Pixel art and Sharpness

[Helm]

I want to discuss something. Is Pixel Art inherently tied to very very sharp art? Can you make pixel art that is blurry or very soft or does that defeat the purpose of the tool you're using? Where does AA stop being useful? Perhaps we're underusing AA? If you pixel something that is for all intents and purposes identical to vector art (only slightly sharper, of course) is that self-defeating? Does Ai (the pixelation user, not Artificial Intelligence) with his shade routines augment or disorient the focus of pixel art?

If pixel art is all about sharpness, then why do we have that 'sweet spot' of AA ? Why don't we just pixel with sharp colors all the time and embrace jagginess? And how about demoscene-like art that is trying so much to hide its individual pixels and edges?

I am not advocating any position here, I'd love to hear your thoughts.

[SwapBrain]

I recently turned a corner in my 'traditional' media where I am ok with letting people see my 'hand' in the media. For a long time my goal was to obscure the medium to create a clearer and higher fidelity image, but I am moving away from that, mostly thanks to a more solid footing in 20th century art history.

All this is to say that one thing that I love about pixel art is the ways in which the images it creates are ambiguous - one of my favorite moments is when I zoom in to get a better look at a particular detail, only to realize it does not exist. I think that to focus on this or that technique is like arguing about the best kung-fu, and we know how Bruce felt about that.

It is true however that I am a hobbyist when it comes to this, and not a professional, or a purist. When it comes to jagginess, my only response is that if it is done in such a way that I as a viewer can recognize it unambiguously as a choice and not a mistake, then I accept it at face value (see Corporate Climber, for a recent example). I feel like intentionality is core to any technical discussion - how many times have you tried to critique someone only to have them blow it off with "I meant to do that." - in the words of Robert Anton Wilson: "Reality is what you can get away with. If you can't get away with it, it ain't real."

Sorry if this is too rambly. I would be glad to clarify or discuss further.

[Helm]

I think I understand what you wrote and thank you for your input.

Intentionality in itself is dubious. An artist makes a lot of subconscious choices in his art and rationalizes later. What I'm looking for in art is beauty (and arete) and the road to these things is through awe. The initial step to awe is seeing effort. What you call intentionality on the choices of the artwork for me is only one part of the effort. A *lot* of intentional (or not) choices shows more effort than the result of those choices does. Good choices or not is not the matter as much as feeling the artist laboured on his piece of art. When I feel a lot of time and sweat has gone in a piece of art, subconsciously I become receptive to the beauty that might be inside (whether I get the message or not is then up to thematics and aesthetics and the ubiquitous 'x factor'). I open up to foreign senses of beauty then, only when touched subconsciously, primarily, by the results of hard work by some other artist.

If it's easy, I'm not interested in it. I don't say this to appear bad-ass or whatever... I'm just describing a phenomenon as it occurs, it's not a value judgment.

Seeing the 'artist's hand' in the art, as you say is usually indicative of *less work* spent on the artwork (the traditional post-Raphaelite level of rendering finish is missing). 20th century art in this context looks perpetually unfinished. Celebrated works of art from that period are just 'rough drafts' in that mindset. Scaling back my expectations as a product of the modern world I consider any piece of art on the level of say, a good Impressionist, a piece of art capable of theoretically achieving some awe in me, and letting me recognise its beauty. In our modest pixel art terms, if a piece of artwork screams 'I did this in 30 minutes' it's not probably worth looking into too much. If it's more like 3-4 to 6-10 hours, then yes, I'm looking at it more closely.

This is all relevant to the size of the canvas and the complexity of the piece of course, but it's a good ballpark figure.


There's autistic pieces of art that belie hundreds of hours of work that has gone into them and yet the result fails to excite in a primary way (it might excite in the way of 'why on earth would anyone do this?' but let's not discuss this now). So effort paid obviously is not the only factor in achieving some awe.

And yes, if something looks very worked yet still retains the artist's 'hand' this is an even more difficult yet capacious for awe situation. This is what I try to do. I never try to draw exactly in realistic terms, I want my art to show my hand and through that my soul. I am not afraid of narrowing my audience in this way. I just need to learn to be even more patient (more effort paid to the art) and better technically (more eyecandy on display to lead to the inner beauty of what I'm doing)

These concerns are sideways from the point of the thread but still very welcome to discuss. Continue from here as you see fit.

