Pixel Art Tutorial Snippets

Discuss the development of other free/open-source games, as well as other games in general.

Moderator: Forum Moderators

User avatar
Jetrel
Posts: 7242
Joined: February 23rd, 2004, 3:36 am
Location: Midwest US

Pixel Art Tutorial Snippets

Post by Jetrel »

A lot of people have wished I could consolidate all of my advice into one place. Lots of times I'll give out one or two snippets of advice contextually, but I'll forget about them when I sit down to write a 'real' tutorial - so I figure I'll just gradually dump stuff in this thread every time I'm dropping advice somewhere else.

Constructive replies are welcome to this thread, but please keep the noise down. Anyone trolling about the validity of pixel art as a working medium is going to get my boot up their ass.



So to begin:
Don't obsess about color count. In fact, try to keep it "out of mind" entirely whilst working; simply by using pixel-art tools, your color count will naturally stay low. If you need a new color, always be willing to just add one; don't ever feel reluctant or hesitant to do that. Being hesitant about that is a massive time-drain when working on pixel art, and can heavily compromise the quality of your piece. Furthermore, it's probably one of the least-important factors in the quality of your piece; no one will care if you used a tiny color-set if the actual picture itself is [censored].

In fact more generally, no one cares about color count. "Well, gee, then - why bother doing pixel art at all?" The reason for doing pixel art is for the look. The unique, crisp look of it is the point to doing it, after all the hardware limitations and other fluff are gone. Other pixel-artists might care about the color-count, but the vast majority of your audience doesn't give one [censored] about either a low color-count, or especially about your adherence to the hardware specs of some ancient console platform. All they care about is the general look. A low color count helps maintain/create the look of being pixel art, but below ~25-50 colors you hit a point of extremely diminishing returns. Obsessing over adding a new color when you've only got 8 is a complete waste of your time. Just focus on making a good drawing, first.

So when do you need new colors? It depends on the size and curvature of your surface. Smaller art/surfaces can transition between wider "gaps" between colors. When you're trying to blend over a very big, very gradually-curved surface; that's when you need more colors. This is why over-sampling, and then trying to re-detail pixel art often ends up being very, very awkward - you can get away with having a low color count when the "gradually-curved" surface is only a few pixels wide, but when you upscale it to a large size, the colors aren't able to "natively" create the gradient, and you need to either dither or insert new colors to fill in the gradient.



Aside: Pixel art, like mosaics, like pointillism, and like cross-stitch just has a unique look to it - like these others, it may have been inspired by some fleeting technological/sociological gestalt, but just because of the raw appeal of how it looks, it will more or less be permanent as long as we have artists. People haven't quit doing pottery, wood-block printing, or doing cross-stitch, even though we've respectively got mass-production, laser-printers, and computer-driven sewing-machines to replace them. I could be wrong, sure, but based on historical precedent wherein I can't think of one single art form that's ever really died out, I think pixel-art is going to be with us for a long, long time. Especially with the mass numbers of "future humans with too much free time on their hands". (That said - if there's a tech singularity or a disaster-induced dark age, all bets are off.)
Play Frogatto & Friends - a finished, open-source adventure game!
User avatar
Jetrel
Posts: 7242
Joined: February 23rd, 2004, 3:36 am
Location: Midwest US

Re: Pixel Art Tutorial Snippets

Post by Jetrel »

Regarding the following image, and choosing palette colors:
FireDragon.png
FireDragon.png (17.87 KiB) Viewed 15732 times
Kaiseto wrote:Oh, wow. It's always beautiful to go searching through all the crap in the Pixel Art gallery of DA and find an absolute gem by somebody who knows what they're doing and takes it seriously This is stellar. I see that there's a green entry in the palette? How did you choose that? It works with the final composition of the piece, but it's something I'd never think to put into such a heavily orange/yellow color palette.
The grand secret to doing crazy coloring is that as long as you get the luminosity right, you really can go crazy with the colors - and this is NOT given the caveat: "within reason". Seriously; go wild. You can go way beyond what seems legal, and the piece will still generally work if you just get the brightness right. (If it doesn't, it's trivial to "rein in" and return it to sane colors, because hey - it's pixel art.

More specifically, the green entry was because metals, especially brass and gold often tend towards greenish (also, dark yellow, which we see as greenish) tones in their shadows. Also, the other thing with the green is environmental lighting; ingame, the dragon often shows up on grass (the rest of the time it'd be on brown cave stones).

But probably the biggest thing is that green is a contrasting color to purple; pairing contrasting colors like purple and green, blue and yellow, blue and red will result in a "vibrating edge". A vibrating edge like that tends to be annoying if it's used in huge areas (imagine a primary-blue billboard with primary red letters - ouch), but when you use it for little bits of shading and such on pixel art, it gives it a sort of sparkle and shine. Generally speaking, that's been one of my big attractions to pixel art; the bizarre "fire/sparkle" it has, that most large-scale art doesn't.
Play Frogatto & Friends - a finished, open-source adventure game!
User avatar
shiremct
Art Contributor
Posts: 116
Joined: August 1st, 2007, 9:23 pm

Re: Pixel Art Tutorial Snippets

Post by shiremct »

This is the number one thing I am working to improve myself at the moment and is, in my opinion, one of the most important things you can learn to make "good" work into "amazing!" work. Stepping through lighter and darker shades of the same hue (forgive me if I'm misusing terms) can indicate depth adequately, but actually using different colors, not just lighter and darker versions of the same, really enhances the depth and makes the image naturally 'pop' a lot more. Using this concept to enhance something's "sparkle" never occurred to me, but I'm going to try to keep this in the mind the next time i'm working on highlights as well as the shadows.

