this has been started in another thread but i thought it may be good to put it into an own one:
gifs with transparent area act faster if you save them without transparent area,then import them to flash,break them apart,cut off the "to be transparent" area and then reduce the vector corner points the shape is made up of as much as possible
here´s the sample file: http://www.stimunation.com/majortom/flas/vektor.zip
(the sample file was on a german flash cd)
you can click on the icons on the topright to see the vector/transparent gif comparison and on the left there´s the millisecond timer,the version with vector method is about 100% faster.
raster graphics don´t look as crisp as before once you break them apart,so this is probably best used for small graphics or moving objects (like particles,rain,bullets etc.).
That is really cool and kind of unexpected. I'm trying to figure out why that would be faster so I'm going to give my theory, if nothing else just to get it straight in my head.
I tried running that with just a unfilled circle and it actually ran slower than the broken apart image of that guys head. So here we go:
Vector images:
Vector images are a set of calculations that when calculated give the image that you have drawn. Its slower because flash has to do the calculations and then draw the pixels onto the screen.
Raster Images:
Raster image files are just stored as pixel data, so flash doesn't have to do any calculations. It just plots the pixels on the screen.
The broken apart image:
Going by the test above I did with the unfilled vector circle I'm guessing flash does not store the broken apart image as a vector. It looks like it just stores it in the same format as a raster image.
I guess since the transparent pixels in the image file are still actually pixel data, flash is still giving them as much attention as a nontransparent pixel. But breaking apart the image and cutting the transparent pixels out means flash no longer has to give any processor time to them.
It's kind of interesting, I always just thought flash would pass over the transparent pixels faster than the nontransparent ones. It's a good thing to know but I don't think I'll be doing it to all my images (mainly because I don't have the patience ) But ya like you said it's really good for bullets and other small things.
[Edit]
Ok I understand what you guys are talking about now. I had never used the 'break apart' option before since I never really saw any real use for it, but I see what it is doing now. It just creates a vector in the shape of the image and just applys a bitmap fill to it. Learn something new everyday.
Originally posted by iopred I guess if you make the shape of the vector pixel perfect, it will incur alot of slowdown tho, instead of a rough vector like they have on the demo.
I was interested to see if that would be the case so I tested it out. I took the steps tomsamson specified and then made the wand tool pixel specific. The head on the right now is my imported and modified pixel perfect one. Runs at approx 2ms slower from what I can tell. Perhaps with a much larger and detailed image it would have a big hit to performance. To that end, I think i'll test that aswell.
Originally posted by iopred Thats so wrong.. Simply because I used to do this before I realised gif's kept their transparency (hey, this was Flash 3 days).
the idea is that you save the gif without transparency and then cut away the area which should be transparent after breaking it apart inside flash.
Originally posted by iopred
I guess if you make the shape of the vector pixel perfect, it will incur alot of slowdown tho, instead of a rough vector like they have on the demo.
the speed difference between vector shape with breaken apart raster (without transparency) fill and the raster with transparency isn´t as big as in the original demo the more corner points your shape is made up of,so its wise to use as few as possible to make the speed impact bigger
The example from StenFLASH is surprising. Not much difference between two really, so the reducing corner points doesnt give as much speed as breaking the image up.
ok,tested again in my coffee break,for seing an extreme comparison,i imported the gif without transparency,broke it apart, and then used it as fill for a rectangle (->only four corner points),here´s the comparison:
high quality:
4 corner points: 24ms
exact cut out: 30 ms
gif with transparency: 64 ms
low quality:
4 corner points: 16ms
exact cut out: 19 ms
gif with transparency: 35 ms
Didn't someone post something a while back like rough speed comparison between the number of pixels, and the number of curves in a vector image?
Something along the lines of 40ish pixels is about 1 curve?
Maybe I imagined it..
I don't know if this is helpful, but when I want a smoother transparent gif or whatever I use 'Trace Bitmap' and just covert each pixel into a shape. This works well for stuff like old game graphics if you're going for a 'retro' look in your flash.
Jeez, if THAT'S the way God made you,
God's a real jerk...
Reply With Quote
) But ya like you said it's really good for bullets and other small things.
