Increasing resolution step-wise?

PF
Posted By
Peter Frank
Dec 9, 2004
Views
738
Replies
19
Status
Closed
Hi,

I have some CCD camera images (no complex motifs, just gray-scale pictures of spots, obtained from a molecular biology experiment) that are too low in resolution to be magnified without getting very pixely (CCD camera just doesn’t give a higher resolution). Therefore, I want to increase the image resolution using Photoshop.
I principally know how to do this but I remember having read somewhere that it is better to increase the resolution step-wise (around 10 % for each step) instead of maybe 100 % at once. Is that true? Are there any other things I should consider when increasing the resolution without getting too many artifacts?

Peter

Must-have mockup pack for every graphic designer 🔥🔥🔥

Easy-to-use drag-n-drop Photoshop scene creator with more than 2800 items.

B
bhilton665
Dec 10, 2004
From: Peter Frank

I want to increase the image resolution using Photoshop.
I principally know how to do this but I remember having read somewhere that it is better to increase the resolution step-wise (around 10 % for each step) instead of maybe 100 % at once. Is that true?

This is commonly known as "stair interpolation" … try it both ways, in 10% steps with simple ‘bicubic’ and, if you have Photoshop CS, in one jump with ‘bicubic smoother’ …

Here’s the web site that best describes stair interpolation (he’s selling an action to do it but it’s simple to write your own).
http://www.fredmiranda.com/SI/

I have some CCD camera images (no complex motifs, just gray-scale pictures of spots)

Something funky like this (as opposed to photographic images with more random visual structures) might be better off with stair interpolation at even smaller steps so try it several ways if you don’t get good results with the two suggestions above.

Bill
P
patrick
Dec 10, 2004
If you are using Photoshop CS thee is no longer a need for incremental scaling.
Select Resample and Bicubic Smooth to upscale and Bicubic Sharpen to downscale.
Good luck! . . . . patrick
"Peter Frank" wrote in message
Hi,

I have some CCD camera images (no complex motifs, just gray-scale pictures of spots, obtained from a molecular biology experiment) that are too low in resolution to be magnified without getting very pixely (CCD camera just doesn’t give a higher resolution). Therefore, I want to increase the image resolution using Photoshop.
I principally know how to do this but I remember having read somewhere that it is better to increase the resolution step-wise (around 10 % for each step) instead of maybe 100 % at once. Is that true? Are there any other things I should consider when increasing the resolution without getting too many artifacts?

Peter
B
bogus
Dec 10, 2004
Hmm, wouldn’t you use Bicubic Smooth to DOWNSCALE and Bicubic Sharpen to UPSCALE?

Because when you upscale an image the interpolation softens the image and sharpening is often necessary.

patrick wrote:
If you are using Photoshop CS thee is no longer a need for incremental scaling.
Select Resample and Bicubic Smooth to upscale and Bicubic Sharpen to downscale.
Good luck! . . . . patrick
"Peter Frank" wrote in message

Hi,

I have some CCD camera images (no complex motifs, just gray-scale pictures of spots, obtained from a molecular biology experiment) that are too low in resolution to be magnified without getting very pixely (CCD camera just doesn’t give a higher resolution). Therefore, I want to increase the image resolution using Photoshop.
I principally know how to do this but I remember having read somewhere that it is better to increase the resolution step-wise (around 10 % for each step) instead of maybe 100 % at once. Is that true? Are there any other things I should consider when increasing the resolution without getting too many artifacts?

Peter

P
patrick
Dec 10, 2004
See Jack Davis & Ben Willmore, "How to Wow, Photoshop for Photography" They provide a rationale for sharpening on scaling down and smoothing for scaling up.
You can set your general preferences for either in the Edit>Preferences function.
.. . . . patrick

"bogus" wrote in message
Hmm, wouldn’t you use Bicubic Smooth to DOWNSCALE and Bicubic Sharpen to UPSCALE?