[Ai]

I want to discuss something. Is Pixel Art inherently tied to very very sharp art? Can you make pixel art that is blurry or very soft or does that defeat the purpose of the tool you're using? Where does AA stop being useful? Perhaps we're underusing AA? If you pixel something that is for all intents and purposes identical to vector art (only slightly sharper, of course) is that self-defeating?

eg. Panda

Does Ai (the pixelation user, not Artificial Intelligence) with his shade routines augment or disorient the focus of pixel art?

I have shade routines now?

Sounds like I am an artificial intelligence.








I have done ridiculous-level manual aa and buffering (eg on bub-n-bros graphics, also on some original gfx). The result was like.. demoscene++ in terms of style (without ever setting out to be that -- just to be extremely smooth.). I didn't actually regard that as a win, cause it seemed very hard to actually add to communicative value that way.. it did LOOK BETTER.. just not enough to warrant the time.
If I did it again, I'd be aiming to get a high quality, gamma correct rendering of a layered CG and touch it up to add pixel value back; less time, more macro-level accuracy.

Personally I feel that if I've spent more than 3h on an artistic work, it's wasted time.. at that point, it tends to become a game of trying to eliminate an error whose nature you do not comprehend.
(FWIW I'm clearly not an introvert, unlike (I posit) most artists -- I care more about doing than understanding.)

[Cure]

If the vector process can yield an equally satisfactorily result then it seems silly that someone would use pixel instead. Not necessarily wrong, just a bit masochistic. Panda is a poor example here, I think, as you can zoom in on his work and see how tedious and intentional it all is. It has a beautiful balance that an automatic tool would be hard-pressed to achieve. It's very zen, everything in its right place.

Demoscene perhaps doesn't take advantage of the tendency of pixelart to produce a natural shaprness, and loses a little bit of the zen aspect when overworked.

I would say we need to define this sharpness we're talking about. Pixel art is abut being 'crisp', perhaps (just to use another term to allow us to speak about sharpness). A non-AA'd line is at maximum sharpness, perhaps, but it is not as fine as it can possible be, we see squares and right angles and not the "crisp-ness" that a good pixel artist can achieve with the same line and a few shades of AA.

I'd say pixel art has a special relationship to 'shapness' that other mediums perhaps do not.

Like painting, some amount of blur is good, you probably need some blending in places, but pixelart can suffer from the same issue that painting is susceptible- the work can get too blurry, too smooth.
Here I use an example of my own art, as it is what immediately comes to mind:
[link]
This painting suffers from its lack of sharpness. While in painting it would've taken me a certain amount of time and effort to tighten up the work, in pixel art the sharpness is practically inherent, the pixels won't mix or behave translucently unless I make a very conscious decision to cause them to do so. And even if I'm playing fast and loose in pixelart, it's much simpler to zoom in to the atomic level of the piece and make the image sharper than is the case in painting. Why pull out the single-hair brush and hold my breath when I've got the magnifying tool!
- - -

apologies If I've veered off topic or rambled, it's pretty late here. this is a very interesting discussion though, I look forward to seeing how it develops.

[Dusty]

I want to discuss something. Is Pixel Art inherently tied to very very sharp art? Can you make pixel art that is blurry or very soft or does that defeat the purpose of the tool you're using? Where does AA stop being useful? Perhaps we're underusing AA? If you pixel something that is for all intents and purposes identical to vector art (only slightly sharper, of course) is that self-defeating?

eg. Panda

I gotta disagree. I find Panda's stuff an amazing feat of pixel technique. As Cure said, I don't think what he does could be duplicated by automatic tools. They're smooth, but so well-defined. It almost crosses the line, but at the same time I consider that more of pushing the envelope of what pixel art can be, rather than what it can be confused with.

On the other hand, I consider work like this(no offense to the artist, of course) to be better suited for vector art. http://www.pixeljoint.com/pixelart/57404.htm
Where is this benefiting from being pixel art at all? There is no AA, there is no attention spent on clusters or overall how one pixel will effect the next. In the end I think this could have been done in vector to save the artist some time, and the art would have benefited from it.


I think we attempt to achieve this "sweet spot" of pixel art because we approach the pieces depending on how the pixels interact to create an overall image. A black line of pixels across a light background is something we probably notice, and have the technique to smooth it out to make the pixels interact with each other better. Contrary to other arts where such a detail wouldn't get the same attention because the art is approached via sections or steps. Not to mention we as humans can approach these smaller pieces with a much more aware degree of accuracy that tools just can't do. We could possibly create a vector similar to the images Panda creates, but can we create the same outcome on something like this:

I doubt it. At that resolution the effects of vectors and tools would create a blur and details would be lost, details that we as pixel artists can make out and create. Just a couple of dark pixels under the feathers that creates such an effect of depth that you'd probably have trouble replicating with a tool. And if you did, you'd end up zooming in and getting just as pin-point accurate as you would just doing it the pixel way.

