Saturday, April 13, 2013

Pixel Art Lessons: Skypehopert's "At Sunset" (Banding/AA, Low Colors, "Story")
Not always a Jeremy Soule fan, I think much of each of his soundtracks ends up as ambient filler, but there's always a few songs on each of them that is really good. This is one of 'em, from Guild Wars 2. I'm no MMO fan, and the game ain't quite my speed, but the soundtrack is pretty good, even for Soule's standards (which, while I'm iffy on much of his work, the production value is always sky-high).

Hey, sorry for the bit of delay on the lesson this week. I was busy the rest of Wednesday and I've been having headaches all week that have discouraged me from writing.

This week is a little different. While I gave crit to the creator of this one, I didn't do quite as much as I'm gonna talk about today. So really, today is a "I take a piece, talk for a moment, then do an edit as demonstration of principles" day.

Skypehopert's "AT SUNSET"

Major Themes: Banding/AA, Low Colors, Story-in-Art

 The main villain wants to destroy the fairy-tale world by nuclear bomb.
This piece came as an entry to the weekly challenge on PixelJoint. Here's the restrictions on it:
Create an image including at least two panels of a new and original comic story featuring the dawn of a new super hero or super villain, such as the radioactive spider biting Peter Parker or the symbiote infecting Eddie Brock.
Canvas Size - Unrestricted.
Colours - Max 10.
Transparency - Optional.
Animation - No.
Now, there's some factors in there. First off: yes, if you've been paying attention, I entered this competition myself, so Skypehopert was technically a competitor. Ain't no hard feelings there though. Second, there's three significant restrictions on the art:
* At least two Panels,
* Origin of a Superhero/Villain
* Low colors, maximum of 10.
Here's my original commentary:
Really liking the visual style in this, and the sixth and seventh panels (the close-ups on the gas-mask man) are pretty freakin' awesome looking, composition-wise. You made some kick-ass color choices too.
The biggest trouble is that I don't see the "dawn of a new SUPER villain." It's really a pretty fine distinction though, I'm very impressed.
On a pixel scale though, this has a bunch of issues, most notably some crazy banding going on. The ore and what it's resting on have some problems defining themselves, and many of the lines could do with some smoothing. Several spelling issues (and needless abbreviations) break the really great style, but if you're not natively English that's not a big deal really.
Really liking the look of it all though!
No lie, I'm way big on the style. Let's see how well he did with the minimum restrictions
* At least two panels. No problem!
* Origin. Personally, I'm not entirely sure I buy it, since when I think Super Villain, I think of them having some sort of Superpower, rather than just being a villain. Bond villains aren't supervillains, in the same way that I personally don't consider Batman a superhero (Iron Man and Green Lantern are fringe cases since their position involve physical changes to their bodies). Of course, ultimately this is the semantics of the competition. I'm not actually that concerned with anything so petty as disqualifying him or anything, but clarity of the story would aid it.
*  This looks like it fits the low color count! A good look gets me a seven count: Black, Orange, Dark Orange, Yellow, Dark Yellow, Green, White. Unfortunately, there's a mess of duplicates in there (racking the count to 40), but since they're not visible they have no impact o the piece, so no problem.

Now, like I said, the biggest trouble is the banding and jagged areas. When I was on Jiinchu's piece, I talked about how Banding was often the cause of a misplaced attempt at transitioning between two colors, a job much better performed by smart curves and anti-aliasing. This piece is an excellent example of the OTHER big cause of banding, where one is attempting to follow a lightsource and illuminate regions tracing along an outline. In general, this isn't even a misplaced mindset, just an error in the exact execution.
The main technique to fix it is pretty easy - simply stagger the actual clusters of pixels so they don't line up directly with the others, thus preventing the banding issue while still keeping the light hitting that surface. It can take a little practice to get used to staggering, especially since our minds are programmed to square things up and make 'em neat. It gets pretty second-nature after a while though.

The other side of the banding coin is the jagged lines, such as the right side of the larger figure in panel 7 or the left side of the hand in panel 3. These are caused mostly by hasty pacing and attempting to get either structures with sharp corners, non-perfect curves, or straight edges at non-standard angles. The solution is the same as the misplaced banding explored with Jiinchu's piece, smart curves and anti-aliasing. Take your time, and zoom out frequently. The key to curves is to get a natural-looking progression of lengths, either increasing or decreasing, while the key to lines is to keep them regular and to keep each of the lengths as close to each other as possible (this sort of thing I might talk about further in the future).

That's the only TRUE problem in the piece as it stands, though there are other minor ones. With that alone, I think that fixing the piece's flaws is certainly possible. Beyond that though, I think there's some interesting potential in the piece that would remain unexplored simple by repairing the broken lines and banding. Here's what I have from my own editing:
At the most basic level, I've cleaned a lot of the lines and repaired the banding. Most of my work was focused on Panels 3, 6, 7, and 8. My greatest contribution was that I made the lighting a little stronger - you get some shape on the figures now rather than just some edge outlines. I've removed all the duplicate colors, as well as the white (which was only in the eye of the gunman and the timestamp). On the finer point, I've refined the stickmen a little, though not much - they're very expressive already.
The hand is a bit of a pride point for me! I don't think that the cloth or rock ever made a huge deal of sense, but I've done what I can to make 'em more interesting to look at.

The big important point when running with low colors is to keep up the contrast. Heavily saturated colors are good for that, since they're bold and need less contrast to stand apart from each other. Saturated colors DO need a little more contrast to differentiate themselves from their own shadows - saturation makes hues contrast each other, but values appear more similar. That's my experience at least.

The last point is about telling a story with your work. Now, I'm of a belief that text is HARD to integrate firmly into an image's story - words and pictures don't get treated the same. If at all possible to just show rather than tell, do it. This piece is a pretty good example of both ways! In the original, I think the text detracts from it. I think there were ways to better communicate "nuclear material" but the motions of the stickmen are really good. If you're gonna add text, be careful - the text at the very least should be written in a stylistically-similar way, not just in its pixels but also in diction. For example, the abbreviations here ("LOL", etc) aren't a very serious sounding conversation, but the drama of the scene is a pretty serious type of thing, with nuclear material and the bold sunset and the death and all that. The dissonance in tone kinda breaks the mood. Now, as I said in the post, there are great reasons why textual issues should be ignored, or at least discounted - in this case, I have strong suspicions that Skypehopert isn't a native English speaker. I might be wrong and he's just typin' sloppy-like, but benefit of the doubt and all that.
Telling a story is hard. Sometimes in pixel art I get lost, lost in all the technical details and single objects and characters. Being zoomed in so close and not always having the full image in mind can make it hard to keep the whole thing in mind, and it can be hard. I confess that I don't feel as careful and deliberate with my art as I do with my RPG design - when it comes to design, every choice I make is very thought-through, including what making that choice says about the principles and theme of my design. With art, I often just fall to what I think looks good.
Keep the story in mind. It's powerful, and if the challenge is to incorporate the story then don't ignore a well-told narrative just because it's a visual challenge. They're linked, and the weakness or strength of one uplifts the other.

With that point, I think I'm done for the day. Sorry for taking so long this week, and sorry for a light-on-the-images day. Hope you liked the edit, and thanks to Skypehopert for being a great test subject.

End Recording,

No comments :

Post a Comment