Creating JibJab's in AS
Moderators: Víctor Paredes, Belgarath, slowtiger
The gist of it is, 24-bit colour, what we usually use for our monitor displays, is about 16.7 million colours, and those colours are made from 3 channels, red, green, and blue, each of which have 256 levels of intensity. 256x256x256 = 16.7 million. When you add transparency, that's another 256 levels of intensity in a fourth channel, and that makes 32-bit colour. GIF, on the other hand, uses 256 colours total, in only 1 channel.
Oops, Touched beat me to it. Oh well, since I'd already typed it, here's my version anyway.
A slight clarification - it has more to do with file format than display technology.
PNG (with alpha/transparency) uses a 32-bit file format.
32-bits (4 bytes) are used to represent each pixel.
One byte (8 bits) each for red, green, blue, and alpha/transparency.
One byte (8 bits) can only contain the values from 0 to 255, or 256 values/levels.
(0000 0000 to 1111 1111 in binary format, 00 to FF in hexadecimal.)
(Because a pure gray is made of equal levels of red green and blue, you can also only have 256 pure grays - counting pure white and pure black on each end of the grayscale spectrum/gradient.)
If you could use a 64-bit file format (16 bits per channel), you could have 65536 levels of transparency (256x256). There are a few programs that use 64-bit file formats, although they are rare, and some programs even use floating point values instead of an 8-bit byte.
(Oh, expanding a little on what Touched said, GIFs use a colour table - the index to the table is only 8-bits, so you can only have 256 colours in total, but they can be any 256 of the 16.7 million colours).
Regards, Myles.
A slight clarification - it has more to do with file format than display technology.
PNG (with alpha/transparency) uses a 32-bit file format.
32-bits (4 bytes) are used to represent each pixel.
One byte (8 bits) each for red, green, blue, and alpha/transparency.
One byte (8 bits) can only contain the values from 0 to 255, or 256 values/levels.
(0000 0000 to 1111 1111 in binary format, 00 to FF in hexadecimal.)
(Because a pure gray is made of equal levels of red green and blue, you can also only have 256 pure grays - counting pure white and pure black on each end of the grayscale spectrum/gradient.)
If you could use a 64-bit file format (16 bits per channel), you could have 65536 levels of transparency (256x256). There are a few programs that use 64-bit file formats, although they are rare, and some programs even use floating point values instead of an 8-bit byte.
(Oh, expanding a little on what Touched said, GIFs use a colour table - the index to the table is only 8-bits, so you can only have 256 colours in total, but they can be any 256 of the 16.7 million colours).
Regards, Myles.
"Quote me as saying I was mis-quoted."
-- Groucho Marx
-- Groucho Marx
I have to correct that info on GIF a bit. GIF can have 256 colors per frame, but it is possible to have more than 256 colors in a GIF image, if you use local color pallets, as in the following GIF image, which has 763 different colors (black to pure red in 255 steps, black to pure green in 255 steps, black to pure blue in 255 steps, and black).

In fact, it is an animated GIF with local pallets and transition "previous". If you want to get all fancy, you can even have a good looking color photo in GIF. Of course, you'll need to separate the colors into 255 per frame and create different frames. You'll need 1 color value out of the 256 possible colors to be all transparent, to let the other colors in the other frames shine through.
Of course, this is of no use for AS, because it cannot import GIF images, only BMP, Targa, JPEG and PNG.

In fact, it is an animated GIF with local pallets and transition "previous". If you want to get all fancy, you can even have a good looking color photo in GIF. Of course, you'll need to separate the colors into 255 per frame and create different frames. You'll need 1 color value out of the 256 possible colors to be all transparent, to let the other colors in the other frames shine through.
Of course, this is of no use for AS, because it cannot import GIF images, only BMP, Targa, JPEG and PNG.
Holy cow!
Talk about thinking outside the box. I never thought of using animated gifs for non-animated images. Amazing.
I wonder if there would be a way to use that technique to create "softer" alpha transparency for a gif? For instance, off set the transparent pixels on an "ordinary" simple gif. Each frame has a different spot for the transparent pixels... that... uh...
My head hurts. I will get back to you on this.
-vern
Talk about thinking outside the box. I never thought of using animated gifs for non-animated images. Amazing.
I wonder if there would be a way to use that technique to create "softer" alpha transparency for a gif? For instance, off set the transparent pixels on an "ordinary" simple gif. Each frame has a different spot for the transparent pixels... that... uh...
My head hurts. I will get back to you on this.