[9_6]

I'll let the "pixel purists" argue about the way all they want but in the end, it's the result that matters and to me, it doesn't matter if you got there with bitmaps, vectors, pre-rendered models or by carving it into stone.

As for the focus of pixel art, isn't it to have as much control over the picture as possible and not to achieve a certain "look"?

[Ai]

Does Ai (the pixelation user, not Artificial Intelligence) with his shade routines augment or disorient the focus of pixel art?

I have shade routines now?

I wonder if this was a reference to my intermediate-color-picker, my granular color selection work, or my auto-AA work?

EDIT2: Aren't you lucky! I just made another one.
color visualization for painting
(4-cornered color blender using L*ch color space, 5 colors/side (25 colors total), support for quantized output (CPC,Genesis, Amiga..))
It's my favorite <3

Re:process vs result
Process (bitmap, vector, pre-rendered model) has some implications about what you understand about the result (due to working closely with that aspect of the art). This can and does effect the kind of results you get, particularly on the larger scale (eg coherent art within a larger work)

[Ryumaru]

Who's to say we must treat every pixeled object the same anyway? People may seem to think that if you don't pixel everything in the same fashion, with uniform amounts of AA overall- style consistency may be lost. But trees should be pixeled differently than rocks, and grass, and humans. I love having sharp edges indicative of individual leaves on trees, and one pixel thick crevices in rocky surfaces, but I wouldn't dare do anything of the sort on say, a female face.

[Conzeit]

I have to agree with Ryumaru here, I feel we've fallen into a trap of framing everything in this question of what is or isnt pixelart, and that limits the amount of stuff we can do with it, or even worse what we can learn from it to use in other mediums. There's just stuff that pleases you and stuff that doesnt...who cares what is pixelart?

Hand AAing a curve to better express it is incredibly boring to me, I only see interest in AA when you're dealing with trying to evoke some bizarre optic illusion with something like 10 pixels, just so you get the right expression on your little sprite. Somehow pixelart has always been about getting these dots of light to interact in some bizarre way I cant understand to create something that isnt there. The blurs of old CRT tvs are really mysterious and enthralling to me because of that.

I do think it would be a lot more fun to see some people use some jaggies at times....which is why the aproach of "should pixelart be x?" seems a little boring, because once you answer that it's like setting up a frame of thought that all pieces made from then on HAVE to have this to be good...like it happened with Selout you know?

Dusty:I was going to say i didnt see anything undoable in that hawk until I noticed the eye. I guess there would be some subtlety lost there.

[Rydin]

Pixel art is not all about sharpness.  It's more about communication in spite of restrictions.  The meanest restriction against AA is Time.  Have enough time for pixel-perfect polish?  Then do as much as you can.  This adds to quality, and quality is good.  If important elements are in place first, however, the piece will still be readable when the deadline arrives.  Crispy AA or not.

As for the focus of pixel art, isn't it to have as much control over the picture as possible and not to achieve a certain "look"?

Is blurry AA is a desirable end result or certain "look"?

[Helm]

Personally I feel that if I've spent more than 3h on an artistic work, it's wasted time.. at that point, it tends to become a game of trying to eliminate an error whose nature you do not comprehend.
(FWIW I'm clearly not an introvert, unlike (I posit) most artists -- I care more about doing than understanding.)

Both these statements are very interesting, Ai. I'd like it if you explained more about how it chasing an error you don't comprehend doesn't lead to some sort of comprehension of *something* (perhaps not how to solve the error itself directly) that's worthwhile. 3 hours per art piece - depending on how fast one is as an artist - might be too low to achieve forward movement and to get better from piece to piece in my opinion.

Cure, I do agree that Panda is 'better' than a vector robot because he chooses to place sharp pixels here and there and his routines are just tighter than the computers. But what if he worked with vectors and then spent 30 minutes putting in per-pixel detail here and there. Would the end result by for all intents and purposes, the same?

I'd like Panda to answer, actually, since we have him here. I'm really interested in the answers to these questions about what pixel art means to everyone.

Cure, that 'crispness' you're talking about, as opposed to just sharpness (which can be found in a jaggy line), is it about clusters? That is, if you can see the artist optimized his clusters but didn't do any AA of note, then you can tell they're beautiful/set up intentionally to be beautiful and this is 'pixel art' as much as a demoscene full-on AA-fest is pixel art?

