Manipulation sequence?

In article <44ef971e$0$1373$>,
"KSL" < dot com.au> wrote:

Adjust the curve
Resample down to 1024×768.
Apply unsharp mask
Convert to 8-bit to save as a JPEG

In what sequence should I apply the adjustments?

In exactly the sequence you just named.

—
Art, photography, shareware, polyamory, literature, kink: all at http://www.xeromag.com/franklin.html
Nanohazard, Geek shirts, and more: http://www.villaintees.com

"tacit" wrote in message

In article <44ef971e$0$1373$>,
"KSL" < dot com.au> wrote:

Adjust the curve
Resample down to 1024×768.
Apply unsharp mask
Convert to 8-bit to save as a JPEG

In what sequence should I apply the adjustments?

In exactly the sequence you just named.

—

Wouldn’t you apply sharpening as the last step, ie. before the conversion to 8 bit? Also, if his resizing involves any cropping, might it not be an idea to do this before the curve adjustment? Just a couple of (probably useless!!) thoughts…..
Cheers.

"tacit" wrote in message

In article <44ef971e$0$1373$>,
"KSL" < dot com.au> wrote:

Adjust the curve
Resample down to 1024×768.
Apply unsharp mask
Convert to 8-bit to save as a JPEG

In what sequence should I apply the adjustments?

In exactly the sequence you just named.

I disagree. I would apply Sharpening while the image still contains the maximum possible information, i.e. before resampling or compression.

1 Apply unsharp mask
2 Adjust the curve
3 Resample down to 1024×768.
4 Convert to 8-bit to save as a JPEG

Only my opinion of course.

John

John Rampling wrote:

I disagree. I would apply Sharpening while the image still contains the maximum possible information, i.e. before resampling or compression.
1 Apply unsharp mask
2 Adjust the curve
3 Resample down to 1024×768.
4 Convert to 8-bit to save as a JPEG

Only my opinion of course.

Read a book about Photoshop, any book about Photoshop, and you’ll see that if there is anything at all that all authors agree upon it’s the fact that you should sharpen LAST. A small image needs a different amount of sharpening, so that’s why the general consensus is that you should sharpen AFTER resizing.

—
Johan W. Elzenga johan<<at>>johanfoto.nl Editor / Photographer http://www.johanfoto.nl

In article <iqVHg.27519$>,
"John Rampling" wrote:

I disagree. I would apply Sharpening while the image still contains the maximum possible information, i.e. before resampling or compression.

If you sharpen, then resample, you lose the benefit of the sharpening. Unsharp mask works by exaggerating areas of high contrast–that is, along a juncture of light and dark, it makes the light side lighter and the dark side darker. The amount of sharpening for maximum benefit depends on, among other things, the resolution of the image.

If you sharpen the image at high resolution, then downsample it, the edge pixels that have been lightened or darkened are averaged out during the resampling process, and the benefit of the sharpening is lost.

—
Art, photography, shareware, polyamory, literature, kink: all at http://www.xeromag.com/franklin.html
Nanohazard, Geek shirts, and more: http://www.villaintees.com

In article <ecotdq$686$>,
"Mike" wrote:

Wouldn’t you apply sharpening as the last step, ie. before the conversion to 8 bit?

I would not, simply because a higher bit depth affords greater room for the lightening and darkening along edges that are how unsharp masking works. In practical terms, however, I suspect the difference between sharpening before reduction to eight bits per channel and sharpening after is probably negligible.

—
Art, photography, shareware, polyamory, literature, kink: all at http://www.xeromag.com/franklin.html
Nanohazard, Geek shirts, and more: http://www.villaintees.com

John Rampling wrote:

I disagree. I would apply Sharpening while the image still contains the maximum possible information, i.e. before resampling or compression.

There are work flows with two sharpening steps, a very light initial sharpening (for example with my Canon 1D series bodies Canon recommends 300%, 0.3 and 0 USM as the first step to get rid of the AA filter blur) then a second more aggressive sharpening as the final step.

But to do your main sharpening before resampling usually makes the final downsampled image look oversharpened, or so it seems with the images I’ve worked on.

