Ramblethread! A brainstorm approaches!

[PixelPiledriver]

Ah I see what you mean.
Building solid shapes from the line tool instead lines or outlines to fill.
I'll have to put something together for this and try it.

Vector art process has a similar sense of "start clean, get messy".
And also supports pre and post manipulation.

It'd be interesting to have a line length limit option to strictly control the size of clusters you could build.

Previews are a wonderful thing but are obviously restricted to math based tools or change.
But I kind of wonder, there could be other places that they might be useful.

[Mathias]

In Photoshop, when using a 1px brush, and I want straight lines I simply hold SHIFT and Left-Click - that draws a line from point to point. Pretty quick.

But you're suggesting actually using a brush that's in the shape of a line? What do you do when you need it to be a different angle? Which would be often. I like a 2x2 or 3x3 when slapping rough forms. Then a 1px clean-up.

[Mr. Fahrenheit]

He means the line tool which when you click and hold, or just click the first time then drag the mouse lets you see the line you are about to lay down.

[ptoing]

Not in Photoshop though. The linetool in PS does not show anything that is really representative of the final line before it is not set down (aka, bullshit).

[PixelPiledriver]

Yah it's odd that it only previews the vector that will lay across the pixels as you move the mouse.
You can see the pixels that it will affect and how it will AA them if you turn on a per pixel grid.



But it's not exactly ideal.

[ptoing]

Eh, that is pretty weird, with the PP grid. Then again, I would never wanna pixel in PS, the only thing that I would ever use it for as far as freelance pixelart goes is perhaps adjustments and other such things which I can not easily do in PM.

[Helm]

Single pixels that work

I know we all like single pixel speculars here and there. Aside from AA it's the other single pixel that's difficult to give up. I am going to present a single pixel solution for highlights that works within cluster theory, and it stands as an example of other similar solutions.

What's the problem with single pixels again? They lack direction, they look like noise. So they reduce motion to a piece and they reduce cohesion of the clusters working together.

However, if we can make a specular one-pixel highlight part of a metacluster, then the metacluster can give the single pixel motion, and it is by virtue of being a meta-cluster, more cohesive than just a single pixel on its own.

Consider a 2x2 pixel square cluster. Now consider another. Overlay them on their corner so they both share a single pixel. make that pixel extra bright, as if the two 2x2 blocks were in an overlay blend mode and they're adding up.




This single pixel has more motion (45 degree brush stroke, in this case) and cohesion (in a pyramid of brightness with the 2x2 squares to support it in the middle) than if it were just standing there just by itself. If the bright pixel has a value and hue relevant (not necessarily resultant on both fronts, it could be antithetical if you want) to the 2x2 squares underneath it, that's an extra layer of pleasing complexity.

The same premise works for a three pixel line with the middle pixel being a brighter single, or any other meta-cluster. Give it a thought.

[RAV]

yeah. I consider this a pixel artist's manual level of detail.
As a general notion I'd call it Embedded Detail.
It's a form of cluster based "anti-aliasing", that is anti-nonClustering.
Especially in works that would suffer from resizes, this approach stabilizes visual quality.

[Cyangmou]

I know that it all started out as an idea / new approach and not by the intent to create a new rule.  If I recall right you even wrote that you, Helm, don't intended to do so.
The results are observable in the net, actually I don't care anyways. I just think it's highly interesting to observe the behaviour and effects something like this can have.


One crazy idea I wanted to share/discuss for quite some time now is "contrasting".

What you need to partake / make up your mind is a perfectly calibrated monitor - If you haven't a high quality monitor, which is pretty decent at giving back values your perception will be quite off.
Although you can calibrate your monitor with this tool (which already can greatly enhance perception): http://www.lagom.nl/lcd-test/

I also thought I'd reduce the concept down to the simplest example I could think of: 

-With contrasting I just mean how small/big the gaps we use between "areas of the same color".
As we all know color is quite a topic on it's own, so I am just focusing on value for this example
(which is basically one of the 3 dimensions of the hsv - color space - (hue, saturation, value) but also has the biggest effect).

I also decided to do my example against the most neutral color at our hands - 50% gray.

I made a chart which contains:
lines: squares at brightness levels starting at 50 and going up to 100 (pure white)
columns: outlines with a value level starting at 45 and going down to 0 (true black)

I think that there are areas where the outline looks "washed out"
I think there are areas where the outline starts to "burn" (the contrast hurts your eyes)
I think that there is an area where the outline just looks alright.

For this simple example I want to pin down the area where background, outline and the square look as a whole nice to your eye.
SO all you have to do is actually look at the chart and search the points where it starts to look washed out and where it starts to burn.

Beware that the human eye adjusts it's color with time. SO you should have looked at something else before you do this and try to not spend longer than about 5 minutes with looking at it, otherwise your perception changes, because of the architecture of the human eye.

[Helm]

Very interesting. Starting from brightest square values, my eye seems to want 30-35-25 outline values for most entries.