Posting photos that look good

[Gary L. Quay]

I'm perplexed. I never can seem to get a photo to look good when I post it to this site. Any gradual color of tonal shifts become banded, and I have to shrink the picture to get it within the size limits. Yet, I see large, great looking pictures all over the site. Any pointers on how to do it right?

--Gary

[Joanna Carter]

I'm perplexed. I never can seem to get a photo to look good when I post it to this site. Any gradual color of tonal shifts become banded, and I have to shrink the picture to get it within the size limits. Yet, I see large, great looking pictures all over the site. Any pointers on how to do it right?
Assuming that you are using Photoshop...

1. Edit | Convert to Profile... sRGB
2. Image | Image Size... Bicubic Sharper (e.g.)700pixels on long side and 96dpi

(I use Genuine Fractals instead; it seems to give better results)

3. Save | Save for Web & Devices... jpg

Joanna

[cjbroadbent]

....700pixels on long side and 96dpi... Joanna
Joanna,
Tell me more about 96 dpi. Isn't 72 dpi standard?

[jb7]

I don't think the dpi matters at all-
only pixel dimensions-

96dpi is the screen res of my macbook pro, though-

I think Gary is asking about pictures embedded in posts,
and not clickable thumbnails-

For that, you have to use tags, little pieces of code that tells the browser where to go to display an image.
After following Joanna's advice above, regarding re-sizing for the web,
you should host the image- either on your own website, or on any of the image hosting sites- imageshack.us, flickr, pbase, whatever-

Once uploaded, the image will have an address- beginning with http...
Placing the tags at the beginning, and at the end of the address (with no spaces in between) will force the browser to look up the image, and display it on the page-

I'm not a computer person, so details might be incorrect, but that is my understanding...

joseph

[Marko]

Christopher, it is only the pixel size and the color space that really matter, dpi is irrelevant for screen representation. It is basically a number that instructs the printer how many Dots Per Inch to lay down on the paper, hence the name. It is not the same as Dot Pitch used for displays.

Hope this makes sense :)

[Frank Petronio]

You can LIGHTLY sharpen your images once they are web-sized to reduce the softening effects of the jpg save. As for the banding, all you can do is use higher quality jpg settings OR modify your file by adding a small amount of noise (which will increase the size of your jpg.)

All of these photo posting sites have automagic software that will resize (and sometimes strip the profile or sharpen) your photos to suit the webmasters (see what myspace and facebook do) and sometime you have to experiment.

Nobody has used a 72dpi monitor in years (unless you are visually disabled and want larger text) -- just like printing, it is the amount of pixels that is important, you can post 1200 dpi images (albeit with small physical dimensions). I think the limit here is something like 500 pixels wide.

[timbo10ca]

Assuming that you are using Photoshop...

1. Edit | Convert to Profile... sRGB
2. Image | Image Size... Bicubic Sharper (e.g.)700pixels on long side and 96dpi

(I use Genuine Fractals instead; it seems to give better results)

3. Save | Save for Web & Devices... jpg

Joanna

Could you tell me more about this bicubic sharper please? I've never used it, but it sounds interesting. I've always downsized using bicubic then applied USM to the degree I feel necessary. I like the control this allows me. Does Bicubic Sharper keep the exact same image sharpness/ ghosting/ etc of the pre-downsized image? This sounds like it could save me a step. My workflow is:

Scan at max ppi (non-uprezed), save as tiff
Downsize tiff in PS to true scanner ppi (bicubic)
Edit and sharpen
Save as tiff for printing, using ZIP compression
Run script in PS to convert to 8bit RGB and downsize (bicubic) to web size (400 pixels high), save as uncompressed jpg
Apply sharpening to this jpg to account for softness induced by downsizing (USM 1 pixel, about 30-50%), save again as jpg, upload to web.

Will Bicubic Sharpen improve my results at either of the downsize stages?