The way the purples and green interact with yellow shades in the tail give an excellent sense of shape and depth; That's something you just can't achieve with lighter and darker shades of straight yellow and orange.

Unfortunately, this is something that takes a lot of practice and training yourself into this way of thinking (at least for those without formal art studies and such.) I'm finding it very difficult to work into my work, even when I'm trying to, and when I do, it isn't always a success... but the places where I have worked it in and it's worked, it has really improved the quality of my work, even when it's just in a small portion of the total image. I know if I can improve my use of this concept to encompass the entirety of an image, it will make my art much better. It's not an easy concept to embrace and put to use, but I think any artist looking to improve their work, and this concept holds true in almost any medium i believe, should try to actively be aware of varying their palette choices. It really does make a big difference in the quality.
User avatar
Jetrel
Posts: 7242
Joined: February 23rd, 2004, 3:36 am
Location: Midwest US

Re: Pixel Art Tutorial Snippets

Post by Jetrel »

@shiremct: I could have written that post a couple years ago. :)

This was also a good post:
http://www.wesnoth.org/forum/viewtopic. ... 70#p465470
Play Frogatto & Friends - a finished, open-source adventure game!
User avatar
Sgt. Groovy
Art Contributor
Posts: 1471
Joined: May 22nd, 2006, 9:15 pm
Location: Helsinki

Re: Pixel Art Tutorial Snippets

Post by Sgt. Groovy »

If dithering should always add texture, I'm interested to know how sky should be rendered in pixel art. A clear sky is one of the few natural objects that has a very smooth gradient with absolutely no texture.
Tiedäthän kuinka pelataan.
Tiedäthän, vihtahousua vastaan.
Tiedäthän, solmu kravatin, se kantaa niin synnit
kuin syntien tekijätkin.
User avatar
Jetrel
Posts: 7242
Joined: February 23rd, 2004, 3:36 am
Location: Midwest US

Re: Pixel Art Tutorial Snippets

Post by Jetrel »

Sgt. Groovy wrote:If dithering should always add texture, I'm interested to know how sky should be rendered in pixel art. A clear sky is one of the few natural objects that has a very smooth gradient with absolutely no texture.
Skies are hard to do in pixel-art. Generally, speaking... people tend to avoid doing gradients and just do a solid tone (c.f. mario). Or if they do gradients, they tend to either obscure the blank sky with clouds, blend the horizon gradient with clouds, or follow a very regular/geometric dithering pattern (frogatto did the latter). Very often, in order to tile vertically without a hard ceiling, they'll abstain from gradienting the upper-half of the sky; instead just blending from a horizon gradient into a solid tone.


http://www.pixeljoint.com/pixelart/49328.htm
http://www.pixeljoint.com/pixelart/54911.htm
http://www.pixeljoint.com/pixelart/50444.htm
http://www.pixeljoint.com/pixelart/46331.htm
http://www.pixeljoint.com/pixelart/43859.htm
http://www.pixeljoint.com/pixelart/42575.htm
http://www.pixeljoint.com/pixelart/39173.htm
http://www.pixeljoint.com/pixelart/50305.htm
http://www.pixeljoint.com/pixelart/19806.htm
http://www.pixeljoint.com/pixelart/33605.htm
http://www.pixeljoint.com/pixelart/36838.htm
Play Frogatto & Friends - a finished, open-source adventure game!
User avatar
Jetrel
Posts: 7242
Joined: February 23rd, 2004, 3:36 am
Location: Midwest US

Re: Pixel Art Tutorial Snippets

Post by Jetrel »

Long post on another forum in response to some guy with a perennial problem: :geek: You've all heard this before; I've certainly said this same stuff dozens of times. Which of course is what this thread is for.
danlthemanl wrote: I can't make art. I've tried, I find myself not being able to make my games because of my lack of art! I don't want to use public domain assets because someone else might be using them and it might look like I'm copying them. I've tried 2D, 3D (waay to hard), I can't do it! Please help :o
Use science!

I was in exactly the same boat as you 4-5 years ago. Zero talent at art - had poked at it, and obviously didn't have "the gift." I was in the middle of college, and was getting really worried because I couldn't get anyone to make art for my games. With my back against the wall, I decided to try anyways, figuring I'd at least learn how to work with artists even if I couldn't become one.

