarily writing yourself into a rage or something.

It's ok.  (at least, I think? I don't even always know if I'm angry.). I don't expect you to know what I'm touchy about [FYI: the tiniest implication that I'm trying to make rules rather than figure out how to break them]. I get incoherent sometimes, but I really am trying to communicate.

It's a good cue, though, thanks.

You assume that I'm thinking that you're promoting hard rules. I am not.

OK.

Ai, I feel swamped. The obsession I feel in all of that is one to explain yourself (presumably you might be thinking that I said "you're wrong" when I did not?) and I'm getting some hostile vibe(or combative as you call it yourself) from it and frankly that does annoy me somewhat (feeling misunderstood simply irks me a lot... just as it presumably irks you?).

Sorry, I think my tendency to edit my message multiple times can give that kind of effect. There are at least a few overlapping / redundant thoughts in some places. And I don't really have that structured a writing process. Definitely room for improvement there.

So the question was about getting a feeling for something and not asking for technicalities or specifics.

Hm. I'd argue that getting a feeling for something is 50% about developing suitable habits and 50% about exercising them in the widest possible set of circumstances.

That's why I think a bullet point list of specific "rules" or facts (or someone elses (seemingly random) observations) is not the most useful answer to the broad/generic question of how to get a feeling for something.

That part of my post wasn't supposed to be an answer, let alone a particularly useful answer.

The problem I see with swamping beginners with "rules" and too many specifics(e.g. counting pixels to make round things) is that if they take them as such(as real actual rules with potential punishment involved upon breaking them or something equally ridiculous), they'll just be programmed robots applying items from a list of rules but do not develop a "feel for something".

.. But I wasn't advocating even -using- those "rules" in any way. I was advocating finding your own principles. Are you saying I presented it in a way that didn't achieve that?

(that's not intended sarcastically; I really do want to know. I just edited out the entire list, but I'm not sure that the update is really an improvement.)

Like I said though, there is a use for rules, I just don't think it's useful right from the start because it takes away the possibility for someone to develop their own feelings/observations about a thing.

I think we agree then. You are possibly far more tolerant of rules than me. 

It is, imo, more likely to cause someone to slavishly follow the list of rules, feverishly trying to adhere/obsess about all of them and giving away the freedom of cheerful play and experimentation which is important to keep up the motivation in the beginning.

Sure, I know what you mean here.  I also wish someone had given me some direction, something to narrow down the problem I was working on, in the beginning; I didn't solve motivation properly until I accepted the idea that meaning is essential for motivation and doing series of X, then series of Y, then series of Z is much more motivating than mixing them up X Y Z X Y Z or something completely arbitrary...

(Also... the original question did not appear to be about "feature fidelity/detail granularity/sampling resolution" to me. I refused to interprete more into it than what it literally said. It seemed about canvas size and developing a feeling for how big to chose it from the start before really knowing what to pixel.)

I agree with this. I intentionally chose the more involved interpretation ('can I put X legibly in Y dimensions') because I regarded the simple interpretation as more or less solved.

(also, admittedly, because the multiscale thing is a bit of a focus for me right now -- I think it achieves the most impression-of-quality per amount of effort put in.)

Communication: It difficult.

Well, thanks for taking the time to explain your position, and I apologize again for any sloppy communication on my part. Hopefully this is a bit easier to understand and less annoying. I did shuffle around your message a little in an attempt to reply more coherently.