Because when you upscale an image the interpolation softens the image and sharpening is often necessary.

patrick wrote:
If you are using Photoshop CS thee is no longer a need for incremental scaling.
Select Resample and Bicubic Smooth to upscale and Bicubic Sharpen to downscale.
Good luck! . . . . patrick
"Peter Frank" wrote in message

Hi,

I have some CCD camera images (no complex motifs, just gray-scale pictures of spots, obtained from a molecular biology experiment) that are too low in resolution to be magnified without getting very pixely (CCD camera just doesn’t give a higher resolution). Therefore, I want to increase the image resolution using Photoshop.
I principally know how to do this but I remember having read somewhere that it is better to increase the resolution step-wise (around 10 % for each step) instead of maybe 100 % at once. Is that true? Are there any other things I should consider when increasing the resolution without getting too many artifacts?

Peter
EG
Eric Gill
Dec 10, 2004
bogus wrote in
news:9Hgud.10$:

Hmm, wouldn’t you use Bicubic Smooth to DOWNSCALE and Bicubic Sharpen to UPSCALE?

No.

Because when you upscale an image the interpolation softens the image and sharpening is often necessary.

Or wrecks the image, bringing out the upsampling artifacts.

However, downsampling also softens the image. I personally do not trust auto sharpening routines and do so as a seperate step, followed by final color correction, since sharpening can result in color shift.

YMMV.
B
bhilton665
Dec 10, 2004
From: bogus

Hmm, wouldn’t you use Bicubic Smooth to DOWNSCALE and Bicubic Sharpen to UPSCALE?

Go to the CS Help file and search for ‘bicubic’ and pick the ‘choosing an interpolation’ link … here’s the gist of what Help says about this (which you can test out for yourself to see that they are right) …

For Interpolation, choose one of the following options:
Nearest Neighbor for the fast but less precise method. This method is recommended for use with illustrations containing non-anti-aliased edges, to preserve hard edges and produce a smaller file. However, this method can result in jagged effects, which become apparent when distorting or scaling an image or performing multiple manipulations on a selection.

Bilinear for a medium-quality method.

Bicubic for the slow but more precise method, resulting in the smoothest tonal gradations.

Bicubic Smoother when you’re enlarging images.

Bicubic Sharper for reducing the size of an image. This method maintains the detail in a resampled image. It may, however, over-sharpen some areas of an image. In this case, try using Bicubic.
J
jjs
Dec 10, 2004
Methinks the OP might be more interested in quantitative depictions, not mere illustration.

I suggest that he check out Reindeer’s products which serve well in that regard. See: http://www.reindeergraphics.com/

There are some shareware and freeware tools that do similar quantitative imagings, including one very nice piece that creates spreadsheets, statistical tables and graphics from plots without legends. Adobe als has/had some papers on using Photoshop for analysis. When I return to the day job, I can post more links.
AM
Andrew Morton
Dec 10, 2004
I have some CCD camera images (no complex motifs, just gray-scale pictures of spots, obtained from a molecular biology experiment) that are too low in resolution to be magnified without getting very pixely (CCD camera just doesn’t give a higher resolution). Therefore, I want to increase the image resolution using Photoshop.

….do make sure you explain with the final image that it is not the image as seen by the camera.
An equally valid way to increase the resolution would be to draw a copy by hand; you might even get better results than the camera by peering down the microscope and drawing from that (but not if your drawing ability is like mine).

Andrew
B
bagal
Dec 11, 2004
Eric Gill wrote:
bogus wrote in
news:9Hgud.10$:

Hmm, wouldn’t you use Bicubic Smooth to DOWNSCALE and Bicubic Sharpen to UPSCALE?

No.

Because when you upscale an image the interpolation softens the image and sharpening is often necessary.

Or wrecks the image, bringing out the upsampling artifacts.
However, downsampling also softens the image. I personally do not trust auto sharpening routines and do so as a seperate step, followed by final color correction, since sharpening can result in color shift.
YMMV.
Wot? It blows the image?

Do you know what you are writing about?

I have had sub 500KB files printed at 15" by 10" using upsampling and they (IMHO) are far superior to 6" by 4" at the original resolution (even better than 8" by 6" too FWIW)

Articulean Genius
O
Odysseus
Dec 11, 2004
In article <9Hgud.10$>,
bogus wrote:

Hmm, wouldn’t you use Bicubic Smooth to DOWNSCALE and Bicubic Sharpen to UPSCALE?