For the first year or so, I approached art with no methodology at all. I just tried doing stuff by "feel", and it pretty much all sucked. After that, I stepped back, and said "okay, hold on - I'm an engineer. Let's try thinking about this problem like an engineer." Let's form hypotheses about what I'm doing wrong. Let's lock variables. Let's test ideas with heavy iteration. It worked.

Most of art is science. Shading .. is merely running a human-optimized raytracing algorithm in your brain. We're poor at casting individual rays, but we're amazing at interpolation. Anatomy is fairly cut and dry - it tells you how the human body can - or cannot, bend and twist. Bend it the right way, your drawing looks graceful. Bend it the wrong way? Your drawing looks uncomfortable. Optics tells us how the eye sees different colors, and how colors are related to each other. Optics tells us you can do stuff like putting grey next to blue, and the grey will look yellowish.


Anyways, besides all the theory, the scientific method can really help as you practice. The biggest danger with practice is that practice alone won't make you good. That's really scary. I've seen people churn out dozens of drawings every week; a huge volume more work than I do, and they've been doing it for years without getting any better. I'd link some particular webcomic artists, but to avoid being a dick, I'll just say that some members of the top-50 comics out there are perennially terrible*, and have put out thousands of pages. So what gives?

The key to having practice actually make you good, is that you must change how you do things, and test if that change makes your work better. You can't just repeat the exact same recipe-for-failure you've always been using, and expect the results to come out differently - they won't. Same input, same output. Practice simply gives you the time to make these tests.

Studies are a key way to do this. The trick with studies, is that they're optimized to get the highest concentration of "tests", in the least amount of time. If you're practicing positioning the nose on a face - the easiest optimization, is just to forget about drawing the body at all. You'll see artists do this all the time - whole pages of nothing but faces, over and over. Or hands, or legs, or whatever the focus is. Likewise, they'll be very sketchy - if the sketch is enough to see if it's right or wrong, why waste 10x as much time "cleaning up" each individual drawing of a leg, when you could do 10x as many trials? 10x as many trials, is 10x as much learning! A key trick in doing studies is to stop after you've crossed the point of diminishing returns for "learning value", and get on to another study.


Another idea I mentioned is locking variables. When testing hypotheses, it's really hard to know if they're really the deciding factor, if everything else is changing too. Like, if you're trying to get the hang of drawing features on faces, it can be nigh-impossible to tell if you're getting it right, if every time you draw the face, the face itself is pointing in a different direction. So do dozens of tries with the same face. Digital tools are a great way to do this, because not only can you cut and paste, but "trying something, then pressing undo, and then trying it again" is itself a huge form of iteration. If, every single time you try to do that, you're drawing a nose onto a face, pretty soon, you'll get the hang of putting it in the right place.
John Sandoval wrote:A piece of advice I would give to the OP is to just ape the [censored] out of artists that you like. Maybe this is a controversial statement, but it's a really quick way to learn.

But when I say 'ape', I don't mean the outright tracing/plagiarism you find so readily on sites like Deviantart. To clarify, you should see what kind of techniques a given artist uses and take from them what you will.
It's controversial only to some modern people with delusions of ownership being a 'natural right'. It's perfectly legal under fair use (as long as you don't claim the work as yours), and it's a time-honored tradition of art known as "Studying from the Masters". In renaissance times, people would do exactly that - they'd actually sit down, as a group, and meticulously practice drawing parts of the master's work - such as directly imitating something from da vinci. Da vinci himself did it, and recommended the practice. It was and still is (at least at some schools) a key part of curriculum. You should always attempt making your own solutions to these "engineering problems", but it's also very good to study how other people solved the same problem. Sometimes it's just not possible to come up with a new, innovative way to it; so use what works.

In fact, back in the day, this used to actually be legal trade. Quite a few people made a white-market (e.g. not illegal, but encouraged) living making duplicates of masterpieces; especially since at the time, there was no photography, and no other means to mass-produce a work if you wanted one for your home (understandably, due to the labor involved, it was expensive enough that only rich people could buy them), but it's interesting that these were not viewed as deceptive "forgeries". They were openly copied, and everyone was fine with the practice. If you've ever read Nathaniel Hawthorne's The Marble Faun, he describes this practice in great detail, and it's a central story element that one of the leading characters was particularly skilled as a copyist; able to infuse her work with a "fire and soul" that some people felt exceeded the originals (IIRC).

But the point is that studying from the masters is a great way to "lock variables", and piggyback on something they kick ass at, but which you're terrible at. You can imitate aspects of their work, in order to practice some other skill of yours - perhaps, you can trace over some professional's anatomy ... in order to practice putting clothes on figures? This can't be sold or distributed as your own work, but it's perfectly fair game for practice. Much more common, is inking and coloring someone else's line art, to practice your coloring and shading.

Also, as John said, you can side-by-side, attempt to recreate the aspects of their work, which you found impressive, and by walking in their shoes, you're walking through solving the same problems they did. If you just "trace" their work, you won't learn anything, but if you force yourself to solve the same problems they did, using your own knowledge and skills, the master's work is an excellent way to tell if your methods are working or not. For example, maybe you're drawing an arm in a similar pose. You look back at the master-work, and realize there's a certain curve to part of their arm that just isn't in your repertoire. Clearly, they (probably**) know something you don't, and you've stumbled on something new to learn! So then you bust open your anatomy book and discover a whole new muscle (or gap between) there (for a simple example, the gap between the triceps and deltoid).





