next thread »

[Rolf]

Film has a built in sharpening by the chemistry on the contrast edges. The dark side gets a bit darker and the light side a bit lighter than the neighbours and so you get the effect of digital sharpening built in into the emulsion.

Digital images are softened by the process of rastering at a contrast border. You get grey in the pixels sitting on the edge due to averaging black and white. You can see that in the #127 video at the Fair Trade and Bio logo. Most consumer cameras are set to strong in camera sharpening for the JPG images. Some seem to be smart enough to soften skin and sharpen eyes, as I remember to have read.

You enlarged your images to 20x30cm or so max. Most people were limited to that size by the size of the baths - and money. Now we look from a distance of 30 cm at a 100% magnification and say: that is not sharp! The full image would be a meter or more in length.

Three reasons why we talk about it today - all perfectly ignorable. ;-)

 

[littletank]

@ Rolf: I do appreciate your sense of humour, keep it going. Your comment about the chemistry reminded me of what I used to do to get a contrasty, or was it a sharpened, image. I used Ilford FP4 film and developed for about 15 minutes @ 20C in a 1% dilution of Rodinol, This was claimed to give an enhanced edge effect.

Basically, then, what you are saying is that the lack of sharpness in digital images is a result of the process itself and not the photographer and, therefore, it becomes necessary for sharpness to be added.

As for size, exhibition prints were made from 35 mm film ( frequently not the whole negative)  enlarged to a maximum of 50 cm X 40 cm so lack of sharpness soon became obvious.

[monoceros84]

What did we do in the old days when sharpening an image was virtually unheard of? If the negative was not sharp we threw it away. What could cause the image to be unsharp? Generally either camera shake, poor focusing or poor optics. Now we have cameras with devices which counteract camera shake, we have automatic focusing we have good optics so why all this hype about sharpening the image?

Before someone leaps in to put me right I know the unsharp mask method of sharpening was developed in the darkroom as were many of the techniques we try to emulate digitally today. However, most amateurs then had either never heard of sharpening or did not have the facilities or the money needed to buy the equipment required.

I want to leap in anyway
Digital images are always unsharp. With exactly the same settings and environmental impacts a digital picture is more blurry than an analogue one. Because of the sensor pattern and the anti-aliasing filter in front of it. Thus, to get the same image with both analogue and digital you need to sharpen the digital one.

[di98jgu]

Digital images are always unsharp. With exactly the same settings and environmental impacts a digital picture is more blurry than an analogue one.

Which leads to the next most interesting steping stone, just how big is this difference and do we talk about black and white or colour negatives versus digital. 

But maybe that is one steping stone to much... 

[littletank]

This has the makings of an interesting discussion. It seems that firstly we must decide what we are comparing with the digital image and how the images were prepared. For analogue I presume there are B&W negative films and colour negative and reversal films and the processes involved in their manufacture and subsequent development. When factors such as film speed, chemicals, processing conditions and so on are taken into consideration we have many, many different images of the same subject to compare.

I do not know enough about the formation of digital images to start this side of the argument so I will leave that to those more knowledgeable than me.

[Dan]

Interpolation.
My most irritating image manipulation sharpness killer!
Often one rotational click, zapp! the whole image.
 

[rayadagio]

Thanks stepanekos for introducing this interesting topic!
@Rolf: High quality show as usual :-)

Here's a simple script for doing some parameter experiments. I could not resist;-)
->  <Image>/Filters/Enhance/Octave Sharpening

I followed Mathias' suggestion and added an option to set the layer mode.

Edit: New version 1.7 uploaded

[Rolf]

Interpolation.
My most irritating image manipulation sharpness killer!
Often one rotational click, zapp! the whole image.
 

Yes, a lot of the post processing introduces additional blur. Rotation, deformation, scaling...

Because of that sharpening should be the very last step in the chain.

[Rolf]

Thanks stepanekos for introducing this interesting topic!
@Rolf: High quality show as usual :-)

Here's a simple script for doing some parameter experiments. I could not resist;-)
->  <Image>/Filters/Enhance/Octave Sharpening

I followed Mathias' suggestion and added an option to set the layer mode.

Now I have to fake a script design session for the next show. ;-) Planned to do just that, but now I have a lot of research time saved.

[rayadagio]

Sorry for pre-empting you. I suffer from impulsiveness. ;-)

[rayadagio]

Sorry again!
There was a little bug concerning merging. It's fixed now. File above was replaced.

[stepanekos]

Thanks stepanekos for introducing this interesting topic!

My pleasure! And thanks for the script. I will try it soon.

[rayadagio]

One question: Why is it recommended to set the threshold of USM to 0? This eliminates the crucial advantage of unsharpened masking, doesn't it? I assume noise could be reduced if the threshold is set >0. What do you think?

[jgsack]

Don't know about the last question, but I would like to know more about the practical effects of threshold, and ideas on where it might be used to advantage.

Re your script: thanks, it permits quick experimentation and comparison. I found that I had to change the layer name if I wanted to keep multiple results around, else the merge layers would do funny things. That's no problem, I just prefix your nice self-documenting names with my initial. [Correction: the difficulty is when there are layers above the active layer. The fix is to move the active layer to the top of the layer stack.]

Also, I changed my copy of your script to use float for the amount, to mimic the basic USM filter settings.

I find that I prefer quite a bit more subtle sharpening than perhaps some people. I might use 0.1 radius and 1.2 amount, for example. There are cases where an even smaller radius looks better to me.

I like to open a couple of extra views with progressively smaller windows but larger magnification, when doing comparisons.

~jim

PS: I discovered console messages "DeprecationWarning: integer argument expected, got float" on lines 39 & 42, so I inserted a line nlayers = int (nlayers) at the top of the function to fix that. I suppose args must be passed-in as floats?

~j

[monoceros84]

Don't know about the last question, but I would like to know more about the practical effects of threshold, and ideas on where it might be used to advantage.

Up to now I only used threshold to prevent noise from being sharpened. Just as Bert said.

I find that I prefer quite a bit more subtle sharpening than perhaps some people. I might use 0.1 radius and 1.2 amount, for example. There are cases where an even smaller radius looks better to me.

A low radius doesn't help for subtle sharpening. It's just the amount and the opacity of the sharpened layer. A low radius can result in really bad images in the wrong pictures. If it's too small, it might even become more blurry. Then comes the region where just noise is sharpened and only above that the real edge sharpening begins to break through. You should always drag the amount to maximum, set the right radius and reduce the amount again.