I agree that unlike say, oil-painting, sharpness is inherent and promoted in pixel art. Which is why a lot of pixel artists have trouble transfusing their skills in other mediums, unless those mediums have really really opaque brushes.

I think we attempt to achieve this "sweet spot" of pixel art because we approach the pieces depending on how the pixels interact to create an overall image. A black line of pixels across a light background is something we probably notice, and have the technique to smooth it out to make the pixels interact with each other better. Contrary to other arts where such a detail wouldn't get the same attention because the art is approached via sections or steps. Not to mention we as humans can approach these smaller pieces with a much more aware degree of accuracy that tools just can't do.

That's very interesting, Dusty. It does seem to go back to 'pixel art = means of fine as hell control' generally.

Ryumaru, your question is fair, but it seems to me that pixel artists *do* tend to pixel everything with the same amount of finish, usually. It's like an inward achievemet, "Render Get!". I very rarely see anyone intentionally draw an object in the background blurry and an object in the foreground incredibly sharp, say in a natural vista. It's usually sharp sharp sharp everywhere, or AA AA AA everywhere. It seems the micro-practises we adopt in pixel art define the macro-effect of the composition. This is something difficult to shake off. When I made that piece, Kitty Empire, with the very rendered dude about to stab a dog and the abstracted background, people in the thread were disappointed I didn't draw the bakground with the same amount of detail that I did the guy in the foreground. The natural end-result of that line of thinking is the beautiful work Fool is doing. Where everything is rendered sharp and detailed on all levels.

I have to agree with Ryumaru here, I feel we've fallen into a trap of framing everything in this question of what is or isnt pixelart, and that limits the amount of stuff we can do with it, or even worse what we can learn from it to use in other mediums. There's just stuff that pleases you and stuff that doesnt...who cares what is pixelart?

Conceit, I am not interested in agreeing or typifying what pixel art is, I am interested in what pixel art means to different people.

Rydin, I do not think necessarily the greatest aspect of pixel art is that it's communicating in spite of restrictions. Some people use as many colors as they need and as big a canvas as they need and the end result is still pixel art and it's still sometimes beautiful. I think the restriction hard-on is more for the artists that can spot it and care.

[Panda]

I don't think I have much to add; but over the years, I've realized that what I really like about pixel art is not the look, but its process.
The way you build up an image, having full control of everything. Sort of like a puzzle or building something with LEGO.
You have an amount of pixels you can use, and by arranging them in different ways, it just works.
Once you figure out how to start, everything else just clicks together.

Of course it could be replicated with vectors and some retouching (saving some time), but that simply doesn't interest me.
If I wanted to work with vectors, I'd be doing so.

Also I guess some sort of obsessive compulsive traits affect my choices as well.
I like order, and I like doing things right.
I just need a level of "polished-ness" to be happy.

So all in all, I just do what I like to do for my own mind-pleasure, following my own criteria.

[Cure]

@Helm: I’d say it could be about clusters or it could be about hiding the resolution (in ways other than well-organized clusters, like aa). If the clusters are managed by the artist then you’re probably less likely to end up with an aa-fest, as you often see in demoscene. But by crispness I also meant well-managed aa, like Panda’s, in the respect that it hides jaggies/the resolution. It works along with the clusters, whereas I think aa-fest work usually isn’t quite as mindful of the clusters, or at least doesn‘t treat aa as something subordinate to clusters.

[Ai]

Personally I feel that if I've spent more than 3h on an artistic work, it's wasted time.. at that point, it tends to become a game of trying to eliminate an error whose nature you do not comprehend.
(FWIW I'm clearly not an introvert, unlike (I posit) most artists -- I care more about doing than understanding.)

Both these statements are very interesting, Ai. I'd like it if you explained more about how it chasing an error you don't comprehend doesn't lead to some sort of comprehension of *something* (perhaps not how to solve the error itself directly) that's worthwhile. 3 hours per art piece - depending on how fast one is as an artist - might be too low to achieve forward movement and to get better from piece to piece in my opinion.

re: comprehension:
I often find the principle behind 'you learn to ski in the summer' works for me:
 It seems to work much better to work for a time on something, then intentionally spend time away from it, drawing other stuff. The downside? disconnection from the original work -- maybe not in spirit, but it may well look like crap later, not worth the time to fix. The upside -- there seems to be a kind of synergistic effect: as long as you're drawing, you can be integrating the knowledge you are acquiring, subconsciously.. until you reach a saturation point for a given area, which gets in the way of noticing anything much consciously any more, like your consciousness gets tired of that for a while...