Tim

[Joanna Carter]

Could you tell me more about this bicubic sharper please? I've never used it, but it sounds interesting. I've always downsized using bicubic then applied USM to the degree I feel necessary. I like the control this allows me. Does Bicubic Sharper keep the exact same image sharpness/ ghosting/ etc of the pre-downsized image? This sounds like it could save me a step. My workflow is:

Scan at max ppi (non-uprezed), save as tiff
Downsize tiff in PS to true scanner ppi (bicubic)
??? Not sure what you mean here. You should, usually, if you intend to print up to the maximum size your scans will allow, then you need to use, providing your scanner supports it as a true optical resolution, 2400 dpi.

This will give you, approximately, 10x enlargement from the tranny/neg and is close to what the film actually resolves.

But, you just scan at 2400ppi, no resizing or scaling is necessary. Once the image is in PS, use the Image | Change Size dialog to change the resolution, without scaling, to 240ppi, as a final working and printing resolution.


Edit and sharpen
You should not sharpen at this stage, unless you are using Genuine Fractals for resizing. Photoshops resizing can really wreck a good image, if you reduce the image size after sharpening.


Save as tiff for printing, using ZIP compression
IF you are using Photoshop, why not save as PSD? It is a lossless compression. Regardless of the file format, I wouldn't use any extra compression.


Run script in PS to convert to 8bit RGB and downsize (bicubic) to web size (400 pixels high), save as uncompressed jpg
You don't need to explicitly convert to 8bit, simply convert the profile to sRGB.

As I mentioned in my first reply, I use Genuine Fractals for resizing, for a couple of reasons:

Photoshop can make a mess of a sharpened image when downsizing, if that were my only option, I would not sharpen the image until it had been downsized, I would use Bicubic sharper to resize and then apply sharpening afterwards.

Genuine Fractals seems to work much better on a sharpened image so, in this case, I would do my final sharpening before resizing. During the resizing, GF allows you to adjust the sharpness of the resized image to accomodate any loss during the process.

I usually use the "Save for Web and Devices.." option to create a jpg file, it seems to handle things better.

Regarding compressing the jpg, I tend to compress as much as is needed to create a certain file size.


Apply sharpening to this jpg to account for softness induced by downsizing (USM 1 pixel, about 30-50%), save again as jpg, upload to web.
In my opinion, it would be foolish to sharpen the jpg if you have already sharpened the original image; best to leave all sharpening to after you have resized, unless you use Genuine Fractals.


Will Bicubic Sharpen improve my results at either of the downsize stages?
If you don't have GF, then yes, it should, but be careful not to sharpen before resizing.

[RJC]

You should not sharpen at this stage, unless you are using Genuine Fractals for resizing. Photoshops resizing can really wreck a good image, if you reduce the image size after sharpening.

If you don't have GF, then yes, it should, but be careful not to sharpen before resizing.

This depends on how you are sharpening. I have CS2 and sharpen in 3 stages - 1. source sharpening of scan, 2. content sharpening and 3. output sharpening on layers and masked for edges as per Bruce Frasers recommendations (ref Real World Image Sharpening with Adobe Photoshop CS2) and implemented in PhotoKit Sharpener. I have absolutely no issues resizing at any stage with CS2.

And in my readings in other fora it appears that the consensus of opinion is that from CS3 onwards Photoshop outperforms Genuine Fractals for resizing.


2. Image | Image Size... Bicubic Sharper (e.g.)700pixels on long side and 96dpi

Ignore dpi unless you want your image to display at a specific measured size on the monitor, e.g. 6 inches on a 120 dpi monitor.


IF you are using Photoshop, why not save as PSD? It is a lossless compression. Regardless of the file format, I wouldn't use any extra compression.

This is irrelevant - TIFFs can be losslessly compressed using LZW or ZIP. Anyhow compression of your master file is just a trade-off between smaller files on disk or slower open/save times depending upon the size of your hard drive and speed of your processor.

Rob

[timbo10ca]

??? Not sure what you mean here. You should, usually, if you intend to print up to the maximum size your scans will allow, then you need to use, providing your scanner supports it as a true optical resolution, 2400 dpi.

This will give you, approximately, 10x enlargement from the tranny/neg and is close to what the film actually resolves.

But, you just scan at 2400ppi, no resizing or scaling is necessary. Once the image is in PS, use the Image | Change Size dialog to change the resolution, without scaling, to 240ppi, as a final working and printing resolution.

