A simple experiment

[pduggan]

Here's a simple little experiment, and hopefully I'll be able to tell you what it's about once a few folk have voted! All you have to do is to compare the A and B versions of the images and vote.<br clear="all">

http://www.petestack.com/forum/fly260a.jpgA<br clear="all">
http://www.petestack.com/forum/fly260b.jpgB<br clear="all">
http://www.petestack.com/forum/sitemap60a.pngA<br clear="all">
http://www.petestack.com/forum/sitemap60b.pngB<br clear="all">

Thanks

Peter</p>

Peat Stack or Pete's Tack?</p>

[pduggan]

Here's a simple little experiment, and hopefully I'll be able to tell you what it's about once a few folk have voted! All you have to do is to compare the A and B versions of the images and vote.<br clear="all">

http://www.petestack.com/forum/fly260a.jpgA<br clear="all">
http://www.petestack.com/forum/fly260b.jpgB<br clear="all">
http://www.petestack.com/forum/sitemap60a.pngA<br clear="all">
http://www.petestack.com/forum/sitemap60b.pngB<br clear="all">

Thanks

Peter</p>

Peat Stack or Pete's Tack?</p>

[pduggan]

Come on folks, it's an honest experiment, not a trick question! (Check the file sizes under 'properties' if you like &#8212 they're subtly different for the A and B versions but they won't help you to answer the question.) Promise I'll tell you what it's all about in a day or two...

Peter</p>

Peat Stack or Pete's Tack?</p>

[gwpriester]

Peter

Is there anyway you can place them side by side?

And what are the little circle thingies? And do I vote for them too?

Gary

Gary Priester

Moderator Person

to rub elm beehives

[pduggan]

Sorry, Gary, I'd thought of placing them side by side, but my hour of editing time for the original post has long gone! Not to worry, however, so here they are again (A versions to the left and B versions to the right):

http://www.petestack.com/forum/fly260a.jpghttp://www.petestack.com/forum/fly260b.jpg
<br clear="all">
http://www.petestack.com/forum/sitemap60a.pnghttp://www.petestack.com/forum/sitemap60b.png
<br clear="all">
As for the 'little circle thingies', they're compasses and, yes, they're part of the vote (had to try my test with both bitmaps and vector drawings)! But nuff said for now &#8212 let's keep voting for another day or so before I spill the beans...

Peter</p>

Peat Stack or Pete's Tack?</p>

[dk_art]

Gee I voted for A the first time ... now with them side-by-side B looks better. WHY ARE YOU TOYING WITH MY MIND LIKE THIS! ..... or is that part of the experiment [img]/infopop/emoticons/icon_biggrin.gif[/img]

David K

[pduggan]

Slight confession (which might make a difference to votes placed before or after 06.55 TalkGraphics time today):

Discovered I'd uploaded the two pics of Fly to my webspace with the names muddled up, so I've swapped the A and B versions at the above time (and everything's now as it was meant to be)! Sorry, not a deliberate ploy (and I must stress that the swap came after David's reply!), but I've taken note of how the votes were standing up until that point.



Peter</p>



Peat Stack or Pete's Tack?</p>

[pduggan]

At the time I discovered my mistake with the file names, 8 votes for A, 8 votes for B and 7 votes for indistinguishable.

At the the time of this post, 9 votes for A, 13 votes for B and 15 votes for indistinguishable.

The A versions were sized through the export controls but the B versions were sized first with the infobar and exported at 96dpi. As pointed out by Sean and others, it's not exactly the same thing (obvious if you open them in Xara).

But the real point of the experiment was to test the theory advanced in a discussion between Gary and bcdeb that resizing through the export process costs detail. My contention is (to quote the figures for the two Fly images) that a 1664 pixel image at 15dpi contains the same amount of detail as a 260 pixel image at 96dpi, and I think this is borne out by the images I posted and the votes cast (more for 'indistinguishable' than anything else) as well as the mathematics. I'm surprised, however, to discover that images resized through the export controls appear to contain slightly fewer bytes, which could be good news for web work. (Whereas designing an image to the correct size in the first place seems to be fundamentally good advice, I often find myself working 'back to front' &#8212 starting with the image and testing sizes and settings to see how many compromises I'm prepared to make. I've always preferred working with the export controls for this because it's so simple to keep experimenting and checking on both file size and finished quality.)

For the record, the Fly JPEGs were exported at 70% quality and the Sitemap PNGs at 16 colours. The A and B images are identical apart from the resizing method and a very few pixels' differences.

So why did people vote for A or B, and can they still see a difference in quality? (I challenge anyone to open both Fly JPEGs in Xara, resize the B version to 1664 pixels height with the infobar and tell me it's more detailed than the A version...)

Hoping for some response on this!

Peter</p>

Peat Stack or Pete's Tack?</p>

[Klaus Nordby]

Pete, it's an interesting issue - but "come sir, you picked a poor example." I.e., your original image is not sufficiently rich in sharp details to really test this.


K

[pduggan]

Fair point, Klaus, but I was trying to spare you all from hellish file sizes. So how 'rich in sharp details' does our image have to be to prove that 1664x15 and 260x96 both equal 24960? IMHO, the reproduction of details on screen has more to do with whether the finished size has enough pixels to show them than whether the image is resized through the infobar or export process. (Think of a chess board and try to render it at 4x4 pixels by any method and you'll see what I mean!)

Peter</p>

Peat Stack or Pete's Tack?</p>