If you understand what I mean when I talk about the difference between epistemological (understanding)
rationality and instrumental (achieving) rationality, I'm basically just saying, working on instrumental rationality is both simpler and more rewarding in terms of boosting both: I have learned to trust my subconscious processes to notice things I don't have the processing power to notice consciously. Then I just need to identify the differences in my behaviour in order to boost to my conscious understanding.
(the four stages of competence might be relevant here)

So as to understanding -- I believe that the gains you can make in conscious understanding are much less than the conscious understanding you can gain through first doing in great variety. eg. 100 5-minute drawings of cats are much more valuable in terms of knowledge acquired per unit of time, than 500 minutes spent on one or two drawings of cats.. basically because most of your computational power is subconscious. (I'm aware incidentally that most people reading this will attribute more of their understanding and actions to conscious motivations and thoughts than can remotely be justified... this is a well documented cognitive bug.)

The 3h thing: in retrospect, I would say it depends on a) experience level -- how many mistakes you've already done, and b) cognitive skill -- your ability to notice elements that are discordant with reality.

Currently I feel that I will never really have 'enough' artistic vocabulary, if you know what I mean by that..
so I'm focusing on getting the feel of various kinds of things; don't really want to hang around too long on one subject . I need more experience. 
Conversely my cognitive skills are quite good. (I found things like LessWrong's explanation of normal and quantum physics had a great impact on my ability to see things more like what they are. various other stuff helped me to train and trust more in my subconscious ability to notice things, rather than constantly trying to notice errors consciously). I still would like to comment that as the duration of time increases, the chances of remaining in 'the zone' drops. Hence the effectiveness of eg. repetition+speed drills -- forcing you to devote all your attention to the one thing, at which point you notice more things consciously but a huge quantity more subconsciously.

This also applies, IMO, if what you are practicing is 'finishing' -- you need to choose things that you can finish quickly, so you don't get stuck.
(actual paid work is different, of course. but paid work doesn't focus on learning, which is what I'm talking about here -- efficient learning processes. actual period of time spent on a single piece has to be short enough that you can sustain that 'HUP HUP HUP HUP! CHAAARGE!' mindset)


* by 'you' in this post I mean anybody. While these are mainly my observations of myself, my understanding is that they reflect general human cognitive behaviours.

</ramble>

hope some of that makes sense

[Ai]

@Helm: I’d say it could be about clusters or it could be about hiding the resolution (in ways other than well-organized clusters, like aa).

Might be worth noting: personally I think of AA as a type of cluster (with most of the pixels hidden), just like buffering (AA= huggy buffering in my mind . This probably effects my ability to work with higher numbers of shades favorably (eg. the gfx reworking I did for bub-n-bros)

My various routines (it would be nice if you could clarify what you meant by that Helm are oriented towards modelling real-world color interpolation.. in much the same way as I used vector experimenting and raytracing to develop my understanding of antialiasing. If I have an axe to grind here, it's that: people don't seem to relate these different areas of graphics work closely enough. They don't seem to use the high level tools (vector) to learn about the low level(pixel -- aa, cluster,..) anywhere near what they usefully could.
(personally I'd advocate alternating between vector and pixel in working on a vector-ish pixel piece. in this process I would feel pixeling such an image was not futile, but an exercise in modeling AA/shapes accurately)

anyway getting back on topic, I personally feel there is not such a thing as too much AA. rather there is such a thing as thoughtlessly applied AA. It's a thing to apply as a strategic art-direction choice, IMO -- the balance of AA in the image effects how the eye travels, just like the more macroscopic idea of focal blur.

My personal preference is significantly more AA than is common, particularly so you gain greater emphasis on non-AAed areas.

Possibly I'm boring in this cause I think most of these things are art direction choices, haha.
I'm even okay with demoscene stuff intentionally blurring away (or sharpening forth) stuff. I think it's somewhat of a bad artistic choice, OTOH it seems like good practice (in a narrow sense -- getting your head into AA at a deep level.)

[Rydin]

Some people use as many colors as they need and as big a canvas as they need and the end result is still pixel art and it's still sometimes beautiful.

Who are you thinking specifically?

[9_6]

Some people use as many colors as they need and as big a canvas as they need and the end result is still pixel art and it's still sometimes beautiful.

Who are you thinking specifically?

I think I can say I'm obsessed when I posted my most recent work...

http://www.wayofthepixel.net/pixelation/index.php?topic=10960.msg115379#msg115379

This you mean?