Because when you upscale an image the interpolation softens the image and sharpening is often necessary.
That’s bass-ackwards. Sharpening an upsampled image is just going to enhance the edges of the original pixels, including anu graininess or noise that was there to start with. The main objective here is to smooth out artifacts — it’s not reasonable to expect more ‘real’ detail to appear. OTOH downsampled images lose definition because small features tend to get blurry by being averaged with their backgrounds; sharpening compensates for this to some extent.


Odysseus
B
bagal
Dec 11, 2004
Odysseus wrote:
In article <9Hgud.10$>,
bogus wrote:

Hmm, wouldn’t you use Bicubic Smooth to DOWNSCALE and Bicubic Sharpen to UPSCALE?

Because when you upscale an image the interpolation softens the image and sharpening is often necessary.

That’s bass-ackwards. Sharpening an upsampled image is just going to enhance the edges of the original pixels, including anu graininess or noise that was there to start with. The main objective here is to smooth out artifacts — it’s not reasonable to expect more ‘real’ detail to appear. OTOH downsampled images lose definition because small features tend to get blurry by being averaged with their backgrounds; sharpening compensates for this to some extent.
Dear Odysseus may I take issue?

It is IMHO to do with perception

The perceived image on a 6 by 4 is not the same as on a 15 by 10

It’s like the proverbial comparison of chalk and cheese

It dunt work!

Well, at the perceptive level though I do readily accept that there are strong differences at the technical level

Aerticeus

ps – kind regards to the Odysei
EG
Eric Gill
Dec 11, 2004
Aerticulean Effort wrote in
news:uZqud.833$:

Eric Gill wrote:
bogus wrote in
news:9Hgud.10$:

Hmm, wouldn’t you use Bicubic Smooth to DOWNSCALE and Bicubic Sharpen to UPSCALE?

No.

Because when you upscale an image the interpolation softens the image and sharpening is often necessary.

Or wrecks the image, bringing out the upsampling artifacts.
However, downsampling also softens the image. I personally do not trust auto sharpening routines and do so as a seperate step, followed by final color correction, since sharpening can result in color shift.

YMMV.
Wot? It blows the image?

Yes.

Do you know what you are writing about?

16+ years in the image biz.

I have had sub 500KB files printed at 15" by 10" using upsampling and they (IMHO) are far superior to 6" by 4" at the original resolution (even better than 8" by 6" too FWIW)

Ah, well, it’s good to dream.
B
bagal
Dec 11, 2004
Eric Gill wrote:
Ah, well, it’s good to dream.

Hi Eric

While I agree that there are differences in technical quality between printed images at different sizes (well, it’s a fact innit? Look at the nearest billboard from 200 yards, 100 yards, 10 yards, 1 foot, 1 inch thru a magnifier then through a microscope

What does one notice?
Well, in the main the image changes,

Now why can that be? We know the arrangement of colored dots on the billboard is for the timescale I have in mind and assuming there are no tornadoes or storms on the go during the comparison, the arrangement of colored dots is constant.

Conclusion 1: perception is influenced by distance from the printed image

8< snipping the rest of the dialogue to conclude that IMHO some perception of image is best optimised by some particular image sizes

eg – a detailed landscape may look cute on 6 by 4, it may look fantastic on 15 by 10

Aerticeus
J
jjs
Dec 11, 2004
"Aerticulean Effort" wrote in message
[…]
Conclusion 1: perception is influenced by distance from the printed image

That is called Standard Viewing Distance. It has nothing to do with your assertion that interpolation makes prints that look better than noninterpolated prints of the same size.

eg – a detailed landscape may look cute on 6 by 4, it may look fantastic on 15 by 10

If you stand far enough away, the 15×10 looks the same as a 6×4.

One of my collector-cars has what is called a 50-foot paint job. It looks stunning on the road but walk up to it and it looks like it was painted with a bucket and broom – kinda like an interpolated paint job. 🙂
EG
Eric Gill
Dec 13, 2004
Aerticulean Effort wrote in news:CxBud.49$mt5.21
@newsfe5-gui.ntli.net:

Conclusion 1: perception is influenced by distance from the printed image

Indeed. But this has nothing to do with upsampling.
J
jjs
Dec 14, 2004
"Eric Gill" wrote in message
Aerticulean Effort wrote in news:CxBud.49$mt5.21
@newsfe5-gui.ntli.net:

Conclusion 1: perception is influenced by distance from the printed image

Indeed. But this has nothing to do with upsampling.

