Dropping a link to this new floodfill/selection option in GIMP 2.10+, Diagonal Neighbors , because the use for pixel art is obvious (filling based on 8way adjacency testing makes using thin features in pixel art so much easier, just click once to change the color of your entire outline, for example.).

I haven't seen it in any other program so far. It's the kind of thing you either really want or really don't want, according to the situation.

(for example of "don't want", imagine you have a generic rounded rectangle or ellipse outline -- you want to fill the inside area but the diagonal adjacency means the fill escapes and fills everything outside too. Solution appears to be filling BG first, so it's already a different color from the inside pixels.)

3/5 of those 'custom wrapping example' images are broken, AFAICS.

Different subject:

There's this concept I've been kicking around that I've never seen explicitly discussed: rate-of-value-change (as a distinct thing from discussing eg 'difference in value indicates difference in depth'. Mathematically, the first derivative of value). "Tension" seems like the most correctly evocative word to me, for a number of reasons:

* rate of value change can be directly mapped to curvature through space ( a straight line through space has relatively constant value, as shown in planes, discs, and lines along the primary axis of a cylinder)

* value change draws the eye (compositional tension) and this can be mapped in reverse from a lineart (areas of greatest angular contrast are the best ones to darken when weighting)

* Actually shading curved sections, the level of mechanical tension in the hand required seems to map exactly to the rate of curvature in 3d space.

That's the idea. I've read a fair bit on art but don't remember any mention of this. Is there something out there that does examine this subject?

Your avatar contrast seems fine to me (though of course I don't know exactly how dramatic it was supposed to be)

That's all general art stuff (light behaviour, color theory, reducing complex blends to a well defined stack of 2-4 tones, working from super general thumbnail to specifics). Arne's art tutorial gives a good sketch of the area .

Opening a ordinary paint program, grabbing a huge, super fuzzy brush and just throwing together hundreds of 'color compos' (fill canvas with primary color; pick secondary color, paint over part of the compo. pick another color, paint over a smaller area than that, and so on.. 3-5 colors) is a good way to get into it. You can also use that method as a quick (2-3 minutes, skip to next one if you haven't finished in time) study of how the artists you like use colors -- not to eyeball all the colors in the palette, but to reduce them to a small, thoughtfully proportioned representative set that 'feels the same', that's the point.

That's my opinion: there's no substitute for a large volume of narrowly-focused work, in developing a feel for things, and so thinking about 'exactly how saturated' isn't that helpful. Being able to look at something and reproduce the basic color relationships quickly and faithfully, is the essential skill, and IMO formal analysis is functionally antagonistic to that. A sense of the meaning of more saturation, less saturation, this color, that color, can arise naturally from that volume of work. Formal things like color theory provide just a few basic landmarks to begin exploration from.

(incidentally, the numbers in most color systems don't have a very consistent meaning, so attempting to reason about them can result in silliness -- eg. RGB #808080 is not a 50% mix between #000000 and #ffffff; a true 50% mix between those two colors is #bababa. Lch, HCY, HSI, and Lab are the closest to reliable, but they are not very commonly available.)

After looking through your gallery.. you have an unusual combination of strong cinematic aesthetic and a very calculated distribution of fine details (which seems to operate on a similar principle to dithering -- distributing a 'quantity of fine detail' into an area and then blending the amount of detail through the picture to integrate it with the composition).

In theory those go together but I haven't actually seen that combination much.  The way you have the gauges oily and smooth vs the grass sharp (but a gradient of sharpness, not just binary 'sharp') is a good example. I guess achieving this level of differentiation really requires a relatively higher resolution.

I don't get the brown blob above the mana gauge though.. is it supposed to be bread? Pretty ambiguous.

Yeah, the distinction between medium and genre is too fuzzy. A medium isn't *exactly* an aesthetic -- which is the main part of what genre implies imo --, but separating out 'pixel art as a medium' from 'pixel art as a genre' is just too hard to do.