I have been advised on a number of occasions that scanning at max dpi then downresing to the true scanner dpi will give a more detailed scan.

I haven't worked with psd, and just use tiff in zip to keep it simple, plus it's lossless. I will look more at saving as psd's for saving disc space though.

I sharpen prior to resizing because it's my ready-to-print final image. I have noticed quality loss when printing at lower dpi settings (300), so I leave the file at it's maximum dpi and print it, regardless of print size. To make it simple, I just resize that image for web and do a touch-up sharpen. I've been happy with the results so far, but I'll have to try saving a non-sharpened image, resize it using bicubic sharpen, then sharpen it and compare to one the way I've traditionally done it. I don't have GF, but have CS4- I just haven't learned it yet (I've been using CS2). I will also have to compare to a CS4 created jpg, done in my usual manner.

Another reason I sharpen my original image is because I've found it to give better results when spotting. On a number of occasions, I've spotted then sharpened, only to find new spots due to the sharpening. So I sharpen prior to spotting, which is now my final step before saving.

Tim

[timbo10ca]

This depends on how you are sharpening. I have CS2 and sharpen in 3 stages - 1. source sharpening of scan, 2. content sharpening and 3. output sharpening on layers and masked for edges as per Bruce Frasers recommendations (ref Real World Image Sharpening with Adobe Photoshop CS2) and implemented in PhotoKit Sharpener.

Rob- could you please elaborate a bit on this? I'm not sure what the difference between source sharpening and content sharpening would be. I'm not familiar with these terms, and you're 3rd step is total greek to me. I've been using USM, and have found the Smart Sharpen not to be useful- no user control, it seems. I've recently learned about sharpening in LAB, but haven't tried it yet (mostly because I'm currently working on a bunch of B&W images, and the way I understand it, LAB sharpening seems to be for color). I've also heard of sharpening using a high pass filter, but haven't figured this one out. There seems to be as many ways to sharpen as there are opinions on it.

Thanks,
Tim

[Joanna Carter]

I sharpen prior to resizing because it's my ready-to-print final image. I have noticed quality loss when printing at lower dpi settings (300), so I leave the file at it's maximum dpi and print it, regardless of print size.
Do you understand the difference between file resolution and printer resolution?

Certainly you have to print at as high a resolution in dpi as you can get but, the file need only be at 240ppi - 300ppi. There really is no connection between the resolution set in the printer dialog and the resolution of the file.

If you scan at 2400ppi without any scaling, you will get an image that is 4" x 5" @ 2400ppi. This is much too high a resolution to make any sense for a manipulation and you have effectively got an image that is 4" x 5".

I would guess that you are resizing to fit a certain size of paper but, do you resample the image at the same 2400ppi, or do you simply alter the resolution until the print is the size you want, without resampling?

When I prepped a couple of my exhibition prints for printing on a LightJet, the final file had to be only 300ppi, and that was perfectly good enough for a 30" x 24" edge-to-edge print.

[RJC]

Rob- could you please elaborate a bit on this? I'm not sure what the difference between source sharpening and content sharpening would be. I'm not familiar with these terms, and you're 3rd step is total greek to me.

Digital capture, whether a scan or using a digital camera, tends to soften the original image due to anti-aliasing filters, the application of ICE to remove dust and scratches etc. Source sharpening refers to a global (subtle) sharpening of the original image to counteract the afore-mentioned softening. This phase of sharpening is done using a mask to isolate the edges within the scene to reduce the possibilty of enhancing noise in the image.

'Content sharpening' refers to 2-phase sharpening based (1) on the frequency distribution of edges within the image (e.g. a portrait requires significantly different sharpening than a high-position view of a dense cityscape) followed by (2) 'artistic' sharpening determined by the photographer's artistic interpretation of the scene and which parts therein require relative sharpening for emphasis etc. This phase of sharpening is done on a blended layer in luminosity mode (similar to sharpening in the luminosity channel in LAB).

The 3rd stage is a global sharpening layer with a high pass filter with parameters determined by the output resolution of the printing device.

