Quality of pictures

[Marianne]

Hi,

I am still struggling with the quality of my show. Like Al advised me I compressed the jpegs with a factor of 5. Seems to the eye the same as a compression of 8 so that is no problem. But my pictures still are about 150 kb in size.

But now I'm trying to get them as sharp as possible on screen without looking unnatural. I am talking here about scanned images, some of them are in medium quality for a start because they are several dozens years old. The oldest picture among them is of my great great grandmother and is made around 1880. Now I'm not looking to get that one perfect on screen, but as they are scanned out of an album which I don't posess, and this show will be part of an archive cd for my whole family, it should look good.

What can you say about sharpening? I have been experimenting with a program recommended earlier in this forum : SharpControl 1.1 by Vtie, but that one seems to make faces a little bit too smooth. I also tried to use Photoshops Unsharp Mask Settings: Amount 100%, Radius 1, Threshold 0. This seems to work although it gives a lot more noise. But maybe that is not too bad for old pictures.

By the way they are not totally black and white as I find that is getting the authenticity down.

So what would you recommend for this show?

And what do you use for your digital images? As my next show is definitely coming out of my own digital camera

Marianne

[Gérard de Lux]

I also tried to use Photoshops Unsharp Mask Settings: Amount 100%, Radius 1, Threshold 0.

Unsharp Mask settings are quite difficult to adjust and depend of the size of the picture because they are in %. I've found that for pictures of about 1000 px wide, the following settings work fairly well : Amount between 150 and 200 % (usually 180); radius 0.3-0.4; threshold 3

I never "sharpen" original images (whether from digital camera or scanned film) but always the final ones, once they are at the final size and after retouching, just before saving.

[Alan Lyons]

Marianne,

Some of the problems with scanned images, in fact all digitized images comes from "noise" caused by the way the image is collected and rendered in digital format. Try www.neatimage.com this programme is designed to deal with this problem by examinine a single colour section of your image. Once you have found a good result you can save a job for a given scanner/ camera and pass all images through this first before going to your image editor. The free version allows you to do one image at a time but if you buy you can batch treat. Have a look at the free before buying,

Regards,

Alan

[JRR]

Marianne:

I have a little chart in front of me that I use when sharpening in PS (Unsharp mask)

"normal" image

75-100%

radius 1-1.5

threshold 4-8

Detail image

200-400%

radius 0.4-0.6

threshold 4-8

smooth image

50-100%

radius 2-3

threshold 4-8

Those numbers give me a starting point with each type of image.

I do get noise the odd time, if the background plain, then I hit it with the FILTER > NOISE > DESPECKLE, but then you lose some sharpness. It is a trade-off I find.

[Guest guru]

Marianne, try also DCEnhancer (Digital Camera Enhancer). It's free, not difficult to use, and in my opinion very effective in several cases.

[think(box)]

Marianne, I would like to second Alan's suggestion to use Neat Image from A B Soft. It beats DCE AutoEnhance hands down. I have used both extensively. DCE AutoEnhance is well made, but it doesn't reduce image noise as effectively as Neat Image. The Neat Image free version and paid version both reduce noise as well. The paid version allows batch operation.

What Neat Image does is to take a noise sample from your selection of a small area that has just uniform shading and color, with no subject or shadow edges. It then processes your photo to reduce noise with unbelievably good results. You do not sacrifice sharpness as you do with smoothing algorithms like despeckle, reduce graininess, etc.

The batch operation is useful if you want to reduce noise in many images from a single source. E.g. the noise profile is always the same from the same digital camera or scanner and source. You sample noise once manually and process all in batch. There are exceptions when a digital camera is auto-adjusting the effective "film speed" or image pickup gain, as noise varies. Likewise, when you are scanning photos from many sources each photo needs individual attention.

I have found that you should reduce noise in the original before processing, but that you can even do it again after a Photoshop unsharp mask usage with very good results. You don't give up sharpness. A photo with detailed subject and clear blue sky with clouds - and noise - simply loses the noise.

[Maureen]

Hi

I always work on my images in Photoshop using a copy above the original (so I can sneak a look at the original if I wish) and then use adjustment layers and layer masks.

When I feel happy with the final image - I always sharpen last (using unsharp filter). However I first create a new layer and then use the alt merge visable option . Thus I put the unsharp filter onto the whole image but have all options saved below.

I use the save to web rather than the save as jpeg, as this gives me the opportunity to view the quality of the saved image and slide along to get the size under 200kb with the best quality. I believe I was told once that Photoshop produces a better 'algorithm' for the save to web than the jpeg save setting so you should get a better quality saving with your compression..(might have been Adobe themselves when questioned about why have the two save setting & what was the difference & which was best to use) .........Have I got that right?

