Re: San Francisco Bay Fog



On Nov 20, 4:08 pm, John Navas <spamfilt...@xxxxxxxxxxxxxx> wrote:
On Tue, 20 Nov 2007 01:59:53 -0800 (PST), mark.thoma...@xxxxxxxxx wrote
in <83279945-946b-472a-8c3c-8ddc0783a...@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>:

On Nov 20, 1:32 pm, John Navas <spamfilt...@xxxxxxxxxxxxxx> wrote:
<http://img440.imageshack.us/my.php?image=p1030699filteredok1.jpg>
<http://img440.imageshack.us/my.php?image=p1030701filteredsh9.jpg>
<http://img440.imageshack.us/my.php?image=p1030708filteredag0.jpg>
<http://img440.imageshack.us/my.php?image=p1030714filteredyq3.jpg>
<http://img442.imageshack.us/my.php?image=p1030715filteredlj3.jpg>
Sorry, but even when I first saw the *reduced* view of the first one,
it was clear there were problems, and then when I clicked to see the
full image... [SNIP]

Putting aside the question of whether your comments were influenced by
knowing in advance these images were from my FZ8 ;), I personally think
these images are very good even though they don't fully measure up to
the originals -- these images have been re-compressed to less than 1.5
MB for free image hosting, which did degrade them somewhat.

They are considerably smaller than 1.5MB. I only looked at the first
couple, but they were 350kb and 550kb respectively and compressed by
Adobe Photoshop level 8 giving 0.5+/-0.1bits/pixel. You would probably
have obtained a better looking result by downsampling a factor of 2
and then encoding at a higher quality at 2bits/pixel. And you would
certainly have avoided some of the degradation by using the IJG codec.

I am surprised at the filter damage to skylined trees and some other
areas. The top of the suspension wire in 669 shows significant JPEG
artefacts that appear to have been rencoded. Slowly changing textures
are also tricky for JPEG and the denoised sky looks posterised with
visible contours as a result. At least some of the problems are down
to a mismatch between the cameras JPEG quantisation and Adobe
Photoshops custom tables.

The sample DMC-F8 image I found online used a compression roughly
equivalent to IJG Q=94 and about 2.4MB for a 7Mpixel image. This is a
little bit overcompressed to avoid some loss of fine detail. A
lossless crop of the original 699 in the vicinity of the top of the
tower, and around its base would be most enlightening.

Regards,
Martin Brown
.



Relevant Pages

  • Re: Original JPG
    ... I understand that a new feature in ACR4 allows the processing of JPEG ... Regarding, I do all of my WB and exposure correction in ACR, then move ... you *do* change pixels when you use ACR. ... To me, who didn't save the untouched originals for a couple of years, this ...
    (alt.photography)
  • Re: (serious, heavy rant) Part of the CS dumb-down...
    ... scanner) just print out one copy and then make a jpeg. ... computer ten times faster than mine, that proves he's ten times better than me, and that is how he measures himself against other people. ...
    (sci.research.careers)
  • Re: (serious, heavy rant) Part of the CS dumb-down...
    ... On Tue, 29 May 2007, BMJ wrote: ... just print out one copy and then make a jpeg. ... This same guy, by the way, is the kind of guy who thinks if he has a computer ten times faster than mine, that proves he's ten times better than me, and that is how he measures himself against other people. ...
    (sci.research.careers)
  • Re: (serious, heavy rant) Part of the CS dumb-down...
    ... Straydog wrote: ... just print out one copy and then make a jpeg. ... Buuuut, ooooohhhhhh..nooooooo, we have to clutter up the internet with 10 megabytes of files, and like never before, eat up my computer time and my time. ...
    (sci.research.careers)
  • Re: Geminus JPEG acceleration and rotation now available
    ... Looking at the choices configurations, the ACCELERATE JPEG decoding ... You can get a demo of the JPEG acceleration. ...
    (comp.sys.acorn.apps)

Loading