jotapeh
|
|
« Reply #80 on: April 15, 2010, 12:48:00 PM » |
|
Still, this all misses the point as neither voxels nor trixels end up looking like pixel art however prettily they are drawn.
Agreed, it seems like none of the posts have nailed the concept that the OP described. I fear this technology may be out of our grasp.
|
|
|
Logged
|
|
|
|
Alex May
...is probably drunk right now.
Level 10
hen hao wan
|
|
« Reply #81 on: April 16, 2010, 12:27:29 AM » |
|
Trixels DO look like pixel-art, at least when you're looking at them head-on/along an axis without perspective. That's sort of the whole point AFAICT, it's an extension of pixel art to 3D Well they only look like pixel art because they are pixel art in that perspective, as drawn by the pixel artist. The intent of the OP was to formulate a process that generated bitmap data from e.g. 3D scenes that looked like it had been pixeled by hand.
|
|
|
Logged
|
|
|
|
nikki
|
|
« Reply #82 on: April 16, 2010, 02:52:20 AM » |
|
maybe you could generate a ['voxel' 3d file] with a much smaller resolution then the iginal 3d file. You could just average the orginal file in. After that you have a "big 3d pixel filled" object wich you could rotate and draw. And look at
|
|
|
Logged
|
|
|
|
Tycho Brahe
|
|
« Reply #83 on: April 16, 2010, 06:00:31 AM » |
|
you COULD use some sort of depth testing method, which ensured that all pixels drawn were the same size, ie; backgrounds would be rendered at a lower resolution than foregrounds.
I dunno, just an idea.
|
|
|
Logged
|
|
|
|
raigan
|
|
« Reply #84 on: May 21, 2010, 08:09:29 AM » |
|
*bump*
|
|
|
Logged
|
|
|
|
Chromanoid
|
|
« Reply #85 on: May 21, 2010, 08:28:05 AM » |
|
the footage was rotoscoped BY HAND in photoshop; nice video anway...
|
|
|
Logged
|
|
|
|
Polly
Level 6
|
|
« Reply #86 on: May 21, 2010, 12:59:42 PM » |
|
That effect can be achieved very easily. Simply render to a low-resolution texture, upscale using nearest-neighbor, and down-sample the colors to 16-bit. Attached are screenshots from such a setup made in ZGameEditor.
|
|
|
Logged
|
|
|
|
Tycho Brahe
|
|
« Reply #87 on: May 21, 2010, 01:08:27 PM » |
|
That effect can be achieved very easily. Simply render to a low-resolution texture, upscale using nearest-neighbor, and down-sample the colors to 16-bit. Yeah, thats sort of what I was thinking, the only thing would be the algorithm used to select the colour when rendering to a low-res texture, Ie, what average do you use (mean, median or mode)
|
|
|
Logged
|
|
|
|
increpare
Guest
|
|
« Reply #88 on: May 21, 2010, 01:16:30 PM » |
|
Yeah, thats sort of what I was thinking, the only thing would be the algorithm used to select the colour when rendering to a low-res texture, Ie, what average do you use (mean, median or mode) Mode wouldn't make sense so far as I can see - there are too many values for looking for duplicates to be useful (I guess you could use buckets and count how many in each ...). Median relies on having a linear ordering (you could do it component-wise I guess). Component-wise average would be fine I'd imagine. You have any reason to think otherwise?
|
|
|
Logged
|
|
|
|
Tycho Brahe
|
|
« Reply #89 on: May 21, 2010, 01:24:42 PM » |
|
Well, imagine "low-resing" an image with a white in one half of the screen and black on the other, with a diagonal line down the centre. You would get a very ugly grey line down the divide if you used the mean, and either black or white if you used the mode or median.
|
|
|
Logged
|
|
|
|
increpare
Guest
|
|
« Reply #90 on: May 21, 2010, 01:30:17 PM » |
|
Well, imagine "low-resing" an image with a white in one half of the screen and black on the other, with a diagonal line down the centre. You would get a very ugly grey line down the divide if you used the mean, and either black or white if you used the mode or median.