Perhaps one of our wonderful technical experts could answer?

If I find when I insert the new image into PTE it looks oversharpened when projected, it is quick and easy to go back to my saved file (which has all its layers ) delete the top one ( alt merge visable, plus unsharpened) create a new layer repeat the alt merge visable and then put an unsharpen filter on this new final version with different settings. Like JRR I usually find my settings vary with the type of image. So it is difficult to recommend settings for unsharpening.

Hope this helps.

Best wishes

Maureen

[alrobin]

I  use the save to web  rather than the save as jpeg,   ...........snip..........

I believe I was told once that Photoshop produces a better 'algorithm'   for the save to web than the jpeg save setting so you should get a better  quality saving with your compression..(might have been Adobe themselves when questioned about why have the two  save setting & what was the difference & which was best to use)  .........Have I got that right?  

Hi, Maureen,

I, too, use "save for web", mainly for faster action, and, as you mentioned, more options.

The main (IMHO) significance of "save for web" is that it saves the image in the sRGB colour profile which has become the "standard" for distributing photos (mainly by web, but this would also apply to distribution of slide shows for others to watch). The reason for this is that apparently sRGB is the profile used on the majority of pc monitors.

Many serious Photoshop users, however, prefer the Adobe RGB "working space" (or some other profile for specific needs), as it has a wider selection of colours. As such, it is the preferred mode if you are planning to convert to CMYK for pre-press applications.

If you are already working in sRGB mode, you can achieve optimum jpeg compression by choosing jpeg save option "baseline optimized". This option is made available particularly for obtaining optimum compression for web applications. It is different from ordinary "baseline" mode in that it compresses by 5 to 10 % more than ordinary "baseline". I do not know if it also converts to sRGB, but I do not believe that it does. However, it seems to create files about the same size as "save for web" in most cases.

One important fact to consider in all this: in Photoshop, in "save for web", the jpeg compression settings are measured as percentages (i.e. values with a maximum of 100). However, in "save as", jpeg compression ratios are based on a maximum of 12. So, a "save as" value of 7 will compress more than a "save for web" value of 70.

Hope this sheds a little light on a very complicated topic.

[Guest guru]

What you said sheds a clear light on this topic, Al. Thanks!

As to sharpening software, don't forget Ultra-Sharpen Plug-in, one of the first and most glorious tools of this type, that has reached now v6. Its action is alike Photoshop Unsharp Filter, but there are not visible artifacts, and the noise doesn't increase in a significant degree. There is a simple but effective free version, and a more versatile "Pro" shareware ($20, 25 on CD).

You can see some interesting samples here.

[alrobin]

Marianne and others,

A while ago I discovered this "piece de resistance" of "sharpening" in Photoshop. It was covered in a tutorial, the author of which I no longer remember, but it is so effective I have to share it with you. It sounds complicated at first, but after a couple of trys you should find it makes a lot of sense. Here goes (I'll try to keep it simple):

1. Make a "copy" layer of your image.

2. Open the "unsharp mask" filter on the copy.

3. Select a reasonable set of parameters, depending on the size of the image, etc.

4. Close the filter and make a copy of the corrected layer.

5. Select the first sharpened layer (copy #1) and change the option from "normal" to "darken". Set the opacity at something like 75 to 80%, whichever looks good.

6. Select the second sharpened layer (copy #2) and change the option from "normal" to "lighten". Set the opacity at around 25 to 30%.

7. Play around with the opacity settings to obtain the best-looking effect.

This procedure, if you had to go through all these steps every time, would be too much work for most every-day images, but the handy thing about it all is that you can set up an "action" to do it all automatically whenever you press the assigned function key.

Just a couple more notes on sharpening:

1. If you use "edge sharpening" instead of "unsharp mask" you can reduce the artifacts and noise (unless, of course, your image is abnormally grainy or noisy to start with). You give up all control over edge thickness, etc. however.

2. My "Photoshop Bible", by Deke McClelland, tells me I should always use a setting of "0" for the "threshold" parameter. To use his words: "I urge young and old to observe Threshold with the utmost scorn and rancor." He uses an "edge mask" to deal with the issue of Threshold. This procedure is even more complex, so I won't go into it any further, unless someone wishes to have the details, in which case you should contact me by email as I don't want to take up any more time on it in the Forum.

Hope this adds to the body of useful knowledge on "sharpening".

Edit: The author of the tutorial mentioned above was Russell Brown, the "Photoshop Guru".