gordonr

There is not much evidence of a pixel level (matrix) difference between algorithms and packages. By tweaking the values of Unsharp Mask you should be able to do the same.

 

What may be different is that the sharpening parameters could vary across the image to take account of local contrast differences. There is some suggestion that functions like Clarify in Paint Shop Pro 7 modifies contrast in an adaptive way.

This should be another reminder to anyone trying to "archive" their digital files: make sure the format you are using is widely compatible with other applications. (Also a reminder to avoid using version X.00 of any program for critical work ;-)

Telegraphic facsimile machines (faxes) were invented in the 19th century, but those were electromechanical rather than digital. Photographic wire service transmissions were developed in the 1920's. See: http://www.ap.org/pages/history/photos.htm

This feature shows off the value of a database! If you type your own user ID into the URL, you can see which photos you rated (not just the big 5!)

Leonard, if all you do is scan and print without any corrections then 8 bits is fine. As soon as you try to do any major adjustments then you start to lose quality. The thread on <a href="http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=004kNm">why image corrections work smoother after changing a 8bit scan to 16bit</a> goes some way to explaining this. Non-linear functions like gamma, curves, etc cause huge gaps in the histogram if not applied at scanning stage.

I think the difference is that expiration of threads are done by a computer script, while individual deletions are done differently. I agree that this might be helpful, but even more useful might be moving them to some staging area for "Uninteresting Threads" (as per Greenspun). Deleting your own e-mails is probably not a wise move...

All modern scanners use 10-bit or higher internally. Whether this information is available to the outside depends on both hardware and software (Twain interface).<br>

<br>

All images need to be converted to some gamma scale for viewing on a monitor (e.g. 1.80 or 2.20). I drew up a table some time ago showing that more bits are crucial to avoid "banding" for gamma adjustment which has a large effect on the low order values. You can see how the roundoff errors come into play. Once the image has been converted to 8-bit this "precise" information is no longer available.<br>

<br>

Gamma function = input ^ (1/exponent) or 255 * (input/1023) ^ 1/exponent (for 10 bit input)

<pre>

Input 1.40 1.80 2.20

0.00 0 0 0

0.25 2 5 11

0.50 3 8 15

0.75 4 10 18

1.00 5 12 21

1.25 6 13 23

1.50 6 15 25

1.75 7 16 26

2.00 8 17 28

2.25 9 18 30

2.50 9 19 31

3.00 11 22 34

3.50 12 24 36

4.00 13 25 39

4.50 14 27 41

5.00 15 29 43

6.00 17 32 46

7.00 20 35 50

8.00 21 37 53

9.00 23 40 56

10.0 25 42 58

12.0 29 47 63

16.0 35 55 72

20.0 41 62 80

24.0 47 68 87

28.0 53 75 93

32.0 58 80

36.0 63 86

40.0 68 91

44.0 73

48.0 77

52.0 82

56.0 86

60.0 91

64.0 95

255.00 254

255.25 255

255.50 255

255.75 255

</pre>

It's called numerical (compuation) analysis. Image arithmetic is done on integers, and it is neccessary to round up/down in order for the result to be an integer. Increasing the number of available integers from 256 to 65536 reduces the significance of rounding errors (assuming the full range of numbers is used). For display purposes the results are always converted back to 8 bit (256 levels).

 

If you start with a value of 1 and want to brighten it by 50%, the result will be either 1 or 2 depending on how you round it. If the original value was 256 the result will be 384, which has higher "precision" (relative to the scale of 65536). For most purposes the differences are small, but the cumulative effect of multiple operations can result in large "errors".

 

"Real" scientists use floating point arithmetic for calculations (e.g. 1.234e5), but this is much too slow for PC based image preocessing...

Something changed in your cookies! Sometimes posting to photo.net updates the parameter "your last visit, which we think was XXX", other times it doesn't. This question comes up periodically, and never seems to get a consistent answer.

It won't work, so try another way of doing it. Like perpetual motion machines, and human powered flight the answers take a lot of time to explain. Previous explanations unfortunately haven't been preserved. This thread doen't contain much value, but is now in the archived forum...

I can't add to the story, but the incident received wide news coverage, including South Africa where the surviving male was born.