After some thought..

@Gecimen: Insulting is a subset of offending -- you could probably call it a shorthand form of offense. Given that insults can contain truth that is usable to others (eg 'cheating bastard'), and that free speech is about protecting citizens from government censorship (not private conflict), you might find that the assertion that insults don't fall within free speech has some quite unpalatable implications.

(obviously there are other legal concepts like libel and slander that come into play when you publicly utter something that could result in loss of reputation. As far as I know, though, it's incorrect to infer that because something is or may be libel or slander, that it isn't free speech.)

AFAICS this reinforces the point that 'free speech' is not really that well suited a concept to use in a discussion of appropriate moderation of a privately run internet forum. It probably has *some* applicability, but what applicability is not at all clear.

I think it's "How should we define the definition of pixel art?"

(ie. what properties should the definition itself have, as opposed to "how should we define pixel art?", which is about what content the definition should have.)

Well, personally I would not put photomanip (or postprocessing/SFX work for films, which you could maybe call 'videomanip') within digital art.. Probably because it depends on a capture device (camera) as well as drawing device + computer.

Agree that there is a distinction digital drawing vs digital painting, although this distinction is fuzzier than analog (traditional) drawing vs analog painting. If you aim to ultimately learn both, then IMO it is useful to mostly dispense with the distinction once you reach a decent level of mastery of each.

The software tools used by these different digital media don't define the media

software doesn't define the media -- with a few exceptions like Oekaki.

And for most cases, software tools don't define the media. But it seems pretty clear to me that pixel art is often, even typically, defined in terms of software tools, or at least the absence of many other software tools. Maybe because we don't have any better way to define it.

When I say 'differentiate from digital art' I mean in the sense of, if we have a robust definition of  'pixel art', you should be able to take a random selection of images and pronounce definitively either 'this IS pixel art' or 'this is digital art but NOT pixel art' for 95% of them. That's the proof of a robust definition. Pixelation and PJ don't care that much about whether pixel art is inside or outside of the category of digital art, but they value the ability to detect irrelevant content.

(this is like, say, differentiating 'speed metal' music genre from just 'metal'. All of the former category is within the latter category, but not all of the latter category is within the former category.)

Considering your earlier post again:

General art topics aren't separate from pixel art, they're what we use pixel art to tackle. Digital painting communities and marker art communities operate the same way, and I'm sure other communities operate similarly. It's about creating effective work in the community's chosen medium, and that includes general concepts, not just ones specific to the medium.

This seems to be a similar misunderstanding. I agree with what you said in the above quote, just about every word. So.. I'm definitely not saying that people SHOULD focus more on "pixel art " specific techniques; Rather, that having a "pixel art" forum means that people *will* tend to overemphasize processes and techniques that are specific to pixel art. Because that's logically exactly what makes it a pixel art forum, rather than just an art forum.

@eishiya:

I'm unsatisfied by the proposition of defining pixel art simply as a medium. I think it fails to differentiate pixel art from digital art. One argument is that if I hand you 10 tools and you only use 3 of them, you have defined down the medium I expected you to work in to a sub-medium. Well, perhaps you are therefore OK with the statement 'pixel art is a submedium within digital art'. Personally I don't think that quite tracks with how either PJ or Pixelation have treated 'pixel art'.

It's 12 midnight, so I'll leave any comments about 'genre' for tomorrow and hope that this comment is somewhat sensible as is.

I think that 'pixel art techniques' that are specifically PIXEL ART techniques are pretty rarified (for example, dither, AA, and clustering could generally be considered as halftoning, smudging, and the study of shape-stacking. You have to go into.. 'dither-walk', 'AA ramp distortion', 'subpixeling', 'specific ways to render small ellipses or circle', 'clean pixel art speculars'.. before you're actually firmly in pixel art's domain.).

