Help me with this experiment

[manuelarean]

I have read a lot about the best file format for import images into flash, so I tried the next experiment:

I use the "famous" Dune.tif image of the samples from photoshop, I changd the image size from 600 x 600 px to 300 x 300, and save It as .jpg and .png file formats (the same image) and I get these file sizes

Dune.jpg (80% quality), size 48 Kb aprox
Dune.png (default settings) 172 Kb aprox

Then I imported these images to flash in separates documents, converted the bitmaps to graphic symbols and made a motion twen of 50 frames from alpha 0 to alpha 100. I saved both fla and I get these file sizes

exppng.fla = 220 Kb
expjpg. fla = 44 Kb

Up to here all looks "logic" bigger input = bigger output

But here is the interesting detail, the swf file sizes

expjpg.swf = 32 Kb
exppng.swf = 8 Kb 8 Kb

So I would like if some of you in your spare time, try some variations of this experiment, and post your resoults here, to get a more definitive conclusion...

Becouse so far I have found a reason to use always PNG to import to flash, and is a reason based on one of the most importants facts in WEB, file size. But before that I would like to probe that this is a "general rule" and not and "unique event"

(sorry about my english)

[fgf]

My results
445k bitmap (photo but with quite a lot of green)
->
166k png
17.6k jpg

png swf 16.5k
jpg swf 17.5k

png settings 24bit png
jpeg settings quality 30%

This had started from a jpeg image so a visual comparison for artefacts etc doesn't make sense and might also have some effect on the results. I'll grub around for some scaned images to test with a bitmap that has never been a jpeg

fgf

[fgf]

Started with a fresh scanned image

final results png.swf 23k, jpg.swf 27k
24bit png versus 30% quality jpeg.

No pig artifacts on the jpeg but it appeared slightly blury around some text in the photo versus the png looking slightly pixelated around the text but preserving the colour range better in dark areas

fgf