Tacit was right.

Bill

"Bill Hilton" wrote in message
SNIP

But to do your main sharpening before resampling usually makes the final downsampled image look oversharpened, or so it seems with the images I’ve worked on.

That’s to be expected. One should avoid *any* sharpening before down-sampling. The down-sampling will create aliasing artifacts, and it will be worse with lots of detail that’s smaller than can be represented in the final image’s pixels. Good down-sampling algorithms (unfortunately Photoshop is not very good in that respect) apply a resampling (low-pass) filter that reduces the artifact generating detail.

That also means that it’s better to *not* apply ‘capture’ sharpening, because it’ll frustrate any subsequent downsampling (e.g. samples/web-publishing/thumbnails/etc.).

—
Bart

For xray or radiographic images, sharpening is unlikely to be consequential in this particular case, since the images are so soft to begin with.

If the OP wants to preserve the look of an xray, using curves is not the way to go. for a variety of reasons, medical images use a level/window adjustment, similar to brigthness and contrast, with presets that depend on the type of image. Any other adjustment to the image will look strange to a medical person who is familiar with such images.
—

Mike Russell
www.curvemeister.com/forum/

"Mike Russell" wrote in message

For xray or radiographic images, sharpening is unlikely to be consequential in this particular case, since the images are so soft to begin with.

Although you are quite right in bringing this up, I beg to differ on your statement with regards to down-sampling. Traditional X-rays have relatively high speed emulsions (unlike mammograms, or single emulsion films, for CRT output, or for industrial use), and they can exhibit graininess. Down-sampling, especially after sharpening, will lead to grain-aliasing.

If the OP wants to preserve the look of an xray, using curves is not the way to go.

That depends on the use. If the images are to be used in print, or in projection, then the look on the output modality is probably more important. After down-sampling, I assume (and hope) they will not be used for diagnosis. We’re probably talking about publication purposes in this case.

—
Bart

Mike

You are quite right about the curves. The software (Sidexis) that we use to process our X-Rays (wrong term, but I

Bart van der Wolf wrote:

"Bill Hilton" wrote in message
SNIP

But to do your main sharpening before resampling usually makes the final downsampled image look oversharpened, or so it seems with the images I’ve worked on.

That’s to be expected. One should avoid *any* sharpening before down-sampling. The down-sampling will create aliasing artifacts, and it will be worse with lots of detail that’s smaller than can be represented in the final image’s pixels. Good down-sampling algorithms (unfortunately Photoshop is not very good in that respect) apply a resampling (low-pass) filter that reduces the artifact generating detail.
That also means that it’s better to *not* apply ‘capture’ sharpening, because it’ll frustrate any subsequent downsampling (e.g. samples/web-publishing/thumbnails/etc.).

I was told by a photography teacher that it’s better, IIRC, to use bi-cubic sharper when down sampling, and bi-cubic smoother when upsampling. Is this true? And is there a memnomic for remembering which is which?

—
John McWilliams

John McWilliams wrote:
I was told by a photography teacher that it’s better, IIRC, to use bi-cubic sharper when down sampling, and bi-cubic smoother when upsampling. Is this true?

Yes …

And is there a memnomic for remembering which
is which?

It’s described this way in the Help files in case you forget …

"John McWilliams" wrote in message
SNIP

I was told by a photography teacher that it’s better, IIRC, to use bi-cubic sharper when down sampling, and bi-cubic smoother when upsampling. Is this true?

That is the recommendation from Adobe (help file), but I do not agree on the down-sampling / bicubic sharper workflow, because it creates worse images than with straight bicubic + sharpening.
Using bi-cubic smoother with upsampling makes some sense, because the pixelation artifacts are suppressed a bit better, but not as good as a few other programs.

This experiment compares the results from several down-sampling methods:
< http://www.xs4all.nl/~bvdwolf/main/foto/down_sample/down_sam ple.htm> I’m sorry having to say it, but Photoshop is not very good at down-sampling, probably the result of choosing speed over quality.

—
Bart