Sure it does, Eric. If an 8×10 print looks very good at 10", and one upsamples 6x then he should view the subsequent print from 6 times the diagonal as far (76") but wait – let me add the JJS rule – if the same person dares to claim it will have the same quality, then the forumula is (Most Distant State Line Boundary) + 100 miles. I mean, like, Get Out Of Town!
B
bagal
Dec 14, 2004
jjs wrote:
"Eric Gill" wrote in message

Aerticulean Effort wrote in news:CxBud.49$mt5.21
@newsfe5-gui.ntli.net:

Conclusion 1: perception is influenced by distance from the printed image

Indeed. But this has nothing to do with upsampling.

Sure it does, Eric. If an 8×10 print looks very good at 10", and one upsamples 6x then he should view the subsequent print from 6 times the diagonal as far (76") but wait – let me add the JJS rule – if the same person dares to claim it will have the same quality, then the forumula is (Most Distant State Line Boundary) + 100 miles. I mean, like, Get Out Of Town!
OK let’s be realistic

6 by 4’s OK for looking close (well you have to)

15 by 10’s OK for not looking quite as close

equally, put a 15 by 10 and a 6 by 4 on a display board – same image printed on different sizes

start from about 4 yards away – which looks best?

progress towards the display in an orderly fashion

at which point does the superior quality of the 6 by 4 overwhelm the inferiority of the 6 by 4?

record the distance and let’s call it X"

If you want prints viewed from 0" to X" then print 6 by 4

If you want prints viewed from X" onwards print 15 by 10

s’easy really

Aerticeus
B
bagal
Dec 14, 2004
Aerticulean Effort wrote:
jjs wrote:

"Eric Gill" wrote in message

Aerticulean Effort wrote in news:CxBud.49$mt5.21
@newsfe5-gui.ntli.net:

Conclusion 1: perception is influenced by distance from the printed image

Indeed. But this has nothing to do with upsampling.

Sure it does, Eric. If an 8×10 print looks very good at 10", and one upsamples 6x then he should view the subsequent print from 6 times the diagonal as far (76") but wait – let me add the JJS rule – if the same person dares to claim it will have the same quality, then the forumula is (Most Distant State Line Boundary) + 100 miles. I mean, like, Get Out Of Town!
OK let’s be realistic

6 by 4’s OK for looking close (well you have to)

15 by 10’s OK for not looking quite as close

equally, put a 15 by 10 and a 6 by 4 on a display board – same image printed on different sizes

start from about 4 yards away – which looks best?

progress towards the display in an orderly fashion

at which point does the superior quality of the 6 by 4 overwhelm the inferiority of the 15 by 10 <====== OOOPS!

record the distance and let’s call it X"

If you want prints viewed from 0" to X" then print 6 by 4
If you want prints viewed from X" onwards print 15 by 10
s’easy really

Aerticeus

there is a typo in the previous post

it’s sorta obvious and OOOPSed above

A
ps – you can use the patented Aerticean method to discern distances for other prints to such as Y" for an 18 by 12 and so on

A

pps – 15 by 10’ed and 6 by 10’ed the same images using PSCS R&R interpolation. The responses made by other people at comparing the images run something like
Prefer 15 by 10 ~ 100%
Prefer 6 by 4 ~ 0%

but this is also content related

A
J
jjs
Dec 14, 2004
"Aerticulean Effort" wrote in message
[…]

Standard Viewing Distance, the resolution of the human eye under varying circumstances, apparent sharpness (accutanc) and all that is PERFECTLY WELL DOCUMENTED, and has been for decades. I’m not going to repeat the literature here, but strongly suggest that you do the same and save yourself the grief of rediscovering the wheel, possibly quite incorrectly.

How to Improve Photoshop Performance

Learn how to optimize Photoshop for maximum speed, troubleshoot common issues, and keep your projects organized so that you can work faster than ever before!

Related Discussion Topics

Nice and short text about related topics in discussion sections