Counterpoint: So I gave all this long jive about being systematic to improve your skill. Be willing to compromise everything I just suggested, for one key factor: keep art fun. If it stops being fun, you stop practicing. No practice, no skill gain. Dead stop. Keeping stuff fun, at all, can actually be incredibly hard. Balancing fun with also frequently doing it is harder still - you won't get very far if the only time you practice drawing is just once or twice a month, on those rare occasions when you "actually feel like it".

I have a bad feeling that the unlucky "rest of us" who don't naturally enjoy drawing 24/7 basically have to "brain-hack" ourselves into making it fun when it wouldn't naturally be. I still have a great deal of difficulty with this, and no bulletproof answers - I find myself cursing my ennui and laziness all the time.




* This speaks to an important point. Art quality matters, but this stuff isn't just one dimension of quality. People can like a work even if it's art is terrible - or much more commonly, with comics, it is different dimensions in just the art, that can outweigh each other. Comic artists sometimes have bad anatomy, but are liked because their characters are funny and expressive and in that particular usage, that matters more. Not something to hang on your wall, but something that manages to make you laugh.

** obviously, they're not perfect, which is why you need to supplement any "study from the masters" with "studying from life". Spend too much time imitating other artists, rather than imitating real-life, and you run a danger of being like Rob Liefeld
Play Frogatto & Friends - a finished, open-source adventure game!
User avatar
artisticdude
Moderator Emeritus
Posts: 2424
Joined: December 15th, 2009, 12:37 pm
Location: Somewhere in the middle of everything

Re: Pixel Art Tutorial Snippets

Post by artisticdude »

Jetrel wrote:For the first year or so, I approached art with no methodology at all. I just tried doing stuff by "feel", and it pretty much all sucked. After that, I stepped back, and said "okay, hold on - I'm an engineer. Let's try thinking about this problem like an engineer." Let's form hypotheses about what I'm doing wrong. Let's lock variables. Let's test ideas with heavy iteration. It worked.
Wow, thanks so much for that last post! :D That's exactly the problem I've been having over the past couple months... doing stuff 'by feel'. As a consequence, I've been kind of slow in developing my artistic skills in the last couple months, because I felt like I wasn't getting anywhere despite the work I was putting into my projects.
Jetrel wrote:I was in exactly the same boat as you 4-5 years ago.
:o Wow, you got so good that quickly? I would have thought it would take, like, at least 10 years to get to that level. I suppose that's a result of the engineering methodology. 8)
"I'm never wrong. One time I thought I was wrong, but I was mistaken."
User avatar
Jetrel
Posts: 7242
Joined: February 23rd, 2004, 3:36 am
Location: Midwest US

Re: Pixel Art Tutorial Snippets

Post by Jetrel »

artisticdude wrote:Wow, thanks so much for that last post! :D That's exactly the problem I've been having over the past couple months... doing stuff 'by feel'. As a consequence, I've been kind of slow in developing my artistic skills in the last couple months, because I felt like I wasn't getting anywhere despite the work I was putting into my projects.
Don't get me wrong - there's still tons of "feel" involved in it. I haven't switched over to producing art like a robot, which would have the obvious result of accurate-but-lifeless drawings. What I've done instead is combined the best of both worlds.

Going purely by feel just randomly changing stuff without rhyme or reason, and just doing whatever feels most appealing. What I'm doing is applying the engineering mindset to the intuitive response. I'm asking "why does this feel better? Are there tangible, comprehensible reasons for it? can I reproduce these results? can I strategize about future changes that I try?"

Without the strategizing, one essentially is applying a "brute force, random walk" to their skillset, and hoping it'll magically land on some state that's appealing to people. As we know from cryptography, brute force .. takes a long while. Dangerously, here, it could easily be longer than a human lifetime. :(
artisticdude wrote:
Jetrel wrote:I was in exactly the same boat as you 4-5 years ago.
:o Wow, you got so good that quickly? I would have thought it would take, like, at least 10 years to get to that level. I suppose that's a result of the engineering methodology. 8)
Yeah. I'm now 27 (my how time flies :cry: ), but yeah, I pretty much didn't start art until I was 21. It was just completely obvious that I didn't have any talent for it, so why bother trying if you're doomed to be a failure?

There's a very, very pervasive bit of conventional wisdom out there, about how different people are born with different talents, and that's all they'll naturally be good at. In a lot of people's minds, this takes what a shape that, when you really think about it, is hilariously RPG-like. That is, if you're talented at something, you'll progress at a faster rate in gaining skill at it, and you'll be open to a higher skill ceiling at it. The corollary of this is that it's futile to play against your "natural class".

