Forum Moderators: not2easy
So it there a waypoint in filesize of images to not being concerned about reducing its size anymore?
It had something to do with the minimum packet size send over the net? So that made a bit of sense as if a packet size is always minimum 8kb reducing the image to 2kb will still send an 8kb packet. This is just my logical thinking not reflecting the reality i would like to verify with you.
thanks,
Bastiaan
It really depends on how many colours the images have, and the complexity of the image. I certainly don't think there is any kind of "filesize limit" that should be implied.
Getting the smallest image size possible without sacrificing quality can be an important issue for higher traffic sites...
//added
I don't think this "minimum 8KB packet" statement is true but someone can correct me if so. AFAIK that's the maximum packet size
[edited by: brotherhood_of_LAN at 11:47 am (utc) on Feb. 17, 2009]
i'm not sure that 8KB is the correct size for the packet either -- i read that it's between 576 and 1500 bytes.
Now, take that difference and multiply it by however many images fit this criteria and the numbers add up very quickly. You can severely hamper the visitor experience on various devices by using larger size images when they can easily be converted and made a fraction of their size.
I come from the camp that every byte should be optimized to the nth degree without any loss in quality. Yes, we are a high speed Internet for the most part but according to statistics, there are still some large numbers in the "less than satisfactory" speed category. Those could be dialup, mobile, etc.
I also have a rocket-sled connection so have to be careful about being minimally considerate to those who don't.
I come from the camp that every byte should be optimized to the nth degree without any loss in quality.
I've left this camp entirely and am not shy about much larger than average pages. The dialup user is mostly not a factor for our considerations.
However, the point about mobile should be taken very seriously. The jury is still out (for me) on how quickly as well as how important this will wind up being; a lot can change in a hurry. Tomorrow's killer app may not make it out of the gate today. However, I have been investing a lot of time in mobile optimization. Mostly by serving up vastly different stylesheets and serving minimum images, not offering the high-res stuff that I otherwise make available - a very heavy hand at stripping the sites down. Sometimes it's easier to just do an entirely different website and keep it super-streamlined for updating, with custom navigation to make moving around fast and easy.
Like pageoneresults, if a page takes time to load on my connection they have got some serious problems; images, JavaScripts for everything under the sun, who knows what all might be going on.
<snip>
... shows a nice graph and if I am reading it correctly the 1500 bytes packet size are used the most or somewhere between 1400 and 1500 bytes.
This is confirmed by
<snip>
that stated that 37% of all traffic is 1500 bytes
so imagine the following t scenarios:
A) I have 3 uncompressed gifs of 1500 bytes (= 3 packets)
B) if I optimize them 50% they will be 750 bytes per image (= 2 packets)
which one do I receive quicker? I guess there is an processing overhead in squeezing the 3 images into one packet of 1500 bytes in option B while in option A the packets are already send by then. also the second packet have space left and will potentially be used for something else too. Again calculation overhead.
I would be interested if this is true.
No URI's please, see TOS [webmasterworld.com]
[edited by: limbo at 4:09 pm (utc) on April 14, 2009]
Yes, we are a high speed Internet for the most part but according to statistics, there are still some large numbers in the "less than satisfactory" speed category. Those could be dialup, mobile, etc
i'm with pageoneresults, even people with fast internet connections can find that their connection is often not working at maximum speed for all kinds of reasons beyond the user's control. [this certainly is the case in the uk and my personal experience, i'm sure it's true elsewhere too]