It's time for the elves to read Philip Greenspun's original charter - don't mess with the user interface: <a href="http://philip.greenspun.com/panda/html.html">Chapter 3: Learn to Program HTML in 21 Minutes</a> (http://philip.greenspun.com/panda/html.html).

<a href="http://www.irfanview.com/">Irfanview<a/> www.irfanview.com

I agree that there is a problem - sorting by rating should only be available if there are 5 or more ratings (to be statistically meaningful). This was the situation previously, and seems to have been overlooked during a recent update.

Try reading the thread in the archives <a href="http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=000gO7">Do digital cameras have a DPI setting</a> (http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=000gO7)

My findings are a little different from Ellis. When doing multiple resampling using bicubic method most of the detail is preserved. Using multiple bilinear resampling produces a very soft result.

 

Paint Shop Pro has an option called Smart Resize which automatically uses bicubic when enlarging, and bilinear when reducing.

 

There is always a tradeoff between image sharpness and Jpeg compressed size, a compromise that will never please everyone.

White background was chosen a long time ago by Philip Greenspun, and has remained that way since. I chose to turn off all colour, and set my background to grey (works for me, but tough on site designs...)

Jpeg re-compression was introduced for various reasons, one of them being that a few people who don't understand the rules (or choose not to follow them) caused problems for other users. The guidelines should probably be updated, but the rules are valid.

<br>

<br>

Jpeg compression levels the playing fields, and has been in place since Philip Greenspun started this site (many years ago):

<a href="http://philip.greenspun.com/panda/images">Philip and Alex's Guide to Web Publishing - Chapter 6: Adding Images to Your Site</a> (http://philip.greenspun.com/panda/)

<br>

<br>

There are many archived threads on this subject: <a href="http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=003j8d">Browser Problems Caused by Photoshop 7 JPEG images

</a> in 2002 (http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=003j8d), <a href="http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=000bKu">Why are Phil's JPEGs smaller than mine</a> in 1999 (http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=000bKu), <a href="http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=000jrw">Ideal compression rate for JPEG Format</a> in 2000

(http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=000jrw), <a href="http://www.photo.net/bboard/q-and-a-fetch-msg.tcl?msg_id=0009CX&msg_id=0009CX">A JPEG question</a> in 1998

(http://www.photo.net/bboard/q-and-a-fetch-msg.tcl?msg_id=0009CX&msg_id=0009CX)

The fact that it is out of production doesn't necessarily mean you wont be able to get it repaired. Many of the Pentax cameras were produced in large volumes and share common components (like cloth shutters), so ask around and try getting a quote for repairs.

This kind of question has been asked before, and covered in the archives. Try reading <a href="http://www.photo.net/bboard/q-and-a-fetch-msg.tcl?msg_id=000jrw">Ideal compression rate for JPEG Format</a> (http://www.photo.net/bboard/q-and-a-fetch-msg.tcl?msg_id=000jrw) and other related threads.

Don't forget the article on photo.net <a href="http://www.photo.net/astro/star-streak">Star Streaks</a> by Philip Greenspun (http://www.photo.net/astro/star-streak).

Statistically you are right in line with the average, since the majority of photos posted never get comments on photo.net. The top of this page indicates 440000 images with 330000 comments...

My old UMAX Astra 1200S flatbed has an 8.5x14in transparency adapter. It's resolution is only 600dpi, but this is more than adequate for web work. I didn't respond earlier, because you can't fit 11x14in on that...

The crime issue is somewhat off-topic, but Allan's point is that it "seems" to be an issue, i.e. it is something that the media hype and influence public perceptions. I have lived in Cape Town for 40 years and have never been mugged or felt personally threatened (nor have any members of my family). Nature photography is one of the safest things I know - other people are a risk if you walk around with expensive cameras anywhere in the world. If you are too scared to travel, then look at my photos of the area instead...

The most loss occurs the first time you convert to Jpeg, and subsequent times the loss may be so small as to be undetectable (if you use the same image editor and quality setting). The combination of editing, closing and re-opening produces more errors due to an accumulation of quantisation (rounding effects). There is no "single" answer to this question, since there are as many Jpeg standards as there are image editors, and at the highest settings the differences are barely detectable (at low settings the results can be terrible). There are already some threads on this subject in the archived forum.