:hmm: I am really, really coming to believe that "talent" - in fact, intelligence writ large, is not a matter of hard-wired brainpower, but is instead a matter of thought-patterns. Cadence of how long you pursue individual threads of thinking, etc. How much you're able to hold in working memory, and with what integrity (human memory seems to be lossy-compressed). How often you pursue ... or cull, branches of thought; what your algorithm for pruning your tree of potential thoughts is. The ability to swap-in-and-out different algorithms for this, appropriate to different situations, seems to be a root of intelligence.

This was the big thing that really hit me during my physics education; I had spent most of my life in my "comfort zone" in terms of thinking. Physics finally was complicated enough that I was forced to adopt different thinking patterns - if I didn't, some of the material was just incomprehensible. I had to do something, or I just couldn't keep it all in my head.


Stupidity really seems to be mental stubbornness. People have chosen to think in a certain style, even if that style is just horrible at getting the job done. From my own experience, at least at first, it's incredibly painful; in fact in some ways feels like an incredible personal violation, to change how you think - because it's changing who you essentially are, as a person. The assertion that how you think is "wrong" is incredibly offending, because it's like saying that everything you are inside; your soul; everything that makes you, yourself, is wrong. It's like the ultimate thing people cling to as their personal identity.
Play Frogatto & Friends - a finished, open-source adventure game!
User avatar
Jetrel
Posts: 7242
Joined: February 23rd, 2004, 3:36 am
Location: Midwest US

Re: Pixel Art Tutorial Snippets

Post by Jetrel »

GILBERT Timmy wrote: The reason sprite like DK or wario are not considered pixel art, is that they are not treat with this technics, they are rendered target of different artistic practice downscale by different process. You can say pixel art is the artistic technic construction of an image.

There is also the style pixel art which derive from the technic and is generally featured by big blocky pixel and retro look.

Sprite is more precisely the technic of using an image in the context of game and define by it's re-usability in many context in contrast to one use image like in animation or one context image like title illustration.
Yeah, sprites are any kind of source art, which is rasterized into premade 2d images of each frame, and animated by blitting those to the screen.  So if you had 3d art, if you're actually rendering the models in realtime, it's not sprites.  If you've pre-rendered them, it is sprites.  Ergo, diablo 1 and 2, pre-raytracing aside, are sprite-based games.  Diablo 3 is not.


Pixel art refers to creating an image where the fact that it has finite resolution is a fundamental part of the piece; that is, where the choice of specific placement of pixels matters.  Pixel art can be any size, but it's such that if you resize a piece, you're losing/adding meaningful choices in the construction.  Such as "I placed these pixels in this exact arrangement because this suggests a slightly crooked eyebrow".  If you double the resolution, you have to completely redo those choices.  If you half the resolution, you might have to decide not to even draw eyebrows.  That's what makes it pixel art - that choices like that were a fundamental part of the piece.  Cross stitching is pixel art; mosaics are pixel art.  Pixel means "picture element"; it means the same thing even for non-computer mediums that are constructed out of individual (usually uniform) chunks of color.

But stuff like cel-shading really likes to pretend it has infinite resolution.  This is why artists always upscale their working res to have the smoothest possible lines.  If you rescale a piece of cel-shaded art, you really don't have to rethink how you drew the details.  Even when it comes to organic stuff like "paint/chalk/charcoal" style lines/fills, artists doing those would prefer they always look that way at any size - that if you keep zooming in on the chalk lines, they remain looking like chalk lines.  That is, that their natural patterns would fractally repeat themselves up to any amount of detail.  (There are exceptions who want their work to pixelate, but by wanting that, they become sort of a hybrid of pixel and CG art.)
Play Frogatto & Friends - a finished, open-source adventure game!
User avatar
Jetrel
Posts: 7242
Joined: February 23rd, 2004, 3:36 am
Location: Midwest US

Re: Pixel Art Tutorial Snippets

Post by Jetrel »

Writing to a guy who asked me about pixel art workflow:


Mirroring:
Audiences generally don't care about mirroring. There are many other things that bug them a lot more; it's pretty low on the "breaking immersion" scale. Mostly, mirroring in sprites only becomes an issue when you "rotate" the sprite around its vertical axis (e.g. having a RPG or platform character, who is drawn mostly from a side-on perspective, "spin" like a ballerina/martial-artist), and it has some *major* bit of asymmetric costuming. (this sort of animation was a common motif in chrono trigger, for one example).

The one huge no-no in mirroring is abandoning it, and brute-forcing an asymmetric costume design by drawing the different directions. It creates about 50-100% more work for virtually no gain at all. (Mind you, this is completely benign if your project is a prerendered 3d->2d game (e.g. diablo, age of empires), since the renderer does all the work for you). Sadly I've actually seen at least one project do this wrong, and it was part of an endemic problem which eventually sank the project - quite relevant to your paper, actually. Basically, the director of the project planned on having art assets, without any regard to how long they'd take to make (regularly adding new categories of stuff to do, and generally asking for an impossibly high degree of resolution and animation). This would be a terrible decision on a budgeted project, but this was an open-source project, and ... he had no budget. He was relying on volunteer effort by amateur-level contributors, and somehow expected work that exceeded fulltime teams of ~ half a dozen professionals.