That created a kind of uneasy relationship with general art discussion, IMO -- so people could and would do it but it was less clearly within the scope of the forum. It was less valued (and of course it's easier too, to criticise narrow technical points rather than the fundamental structure of a picture)

That's an ongoing problem for Pixelation IMO, and it's not obvious whether it's even possible to solve it. It looks to me like it mostly overlaps with the ongoing debate "What *exactly* IS pixel art, anyway?"

EDIT: If you compare pixel art to other forms of art, for example concept art or illustration, it highlights that pixel art is defined technically -- even if we can't entirely agree on that definition -- whereas other forms are defined functionally (a picture falls into 'concept art' insofar as it is laser focused on 'conveying a concept' rather than being technically impressive or especially beautiful). I think that's the key factor that creates this uncomfortable relationship with technique. It would be interesting to compare with the communities of other technically-defined artforms.

Insulting and offending others is really the main thing that IS within the definition of free speech, AFAICS. You can't say anything of any real interest without potentially offending someone. We don't need to legally protect inoffensive speech because, who's going to object to inoffensive speech? Nobody.

TBH 'free speech' seems like a red herring though -- If there is any moderation of posts beyond 'remove obvious spam', it seems to me obvious that what you have cannot be free speech. Instead you have a subset of speech which is defined adhoc by what posts and users you choose to remove -- for example in this case, it's not simply that RAV was suddenly insulting on day; he had a history of sporadically being intensely and perhaps pointlessly insulting, as well as going on long tangents. He also at times expressed curiously definitive opinions on pixel art -- curious because when pressed for pixel art work he did, he produced nothing. These things were, as far as I know, pointed out to him, with no noticable effect. There was then a reasonably clear case that he wasn't there to participate constructively.

So the idea that RAV was participating in a community with unrestrained free speech seems, on the face of it, wrong.

I won't pretend I know what's going on, but I certainly have questions:

* What is the practical meaning of Ptoing becoming a mod again? He doesn't seem to have said anything about it on Pixelation.

* I'm not satisfied with the account that Helm has made of himself. At all. I don't accept his claims of what was going on without evidence, and the (im)precise manner in which he has phrased these claims only makes me much more suspicious of them. The attempts by Indigo to play off these events as no big deal, also really doesn't jive with *any* of the other accounts at all.

@pistachio

"think you have that backwards"

> proceeds to strongly and continuously affirm and reinforce the point that I was making, but still implying that it's a rebuttal

lol

I guess that's down to ambiguous wording ("hold back the drama") on my part.

What I meant was to enforce gradual, continual, and overall development of the picture. Vitality and efficiency are what enables that, absolutely.

Now, what you are saying about polish seems confusing and not quite on this same track. While it must be restrained and applied precisely, it seems to me that it adds most of the actual drama ("read", "pop",  contrast, etc). This shouldn't be construed as saying that gesture is not the most important thing that everything else is built around; it is. But makeup, so to speak, is what creates the dramatic effect. structure AFAICS 'makes ready for' the dramatic effect.

Well, it's hard to talk about. It's not like doing gesture studies doesn't have a great sense of drama *for me as I am doing them* -- but only another artist really appreciates a gesture drawing; for the average viewer, a skeleton is not dramatic, it must have flesh and clothes first :)

@pistachio:

"Drama is the backbone of an effective piece"

Strongly agree, but I'm not just promoting more 'blueprint' manner of working for only symbolic images (FWIW I actually think symbolic images should be even more dramatic insofar as it doesn't hurt clear identification of the elements of the image).

Rather, it seems to me that the thing to aim at is to remember what drama you are trying to present, but then hold yourself back from pulling that drama out fully until the picture is, say, 80% done. That seems to me to properly separate the concerns of design and construction from those of feel and drama, so that each can be developed with a minimum of confusion.

(Not to neglect thumbnail studies. They seem to me to be a way to efficiently arrive at a suitable dramatic vision, rather than something that directly participates in the development of the image)

...