I always sharpen non-destructively on layers, and where necessary with edge-masks as previously mentioned. Bruce Fraser, from whom I picked up these techniques from his afore-mentioned book, is a renowned expert on Photoshop, and in conjunction with another expert, Martin Evening, developed the Pixel Genius PhotoKit Sharpener that employs thses techniques. At the time I read his and Martin Evening's books I could not afford to purchase their software so wrote my own set of Photoshop actions to implement the same sharpening techniques (all I had to do was to attempt to fully understand everything in their books !).:rolleyes:

Rob

[Frank Petronio]

Bruce was the expert but he passed away so don't go looking for his CS4 book....

You don't have to use sharpening layers, although they are nice, but piling up a stack of adjustments layers really adds to the file size and performance issues with larger files. The important point is that you need to save an "archive" or "master" file at your maximum useful resolution and you'll want to sharpen it just enough that aren't degrading anything in your image (usually you get subtle halos around the edges of the most contrasty transitions.) Once you save (and back up) your master archive file, then you can resize -- and adjust, usually with more sharpening -- various sized "output" files for different sized prints, reproduction, and internet use using the "Save As" and "Save for Web" File commands.

I save single layered archive/master files. It forces me to make decisions about the file. That is not necessarily a bad thing -- leaving every aspect infinitely adjustable means that it will be infinitely adjusted!

[RJC]

I agree entirely Frank. My master/archive file is almost always a single layer file saved after flattening the layers after source and content sharpening as a use-neutral master. I occasionaly need to save the sharpening layers but tend to avoid this if possible- however during editing sharpening layers are extremely useful to adjust the sharpening until completely satisified with the result rather than to keep re-opening the image and repeating the sharpening steps.

These are ideas culled from Bruce's book, I don't claim any originality in this regard, but I'll hang onto my copy for CS2 given the news that he has sadly passed away (I'd not heard this news before).

Rob

[Daniel_Buck]

are you scanning B&W film? Are you scanning it as greyscale? if you are, you want to convert it to RGB before you save it out as a .jpg for the web.

[Frank Petronio]

When you save as a jpg the greyscale automagically becomes RGB. Using Adobe's Save For Web command allows you to retain a color profile and metadata w the jpg, although people suggest that web jpgs should either use the sRGB profile or none at all (the Save For Web default strips the profile off).

[timbo10ca]

Do you understand the difference between file resolution and printer resolution?

Certainly you have to print at as high a resolution in dpi as you can get but, the file need only be at 240ppi - 300ppi. There really is no connection between the resolution set in the printer dialog and the resolution of the file.

If you scan at 2400ppi without any scaling, you will get an image that is 4" x 5" @ 2400ppi. This is much too high a resolution to make any sense for a manipulation and you have effectively got an image that is 4" x 5".

I would guess that you are resizing to fit a certain size of paper but, do you resample the image at the same 2400ppi, or do you simply alter the resolution until the print is the size you want, without resampling?

When I prepped a couple of my exhibition prints for printing on a LightJet, the final file had to be only 300ppi, and that was perfectly good enough for a 30" x 24" edge-to-edge print.

I know the difference between ppi of scan and dpi of output. I just have never altered my file ppi size prior to printing, other than one time and found the print resolution to be poor in comparison. I'll give you the scenario- I scanned a 35mm slide with a Nikon Coolscan VED at max resolution (4000ppi). File size was not a problem for speed in PS so I left it. I printed an 8x10 with the file document size resolution in PS set at 4000 and then at 300. I could see a degredation in print quality with the 300ppi setting. So now I print at the max resolution I scanned at. Now that I'm doing LF film, the files are too bulky to leave at max scan size for editing, so I reduce to "true" scanner resolution (I'm optimistic and use 2400ppi on my V750). Regardless of the print size, I leave the resolution at the max, and just change the print dimensions. What the software does then, I don't know...... Sometimes I use the Windows print tool, sometimes Photoshop.

Tim

[timbo10ca]

Bruce was the expert but he passed away so don't go looking for his CS4 book....