Asset Recreation:
Huge, huge problem. This is the bullet that killed Duke Nukem, Daikatana, and many others. In fact, there are probably many more games that succumb to this than we ever hear about, since I know of quite a few developers who are shy about showing their stuff off. For every success like "cave story", there are probably several dozen games that are equally good or better, but which never get finished because the author keeps heavily reworking stuff over and over and over. Examples: http://darkfalzx.blogspot.com/ http://carnivacgames.blogspot.com/ Frogatto itself had some problems with this; although the boost in graphical quality was comparable to "warcraft 1 -> 2", we could have finished the game nearly a year earlier if we'd not rebuilt everything. Note that this applies almost identically to code. Daikatana committed suicide by switching engines from quake 1-> 2. Their reason? Visual quality.

I'm as big a sinner as anyone, but almost always, you should tie off what you've got, make it a finished title, and then do your improvements as a new sequel, if they're really compelling, rather than doing the same amount of work and having only one game to show for it.


Asset Recreation Counterpoint:
That said, you can't just sling everything out and call it good. You're going to inevitably make some mistakes which are so bad they *must* get redone. Frequently, I end up making animations outside of the game, and then porting them in and finding out they jar badly with character movement in the actual game. A good strategy to deal with this is to _sketch_ the animations; literally going as primitive as stick-figure/skeleton animations, or blocking with blobs of color (my preference). Once sketched, I directly import just the sketch into the game, and feel-out how it plays before committing time to polishing it. It's bad if you do this to one character, but it can be devastating if (for whatever reason) you make a large portion of the art assets before having a chance to try them in game.

The same thing applies to tiles - you might make a mockup with a great-looking group of tiles, but only when you actually try it in game, find out that it has major issues (be it tiling, repetition, etc). If you don't test this contemporaneously with developing the thing, you'll throw effort at fixing spurious problems (such as making duplicates for some tiles that aren't used much in practice), and will miss real problems (such as some motif in a tile being combinatorially impossible to make tile nicely). A mockup is a lie - stuff that looks great in a still image might not hold up in an actual game that tests it from all fronts. Get it in an actual engine ASAP! (This is basic prototyping, and applies to every part of the game, but it's so important that it bears repeating).


Less is more:
This is dead on. Pixel art is in its forte, is in the domain where it outshines other types of art for 2d sprite rendering, at sizes below 50x50 px. There's not enough space to directly render all 'features' of a character (such as facial features). In a medium where you have no direct control over individual pixels (3d, whether prerendered or not), any feature smaller than a pixel basically gets blurred into a skin-colored soup. Consider the faces of player characters in diablo 2. Even with its cartoonishly big heads, warcraft 3 had the same issue; blurry, static faces. In pixel art, you're able to control every pixel, and thus, you can choose to render - or NOT render, sub-pixel features at will. By not rendering the unimportant ones, you can put great emphasis on what is important.

To consider a few squaresoft SNES classics, both the famous "chrono trigger" and the more obscure "seiken densetsu 3" were notable for having expressive, readable faces, able to show dozens of distinct emotions, even though their sprite's head sizes were similar to those of other games like diablo 2. In fact, they even managed to pull off readable fingers(!) - a character could pull out a flute, and you could see the character's fingers lifting and falling on the holes of the flute. That'd be nigh-unthinkable in diablo - in fact diablo's modelers may not have even made individually articulated fingers on their player-models. This was possible in part because of what they chose to simplify - when playing the flute, their hands were simplified to a mere 3 fingers. In drawing their faces - especially in chrono trigger, the characters did not have mouths, unless the mouth was actually open. Nor did they have noses.

A clever expression I've heard for this is that "regular drawing is prose, and pixel art is poetry". Specifically, pixel art is the short, compact kind of poetry, like a haiku. Selective omission, to allow other selective exaggeration, is the essence of the craft. Features should not be directly rendered, but should instead be simply "suggested" by simplifying them into a general geometric shape. A posed hand is often just a flesh-colored blob in the shape that pose of the hand would take. The thousands of strands of hair which might actually get drawn on a regular-sized head, should instead get drawn as just a few, unbroken "locks" of hair on a pixel figure.


Concept Art:
Contrary to 'industry' expectations I never use concept art for pixel games. Ever. This is perhaps a way of saying that my pixel art IS my concept art. When making a pixel-game, a dedicated "concept artist" is only indirectly useful - unless they double as a creator of finished non-pixel story/character/chapter/etc art, they can't actually create any assets for you, and you could hire a second pixel artist, thus doubling your production, for the same cost. Second of all, any pixel artist unable to come up with good concepts on their own is a terrible pixel artist. It's as basic to the skillset as doing concepts is to illustration; if you hired an illustrator to do character profiles, the likelihood of them being great at doing character drawings, and bad at coming up with visual character concepts is virtually nil, because there's so much overlap between the fields.

Last but most important, by doing my concepts directly as final assets, I never have any disjoint at all between what I'm concepting, and what I'm able to portray. I never create character features/costuming/etc that are impossible to render in pixel-art. Everything I create is immediately proven "feasible" in the game I'm working on.