Nicolaïdes' book includes a gesture exercise that relates to 'the illusion of movement through multiple frames'. Basically, look at someone doing a repeated action, and draw the gesture both of individual extreme 'snapshots' and of the whole thing, simultaneously (one continuous line + don't stop moving, as is the case for the base gesture exercise proposed by Nicolaïdes). Sort of consider a gesture as inherently being a time series/ animation rather than a static snapshot, so you can draw a somewhat clear 'thumbnail' of the entire animation.

Since that explanation is pretty inadequate, exact text:

The model is asked to take a moving pose. This needs an example: The model stands facing the class and turns, keeping his right leg in the same position but swinging his left leg around, twisting his torso and reaching with his left hand to the right; then he returns to the original position. He repeats this over and over again for three minutes, sometimes moving slowly, sometimes moving naturally.

Starting with that part of the body which remains more or less stationary, make a gesture drawing of at least two poses, one at the beginning and another at the extreme of the action. In the pose described above, the right leg remained more or less stationary and the extreme of the pose occurred when the left hand was reaching to the right. These two drawings are made on the same paper and are superimposed, the part of the body that was the same in both poses being drawn only once. Keep the pencil continually moving as the model moves.

Later you may use the moving-action pose to make drawings that are more detailed, but for the present stick to the spirit and the style of a quick gesture study. Try to draw three positions whenever the pose permits.

@pistachio : I think we are more or less on the same page. I especially agree with "They're not discrete but they reinforce each other in some way"

Another way of splitting it up could be analytic vs dramatic. Perhaps my comment about pixelblobs makes more sense in that context: The emphasis of an edge adds drama; the size of a flat-color blob before it transitions to another color == further emphasis of that edge. ergo, working with larger brushes or shape-cuts ~= rapidly dramatizing the picture. Working with thin lines ~= conserving drama so you can work out problems more clearly.

Vector = force? I'd prefer to categorize vector as motion, which can be considered as a superset or abstraction of force (the geometry of an object can be considered as a collection of motions through space, but an animated object also describes, as a whole, a motion through space).

I haven't had much luck finding discussion of the abstract question "How should I think about art?", personally. But Nicolaïdes' "Natural Way to Draw" is a stand out in this regard, focusing heavily on sound mental framing rather than technique.

..Although with an external tool, you can hook it up to file notification[1] so that every time you save your work, an image viewer pops up showing the result of isolation-checking it.

On a slightly different subject, it should be possible to override the output filename via parameter (eg. check pic.png output.png or check -o output.png pic.png ). This is a matter of being a proper "CLI citizen" -- and the latter form would also allow the user to make a custom launcher that uses their preference of output filename.

[1] eg. inotify on linux; there is a comparable system on Windows IIRC.

@eishiya: That (explicitly specifying a non-continuous method) explains it. Non-continuity is something I scrupulously avoid, because it seems to impair my sense of the image as a whole. I'm always aiming to draw as continuously as possible, and to complete the basic layout within 60 seconds (with the context that working this fast gives me more ability to try out alternatives before committing). That IME pushes dynamism to a level I'm kind of happy with.

So I guess you could say personal preference or individual weakness (Dynamism/feel seems very difficult to maintain a grip on, for me). I agree with your observations in general so it makes sense the particular approach you describe could work for others who are better at hanging onto that initial feel.

@pistachio:

IIRC I'd called it 'graphic vs vector'. But perhaps 'graphic vs blueprint' is more apt. I've kept on thinking about it, but not sure if anything is new. The recent discussion about working from blobs could probably be considered in those terms (as a clearly 'graphic' way of approaching design, with the relative size of your brush indicating degree of graphicness.)

Still not sure whether it maps onto symbolic vs representational. To me, it looks like more of a lean -- working graphically pushes you somewhat towards symbolic presentation, working based on vectors pushes you somewhat towards representational presentation.

Eh, blobs+large to small is my typical method of working, but I can't say it's even anywhere close to being as easy with mouse as it is with tablet.

Mainly this seems to be related to the weight of the mouse relative to tablet stylus.