-vern
- Víctor Paredes
- Site Admin
- Posts: 5815
- Joined: Wed Jan 26, 2005 12:18 am
- Location: Barcelona/Chile
- Contact:
for my block www.tazatriste.cl/?bubu this would be great too.
the page is about moving things (as images for example)
unfortunately, since four or more messages ago i haven't understand anything.
edit: oye, heyvern, the last awful IE supports pngs with alpha.
the page is about moving things (as images for example)
unfortunately, since four or more messages ago i haven't understand anything.
edit: oye, heyvern, the last awful IE supports pngs with alpha.
That would be impractical, because GIF has no alpha values, only transparent or non-transparent.heyvern wrote:I wonder if there would be a way to use that technique to create "softer" alpha transparency for a gif? For instance, off set the transparent pixels on an "ordinary" simple gif. Each frame has a different spot for the transparent pixels... that... uh...
There is a workaround, but that would cost an enormous amount of computing power (cycle the colors very fast, so they become kind of translucent). This was a similar trick used on the Commodore 64 to create more than the 16 possible colors in multicolor hires mode, by rapidly swapping colors on screen. Of course, this created a flickering of the screen, and using this technique in GIF animation would probably do the same.
Another solution could be to not use transparency, but rather use the background color of the website, which gives the same effect as transparency.
See these JPEG images:


Here is a workaround to get PNG alpha transparency to work in IE5.5+/Win:
http://www.alistapart.com/stories/pngopacity/
I can't test this solution, because I don't have Windows.
http://www.alistapart.com/stories/pngopacity/
I can't test this solution, because I don't have Windows.
More than 8 bit (256) gets real expensive real fast. I am working on a movie where Warner Bros is requiring 10 bits per channel images for the compositing and the files are HUGE, even at the same pixel size as the 8 bit counterpart.
Because everything works in multiples, the next step up from 256 is a pretty big jump.
Because everything works in multiples, the next step up from 256 is a pretty big jump.
----
Terrence Walker
Studio ArtFX
LEARN HOW TO Make YOur Own Animated Film!
Get Video Training to Show You How!
Terrence Walker
Studio ArtFX
LEARN HOW TO Make YOur Own Animated Film!
Get Video Training to Show You How!
Artfx,
Does this have to do with the "EXR" format? That is becoming almost... nearly... a standard now. Animation:Master recently got EXR support for compositing and rendering. Huge files but the image quality is unreal. It is scary how much dynamic range is in there.
A bit soon to think about, (and I don't don't need it yet
), but at some point maybe AS should have this feature. Especially with HD getting so hot.
-vern
Does this have to do with the "EXR" format? That is becoming almost... nearly... a standard now. Animation:Master recently got EXR support for compositing and rendering. Huge files but the image quality is unreal. It is scary how much dynamic range is in there.
A bit soon to think about, (and I don't don't need it yet

-vern
-
- Posts: 246
- Joined: Thu Mar 10, 2005 7:29 pm
- Location: USA!
When I'm digitizing in video from a BetaSP I've used 8-bit and 10-bit "uncompressed" settings and the size difference between them is considerable. For most of the stuff I do (almost never any compositing) the 8-bit is more than good enough since I don't really need the extra info in the files.artfx wrote:More than 8 bit (256) gets real expensive real fast. I am working on a movie where Warner Bros is requiring 10 bits per channel images for the compositing and the files are HUGE, even at the same pixel size as the 8 bit counterpart.