Workload:
http://www.gamasutra.com/blogs/AdamSalt ... elines.php
As Adam "Adam Atomic" Saltzman noted in a recent gamasutra interview, pixel art's control over individual pixels has a downside in that the higher resolution you have, the more work your artists have to do. It scales to the power of 2, and increased resolution doesn't give very much benefit - before long, you're outside the sweet spot of pixel art, and having that control over individual pixels becomes a negligible benefit, at an extraordinary cost. Moving from a 320x240 canvas, to a 1280x1028 canvas isn't 4x as much work - it's _16x_. To say nothing of the additional multiplier Adam talked about in doing animations.
Play Frogatto & Friends - a finished, open-source adventure game!
User avatar
Jetrel
Posts: 7242
Joined: February 23rd, 2004, 3:36 am
Location: Midwest US

Re: Pixel Art Tutorial Snippets

Post by Jetrel »

Something I posted over on the (pretty much dead) allacrost project's forums:

nicknameminivan wrote:I can play the game in a high resolution, but WHY :twitch: are the sprites so small?
It is absurd to have a 10x10px sprites on a 1028x768px screenresolution. It looks bad. The sprite have to be at least 50x50px.....
The bosses looks good.
On these old screenshot you can see a much bigger sprite and i looks so much better
http://www.allacrost.org/media/screensh ... en_03d.jpg
I'm hopefully doing an educational "Public Service" by responding to this. This is sort of an entry for my burgeoning pixel-art tutorial I haven't finished yet.

Increased resolution in a game is in itself, neither beneficial nor harmful. It's simply a technical bullet point - increased resolution only benefits you if you're able to fill all of that additional resolution with additional details. And I specifically mean new details; not just higher-res, sharper textures, but higher res textures that actually show new stuff. For a simple example of improving a low-rez tree, this would mean going from an amorphous green blob, to actually showing individual leaves. A high-rez "amorphous green blob" isn't an improvement.

The benefits of additional detail are:
1] better clarity makes in-game status/cues more comprehensible. For example, if you have a 3d wargame with enough art+scripting muscle that you're able to switch wounded soldiers to a clearly "wounded" graphical state, you can convey useful information without cluttering the screen with hp bars. (Age of Empires 3 does this with building damage). It can spare you from having to clumsily convey things like emotions through separate portraits - these convey much better if you can just look at someone's face, and see their emotion just like a real person would display it.
2] piggybacking on the former, high-level cinematic "emotional involvement" doesn't work unless you can really do human interaction at the same level of involvement as a movie; unless you can look closely and read every little facial tick of a persons face and see into the "inner world" of the character. They say "the eyes are the mirror of the soul", and if you can't look a character in the eyes and have them look convincingly "true" (whether as a cartoon, or a realistic render), the emotions just don't travel in the same way.
3] It looks cool.


These are all nice benefits, but at the same time, it must be remembered that all of this comes at a steep cost; even if it doesn't (as it often will) incur a performance loss, it costs an enormous amount of work on the part of modellers, texturers, level decorators, and just artists in general to fill in all the details with unique, valuable detail. Even on a project like this, with no monetary budget, you are working within a time-effort budget: how much are you, and the people on the internet at large willing to work on this? The key thing to take to heart is that wiki-magic/deus-ex-internet isn't going to save you - contrary to popular belief, the internet isn't going to show up and do an unlimited amount of work for you. This resource is limited!

I'm not some armchair philosophizer, here - I'm saying this as the art director for Battle for Wesnoth, a contender for the most successful open-source community-driven game ever made. I have some experience in these matters. There are a lot of factors in how much contribution you can get from the internet, but in my experience, it's quite limited, and the key limitation is that you've got to finish a more-limited-in-design game before others come in and start building on your core. You can't have the start of a really impressive game, you need to have the entirety of a modest-looking, but totally complete and very enjoyable game for people to help out, or even give a [censored] at all. Thus the thesis statement is: Do not count on internet contributions at all, budget your design based only on what your current team can do.

Budget based on time; you're not going to be able to work on this more than a few years; likely 5-10 at the top. Life happens. Team members get things they care about more (wife, kids, dream job). Sometimes they die. Far more than any cocky wannabe game dev will ever admit, most people realize after a few years that game development actually sucks - it's a hobby that basically requires you to spend the time equivalent of a second, full-time job to get anything done in years, rather than decades (at least, if you're doing anything more complicated than a cheap little puzzle game). So the team and momentum WILL break up, and you've got to plan small enough to have it done before then.



So, the postmortem regarding allacrost in particular:
Roots chose to make the screen 1024x768, and stood steadfastly by that decision. I don't know why, I guess he just couldn't bear the thought of being anything less than (what was at the time) native res for many screens.