You don't have to use sharpening layers, although they are nice, but piling up a stack of adjustments layers really adds to the file size and performance issues with larger files. The important point is that you need to save an "archive" or "master" file at your maximum useful resolution and you'll want to sharpen it just enough that aren't degrading anything in your image (usually you get subtle halos around the edges of the most contrasty transitions.) Once you save (and back up) your master archive file, then you can resize -- and adjust, usually with more sharpening -- various sized "output" files for different sized prints, reproduction, and internet use using the "Save As" and "Save for Web" File commands.

I save single layered archive/master files. It forces me to make decisions about the file. That is not necessarily a bad thing -- leaving every aspect infinitely adjustable means that it will be infinitely adjusted!

I'm still trying to figure layers out for some of this stuff. So far, I can't seem how to use the clone stamp, dodge and burn, or sharpen in a layer- I do the action, but nothing happens on the picture.

Is your single layer master file pre or post editing- if it's post, that's exactly what I'm doing- creating a master file that is print ready, then reducing size of the ones I post..... If it's pre-editing, I can't see why anybody would want to waste the time re-editing a photo once it's been reduced to the size you plan to make that single print as. Either way, I'm confused as to why I'm being told that what I'm doing is the wrong way to do it :rolleyes: It seems to me that my workflow is exactly what you've just described. I do admit though, that my sharpening thus far lacks finesse. The methods Rob is describing sounds like what I'm trying to get my head around. Gotta find that book, or something on the net describing the process..... At least I now know about LAB- I just haven't used it yet.

Tim

[timbo10ca]

are you scanning B&W film? Are you scanning it as greyscale? if you are, you want to convert it to RGB before you save it out as a .jpg for the web.

Hey Daniel- I'm scanning color trannies and B&W negs. For B&W, I've been playing around with a couple different methods- scanning as greyscale 16 BIT, and scanning as RGB 48 BIT then using the Green channel (apparently the sharpest/best to use from reading around here and elsewhere) or using the Red channel (may seem to look better sometimes, but I may be fooling myself????). To be honest, I don't know if I can see much of a difference and scanning in RGB just seems to make bigger files. I've been pretty happy with the Greyscale files I've produced so far. Mind you, I haven't printed many of them yet. I run files through an action to reduce file size, convert to RGB, and save as jpg. I do not use "Save for Web".

Tim

[Daniel_Buck]

Hey Daniel- I'm scanning color trannies and B&W negs. For B&W, I've been playing around with a couple different methods- scanning as greyscale 16 BIT, and scanning as RGB 48 BIT then using the Green channel (apparently the sharpest/best to use from reading around here and elsewhere) or using the Red channel (may seem to look better sometimes, but I may be fooling myself????). To be honest, I don't know if I can see much of a difference and scanning in RGB just seems to make bigger files. I've been pretty happy with the Greyscale files I've produced so far. Mind you, I haven't printed many of them yet. I run files through an action to reduce file size, convert to RGB, and save as jpg. I do not use "Save for Web".

Tim

Yes, scanning RGB will be larger than B&W. 3 times larger, because you'll have 3 channels (R, G, and B, each channel with 16 bits, which equals 48 bits) instead of just one channel (grey, 16 bits in one channel)

[timbo10ca]

Yes, scanning RGB will be larger than B&W. 3 times larger, because you'll have 3 channels (R, G, and B, each channel with 16 bits, which equals 48 bits) instead of just one channel (grey, 16 bits in one channel)

Yeah- I realize that. I was doing it only to see if the bloat was worthwhile. I would open in PS, convert to greyscale using the G channel then go to mode and select greyscale to unload the bulk. This is the only way i could figure out how to "drop" the B and R channels, as I'd read to do....

Tim

[Gary L. Quay]

Thanks for all the responses. I just picked up a copy of Photoshop CS4. I had been using Elements 4.0. I looked into some of the plug-ins, such as Genuine Fractals, but I can't afford them just yet. I use the Save For Web function on Elements, and sharpening never seems to be helpful after that.

I put the photos right onto the post, and they show up as thumbnails. I've noticed that some images show up right in the post instead of in thumbnails. Are those the ones that come from other sites like Flickr or Facebook? The quality is usually very good on them. I'm still digesting the other info in the replies.