How would you define the median in this case? (fair point about mode potentially being useful for stuff with low number of colours).
|
|
|
Logged
|
|
|
|
Tycho Brahe
|
|
« Reply #91 on: May 21, 2010, 02:15:29 PM » |
|
Well, imagine "low-resing" an image with a white in one half of the screen and black on the other, with a diagonal line down the centre. You would get a very ugly grey line down the divide if you used the mean, and either black or white if you used the mode or median.
How would you define the median in this case? (fair point about mode potentially being useful for stuff with low number of colours). the median things a good point. Possibly the individual medians if each of the red, green and blue values of the pixel?
|
|
|
Logged
|
|
|
|
Tycho Brahe
|
|
« Reply #92 on: May 23, 2010, 07:41:50 AM » |
|
Sorry for double post but I felt this deserved more than an edit: I mocked up a simple program to compare the mean, median and mode and I think mean wins out (tested on this image: http://en.wikipedia.org/wiki/File:Apollo_17_Cernan_on_moon.jpg) How does that look to people, is that a acceptable pixel art effect?
|
|
|
Logged
|
|
|
|
increpare
Guest
|
|
« Reply #93 on: May 23, 2010, 07:52:00 AM » |
|
Sorry for double post but I felt this deserved more than an edit:
I mocked up a simple program to compare the mean, median and mode and I think mean wins out
can we see comparison pics? How does that look to people, is that a acceptable pixel art effect? No, it just looks like it's been resized - pixel art isn't just low-res (see earlier on in the thread - about dithering in particular) - it encompasses a quite wide variety of compositional and textural effects/techniques. Also, a main concern in this thread is having something that is in 3d and looks pixelly, while preserving its aesthetic during rotation.
|
|
« Last Edit: May 23, 2010, 07:56:22 AM by increpare »
|
Logged
|
|
|
|
Tycho Brahe
|
|
« Reply #94 on: May 23, 2010, 07:58:10 AM » |
|
okay then, the main difference with this and most resizing/lo-rezing algorithms is that it takes into account the average colour of a rectangle or square instead of just the four corners, which most algorithms (eg its the default in opengl and in processing) do.
and I'll put some comparison pictures up when I have time
|
|
|
Logged
|
|
|
|
gimymblert
|
|
« Reply #95 on: May 23, 2010, 08:53:27 PM » |
|
We need something that process high frequency and low frequency separately
|
|
|
Logged
|
|
|
|
Lon
|
|
« Reply #96 on: May 24, 2010, 12:58:06 AM » |
|
How does that look to people, is that a acceptable pixel art effect?
Interesting, but I am afraid increpare is correct, it looks like a resized low-res image (try resizing it small to kinda see the inverse effect). I think an important part of pixel art (though I do not have much experience making pixel art) is using a limited pallet and having more contrast between different objects.
|
|
|
Logged
|
“We all sorely complain of the shortness of time, and yet have much more than we know what to do with. Our lives are either spent in doing nothing at all, or in doing nothing to the purpose, or in doing nothing that we ought to do..." -Seneca
|
|
|
Alex May
...is probably drunk right now.
Level 10
hen hao wan
|
|
« Reply #97 on: May 24, 2010, 01:12:01 AM » |
|
Using image filters is probably a bit of a red herring. I think something like this would need some AI.
But I still think it's impossible.
|
|
|
Logged
|
|
|
|
Mikademus
|
|
« Reply #98 on: May 24, 2010, 01:56:56 AM » |
|
How does that look to people, is that a acceptable pixel art effect?
2D art is an artistic style, not mere pixelation.
|
|
|
Logged
|
\\\"There\\\'s a tendency among the press to attribute the creation of a game to a single person,\\\" says Warren Spector, creator of Thief and Deus Ex. --IGN<br />My compilation of game engines for indies
|
|
|
nikki
|
|
« Reply #99 on: May 24, 2010, 06:47:18 AM » |
|
to me it looks like a simple resized image too (wich it is) the original image is a simple 2d bitmap too, so it would be extra complex to magically analyze that image, whereas a 3d image would give out more metadata..
then again, when a simple pallet is applied over this image, i'm sure you can fool some people in believing that its pixelart...
|
|
|
Logged
|
|
|
|
|