This isn't inherently a bad decision, except that he also chose to be a pixel-art game, because that's what he was copying (FF5), and that's the combination that proved toxic. Pixel art games can't just be made bigger, because for pixel-art, you have to manually place every pixel; an increase in scale is a proportional increase in work. It also defeats the purpose of pixel-art, which is making cool exploitation of individual arrangements of pixels - pixel art allows you to essentially "hack" the medium to show more detail at a tiny size than other forms of art (vector, cg, 3d) are able to. By individually controlling the placement of pixels, you can make visible eyes and faces on characters which you couldn't in 3d, because you don't have that control (the rendered humans for example, in AoE3 are mildly bigger than chrono trigger's, with surprisingly similar proportions for heads and such, but only chrono trigger has visible facial expressions. But once you're over a certain size, this advantage disappears, and the costs for still being pixel-art skyrocket. I would say this advantage really hits the foul side of cost/benefits when your character is over 100x100 pixels, or when your environment is bigger than 300x300 pixels.

Allacrost's sprites are tiny because that's all the art team could do - in fact, they couldn't even do "a whole game's worth of" that. I was one of them, it was too much work, and we all gave up. One of the reasons for this mistake was that we as artists were simply dumb in being too cocky; but the other reason is we were trying to fill a canvas leagues bigger than our abilities. We would be lucky to fill a the canvas of a SNES game - 256x224px, which pushed the teams of geniuses at squaresoft et al to their limits. We didn't know if we could do that, alone; match up with the pros - that would have been an enormous accomplishment in its own right. To then go further and choose to fill a canvas roughly 14 times the size, was far past "daring and foolhardy", and was just plain stupid.



Yeah, but with 3d...
Can this be avoided by just doing (prerendered or realtime) 3d (or vector art for side-on stuff) instead of pixel art? Sort of. Yes, you of course can trivially render your assets at any size. But you still have the same trap of "wanting more meaningful detail", and that's the real killer.

Pixel-art simply exacerbates this issue because it forces you to put meaningful detail into every pixel. But 3d art has the same trap - you still have to make tons of meaningful decisions about details in your models and their animations. You still have to make lots of meaningful decisions as you draw the textures for these models, and decorate an environment with them. You still run the core risk of wanting a game environment more detailed than your art team can provide. I've watched thousands of 3d projects destroyed by this, and it's a damn shame. They've got like .. 2 modellers, but they're shooting for the same quality as AAA teams that sometimes have over a hundred, these days. Many of these projects would have succeeded with a cool, simple, low-poly look like an N64 game. Minecraft did that, and it worked; Fez is doing that too.

Don't overbudget the time it will take to finish your project, or you will fail miserably. I guarantee it.
Play Frogatto & Friends - a finished, open-source adventure game!
User avatar
Jetrel
Posts: 7242
Joined: February 23rd, 2004, 3:36 am
Location: Midwest US

Re: Pixel Art Tutorial Snippets

Post by Jetrel »

more directed at game-making in general, but:

Another idea for why to KISS:
The hardest thing about working on a game is maintaining motivation. I hate to use the term "honeymoon", but that's pretty much exactly how it works - there's this initial period of awesome motivation, when you feel like kicking ass all day long and you get tons done. But then, no matter how cool your project is, you've just been way too exposed to it, and it loses the magic. It really has nothing to do with the quality of the project; no matter what it is it just stops being fun once the novelty wears off. I suspect even if you're "john carmack on the final laps of finishing doom for release", every game project eventually starts to suck because you want to move on and do something else already.

Sometimes the magic will come back, but the key thing is that there WILL be some major periods of time when you hate your project. Half of those times you'll just want to reboot and do it differently. It's times like that that you need to stick to your guns and actually -finish- the project you start. Not reboot it, not switch to doing something else - finish it. It hurts, but it's the key different between being an *actual* game developer, and a "wannabe" game developer.

The huge reason we always tell people to start small isn't that they aren't *able* to finish stuff; most people are skilled enough from free-time hobby-exploration that they're fast/skilled enough to finish a game. The problem is that by the time they're halfways there - they usually no longer *want* to. If you keep it small, then finishing it after you've started to hate it is easy. You knuckle down for a week or two and you're done. But if your game is this huge, ridiculous, sprawling dream-project, there's no way you can finish it after you start to hate it.
Play Frogatto & Friends - a finished, open-source adventure game!
User avatar
Frogger5
Posts: 951
Joined: November 7th, 2009, 11:45 pm

Re: Pixel Art Tutorial Snippets

Post by Frogger5 »

That is so true. And not just for games either, books, art, music. I have had those times when writing when I just want to start all over again in the hope of it being better, and it usually end up with me abandoning it. My willpower sucks.
My spritework can be seen here.

Want to play Roll 2 Dodge, or even start your own game?http://rolltododge.freeforums.org/index.php We need you!
Stilgar
Posts: 465
Joined: January 21st, 2006, 8:22 pm

Re: Pixel Art Tutorial Snippets

Post by Stilgar »

Regarding what you said about Allacrost, how has the experience with Wesnoth differed, in your opinion? Wesnoth's sprites/screen size are on the large side according to the measures you gave, too.

Also, would you have any handy tips when it comes to breaking out of a "rigid" feel in sprites due to difficulties with the concept of a grid of squares getting in the way of planning out more free shapes?
Post Reply