--Gary

[jb7]

I knew I was the only one who deciphered the question correctly...


j

[Gary L. Quay]

After reading your reply again, I'd say you were pretty close.

--Gary


I knew I was the only one who deciphered the question correctly...


j

[Gary L. Quay]

http://farm4.static.flickr.com/3385/3511179031_5b5c115cd9.jpg?v=0.jpg

Okay, I tried to get the address from the Flickr site, and put and at either end. This is what I got. I got the URL, not the photo. What did I do wrong? Is there a place on this site with instructions? I didn't find one... yet.

[jb7]

In that address, you'll notice that there's two 'jpg's

If you remove everything after the first 'jpg'
then use the tags with no spaces,
this appears-

a very cheeky picture...

http://farm4.static.flickr.com/3385/3511179031_5b5c115cd9.jpg

[Gary L. Quay]

Thanks!

--Gary

[Gene McCluney]

http://farm4.static.flickr.com/3385/3511179031_5b5c115cd9.jpg?v=0.jpg

Okay, I tried to get the address from the Flickr site, and put and at either end. This is what I got. I got the URL, not the photo. What did I do wrong? Is there a place on this site with instructions? I didn't find one... yet.

Do you use the brackets around the img? Like [img] ?

[Gary L. Quay]

Yes. at the beginning and at the end.

--Gary

[D. Bryant]

Another reason I sharpen my original image is because I've found it to give better results when spotting. On a number of occasions, I've spotted then sharpened, only to find new spots due to the sharpening. So I sharpen prior to spotting, which is now my final step before saving.

Tim

Spot first always prior to sharpening. If you are missing spots then roll back the sharpening and repair the spots as needed. Always use a non-destructive path when sharpening and preserve that as part of your archived file. Then final sharpen for output media and size. Don't apply ouput sharpening for archived images.

Don Bryant

[D. Bryant]

Joanna,
Tell me more about 96 dpi. Isn't 72 dpi standard?

72 dpi is a myth. To determine what dpi your monitor displays images divide the horizontal screen resoultion with the physical width of your display. That usually doesn't result in 72 dpi. IOW, 72 dpi isn't a standard.

Don Bryant

[Brian Ellis]

Joanna,
Tell me more about 96 dpi. Isn't 72 dpi standard?

72 hasn't been the standard for many years. Current monitors are around 100 - 110 or so I've been told by people who are supposed to know that kind of thing.

[Brian Ellis]

Could you tell me more about this bicubic sharper please? I've never used it, but it sounds interesting. I've always downsized using bicubic then applied USM to the degree I feel necessary. I like the control this allows me. Does Bicubic Sharper keep the exact same image sharpness/ ghosting/ etc of the pre-downsized image? This sounds like it could save me a step. . . .
Tim

Following is what I was taught while attending a Deke McLelland workshop a couple years ago. Bicubic Sharper is used as the name implies, to gain a small degree of additional sharpness. Since you presumably are starting with a photograph that's already been sharpened to your satisfaction, it should be unnecessary and may even be harmful to use Bicubic Sharper. Bicubic Smoother is used if the photograph has excessive noise. Plain old Bicubic is the one to use unless you have a specific problem you're trying to solve (i.e. you need a sharper image for some reason or the photograph has excessive noise). Never use Nearest Neighbor or Bicubic Linear, they're both useless. And ignore the editorial stuff that Photoshop says about using one method for enlarging and another for something else.

[Gary L. Quay]

72 hasn't been the standard for many years. Current monitors are around 100 - 110 or so I've been told by people who are supposed to know that kind of thing.

I tend to use 96 dpi. If it is a bit fuzzy at the largest size that Flickr will allow, that's OK. I don't want anyone to be able to copy one of my images and print it 8x10.

Thanks again for all of the help getting me online.

--Gary

[Brian Ellis]

"I'm still trying to figure layers out for some of this stuff. So far, I can't seem how to use the clone stamp, dodge and burn, or sharpen in a layer- I do the action, but nothing happens on the picture."

This is most likely caused by having the wrong layer active at the time you're using the tools.