Re: [Gimp-user] Web development question?

[Steve Kinney <admin pilobilus net>]

On 02/16/2012 08:59 PM, Burnie West wrote:
> > > create artifacts - or so it seems to me.
> > I did not know that jpg had an indexed mode.  I knew I would be
> > getting some interesting feedback when I posted that answer.
>
> I didn't mean to imply that jpg has an indexed mode. It just seems
> to me if
> one exports a GIMP image in indexed mode to jpg the file comes out
> smaller.
> Haven't run extensive tests on that, though.
>
> GIMP actually transforms back to RGB before going to jpg (asking
> if that's
> what you want first). So YMMV.

That makes sense.  Blast this "e mail" stuff, so easy to
misinterpret things that made perfect sense when they were typed.

I have done a lot of tests and I am confident that indexed images
consistently make larger files than RBG, no matter how thoroughly
tweaked they are.  There are some cases where this effect is
reduced:  Images that have a very limited number of colors in large
contiguous areas seem to be the least affected.

In re the jpg vs. png thing referenced in other posts, the alpha
channel in png still seems to me the only real advantage the format
has for web applications, because png files are consistently larger
than jpg files with the same content and resolution.  Inkscape
exports to png because it is lossless and has an alpha channel, both
of which are important when importing the image to a raster editor
for final adjustments.  These features also make png the most
"universal" and error resistant web graphic format, and for that
reason the right format to prescribe in high level web design course
work. 

But there are still lots of cases where jpg is the right choice for
most web images:  Like sites that will be accessed by mobile
devices, or in countries where dialup is still common, or on
commercial sites that serve millions of page views a day.  These are
places where image file sizes can substantially impact the user
experience and/or server load.

I am looking forward to the day when all significant browsers
support the major features of the proposed html 5 standard.  Direct
support for vector graphics will be a real revolution, enabling
substantial expansion of CSS to eliminate many common (and presently
useful) decorative layout hacks, and reducing many common kinds of
graphics to a vanishingly small fraction of their present on-disc size.

:o)

Steve