* making two strokes in quick succession is difficult to do at all; it's stroke, stop, reposition, stroke.

* large strokes work (if you have the right technique - lock elbow+wrist, rotate shoulder), small strokes get polygonized (by muscle mechanics, not by the computer)

It's interesting that Henk still prefers mouse so heavily over tablet for PA. I guess the slack (dynamicness / design speed / etc) must be being taken up by preparatory sketches in traditional mediums.

Which is something that he doesn't comment on in this interview, but I think someone else has interviewed him previously..

pxltrm : a keyboard-based (vi keys!) pixel art editor for Linux/OSX terminal.

Written in pure BASH, which is possibly more hilarious than the concept itself.

Kind of touches on ascii art a little, but REXPaint is probably better if that's what you're interested in.

GIMP 2.10 released (finally)

Release notes here (long, has videos of fun things like N-Point deformation tool)

Seamless indexpainting has been in for a long time now (more or less anything you would do to a 24bit image works on an indexed image -- airbrush painting, nonbinary selection masks, painting modes and layer modes, alpha masks, filters (eg. blur or sharpen.. j/k))

Other things like HDRI, interactive previews, on-canvas IWarp and blend, symmetry painting, people are getting excited about elsewhere but are probably less relevant to PJ.

The symmetry painting does include tiling though (implemented for texture painting, of course..) .. haven't gotten a chance to try that out yet.

Downloads currently available for Windows and Linux, AFAICS.

https://github.com/0ion9/filehistory <- implementation of the below idea.

I did it before with git and `inotifywait -m -e close_write`. Inotifywait tells you when i/o events occur on particular files.

Maybe git is not appropriate, though. Keep it more basic.

Probably just sit in a `while` loop and copy the file to a $FILENAME.hist directory (eg /tmp/foo.png -> /tmp/foo.png.hist/) with an index and a md5hash for a filename (eg. /tmp/foo.png -> /tmp/foo.png.hist/00001-d8362b319cfc5588aeb1a608cc1702b0.png).

Making some kind of commit log would also be easy (prompt for a log message with zenity or dmenu, append it to a file `log.txt` like this: "00001-d8362b319cfc5588aeb1a608cc1702b0.png Initial revision".)

If you wanted branching history then I think you would have to get into git. At any rate things would unavoidably increase in complexity if you want branching history.

@pistachio:

Sort of? I'm still figuring out exactly what I mean by 'graphic'. I think that a fully rendered image is by definition graphic, no matter how realistically you choose to render it or in what medium. As you render things, you repeatedly experience the affect (spelling is correct) that you are trying to communicate to the viewer. I'm not sure if the result of that is that your conceptualization of the image becomes more symbolic (an arrangement of effects), but it may be something like that.

Content dictating the technique is actually the inverse of my main concern. My experience is as soon as you begin to engage with technique, it tends to take over and deform the content into something that is conveniently rendered with the techniques you currently possess. Therefore it seems that technique must be deferred or minimized as much as possible, if you want to create an image which is fit for a purpose rather than partly technique-porn.

In the case of a medium as coarse as pixel art, it's actually very hard to avoid diving into technique near immediately. Once you layout some blobs indicating the broad design, what is there to do other than rapidly rerender those blobs in increasingly refined ways? If you prefer beginning with a lineart, then you are also engaging with techical specifics near immediately, just different ones -- clean lines, avoidance of tangents or banding.

(rather than 'technique', maybe I should be more specific? 'medium-specific optimizations / methods'? Because the most general techniques don't seem to be as problematic in this way (eg. 'perspective' doesn't seem to be a strong attractor, but 'clean line convergence in pixel art'  does seem to be. It seems to be related to both the speed with which you can iterate through variations on a rendering, and strict constraints that mean the number of correct solutions is relatively small.)

There probably are things more 'graphic' than pixel art. Color field paintings, possibly -- they seem to be explicitly just an arrangement of effects.