Resizing a cropped image..

[bnnrcn]

Does resizing a cropped image decrease the sharpness?As far as I understand so far,yes.So what are the limits of cropping or resizing an image to get an acceptable quality.I started a thread today and my image doesn't seem as sharp in CIC as it is on my computer's screen

This is my thread with less sharp image
Mine is a very little ant:)

[Geoff F]

Cropping an image reduces the size; so if you want to return the cropped image to the original dimensions, which means there is some upward resizing, then a little softness may develop. And is likely to be worse if there is a substantial resize. So a little extra sharpening may be desirable.

But with modern image sizes, it is often possible to crop and still find that the cropped size is sufficient for the intended purpose. In which case there is no softening.

However, after doing a substantial resize for internet use, for example, I often use a little Unsharp Mask sharpening before uploading. How much to use is always a matter of conjecture; but I tend to go with something around: amount 50 radius 0.5 and threshold 1. That is just a suggested starting point though.

The amount of resizing and method of uploading can also cause problems. When using the Tinypics option, which is usually simplest and best, I work around 1,000 to 1,200 pixels on the long edge and 96 ppi although the ppi is automatically adjustable anyway. And a Jpeg compression of 80% quality (level 10).

Going too large in image size can result in auto resizing during uploading which may cause some softness. For example, the CinC Gallery method has a smaller maximum size; which I think is around 700 pixels and 150 kb. Anything over this limit is auto resized.

[Manfred M]

Hi Binnur - there is no straight forward answer to your question, because there are a number of variables involved. Resizing will ALWAYS impact sharpening, but it will not necessarily decrease sharpness. It is all a matter of degree. The important thing is to output sharpen at 100% of your final image size, regardless of whether you upsample or downsample the image. Upsampling means that data has to be created, and this will tend to result in overall softening. Downsampling will have the opposite effect as data is thrown away, and often a sharper image results.

Unless your image is a perfect match for the display size, it will always be upsampled or downsampled to match the display size. In general, digital camera have far more resolution than a computer display, so downsampling will virtually aways occur when you display an image. Unless you crop very drastically, so that the crop is less than the resolution of your display, it will still have to be downsampled.

The problem with images on the internet is that they are subject to all kinds of processing that you have little or no control of. Depending on the host you are using for the files that you are linking to at CiC, other processing will occur. Files are often compressed by the hosting service, in order to save storage space. A highly compressed file will show a loss of detail, so this could be your issue. When I download your image, the metadata shows that it is only 639 x 578 pixels, so it is quite small, so I assume a lot of data has been thrown away, and viewing it at a larger size will certainly show a high level of softness; and I suspect this is where the bulk of the softening occurs..

[bnnrcn]

Thank you Manfred,my camera shoots images with dimensions 4272x2848 and after cropping the dimensions for the image was 947x856 .When uploading image to CIC I chose 'message board option' and you say that metadata shows 639x578 (in message board option that I chose dimensions were 640x480 so I really don't understand how I got 639x578)what I understand from your answer my image is less sharp in CIC because of losing some more pixels.

I'm also still confused about getting a sharper image with downsampling because you say that data is thrown away when you downsample an image

Hi Binnur - there is no straight forward answer to your question, because there are a number of variables involved. Resizing will ALWAYS impact sharpening, but it will not necessarily decrease sharpness. It is all a matter of degree. The important thing is to output sharpen at 100% of your final image size, regardless of whether you upsample or downsample the image. Upsampling means that data has to be created, and this will tend to result in overall softening. Downsampling will have the opposite effect as data is thrown away, and often a sharper image results.

Unless your image is a perfect match for the display size, it will always be upsampled or downsampled to match the display size. In general, digital camera have far more resolution than a computer display, so downsampling will virtually aways occur when you display an image. Unless you crop very drastically, so that the crop is less than the resolution of your display, it will still have to be downsampled.

The problem with images on the internet is that they are subject to all kinds of processing that you have little or no control of. Depending on the host you are using for the files that you are linking to at CiC, other processing will occur. Files are often compressed by the hosting service, in order to save storage space. A highly compressed file will show a loss of detail, so this could be your issue. When I download your image, the metadata shows that it is only 639 x 578 pixels, so it is quite small, so I assume a lot of data has been thrown away, and viewing it at a larger size will certainly show a high level of softness; and I suspect this is where the bulk of the softening occurs..

[Manfred M]

Binnur- what I (and others) have seen is that the down-sampling algorithms seem to enhance areas of contrast. Ultimately, in down-sampling, some pixels have to be thrown away, as there is no other way of creating a smaller image. Fewer pixels are required for a smaller image and the computer screen has a fixed number of pixels, so the only way to create a smaller image is to physically used less pixels.

If there are areas of similar colours, some of these are "thrown away", so at the boundaries, where lighter and darker colours meet, these appear to be more defined. The results in the image looking “sharper”.

This is the same way that sharpening works; the algorithms look for areas of high contrast, and at the point the light colours and dark colours meet, the light pixels are lightened and the dark pixels are darkened.

As for the actual number of pixels, all I can do is to report the dimensions reported by my computer; in my case I looked at the image parameters as reported by Bridge CC.

[bnnrcn]

Manfred-What I understand is that 1- I don't normally get a softer photo by downsampling and losing some pixels but on the contrary I might get a sharper image if I do this operation on my computer 2-I might get a softer photo while uploading to CIC by choosing one of the upload options with smaller dimensions ,because files are compressed by the hosting service and losing pixels in this way may cause a softer image. Is this what you mean?

Binnur- what I (and others) have seen is that the down-sampling algorithms seem to enhance areas of contrast. Ultimately, in down-sampling, some pixels have to be thrown away, as there is no other way of creating a smaller image. Fewer pixels are required for a smaller image and the computer screen has a fixed number of pixels, so the only way to create a smaller image is to physically used less pixels.

If there are areas of similar colours, some of these are "thrown away", so at the boundaries, where lighter and darker colours meet, these appear to be more defined. The results in the image looking “sharper”.

This is the same way that sharpening works; the algorithms look for areas of high contrast, and at the point the light colours and dark colours meet, the light pixels are lightened and the dark pixels are darkened.

As for the actual number of pixels, all I can do is to report the dimensions reported by my computer; in my case I looked at the image parameters as reported by Bridge CC.

[bnnrcn]

Hi Geoff,do you mean auto resizing in CIC may cause some softness and it is better to resize my images on my computer before uploading?

Cropping an image reduces the size; so if you want to return the cropped image to the original dimensions, which means there is some upward resizing, then a little softness may develop. And is likely to be worse if there is a substantial resize. So a little extra sharpening may be desirable.

But with modern image sizes, it is often possible to crop and still find that the cropped size is sufficient for the intended purpose. In which case there is no softening.

However, after doing a substantial resize for internet use, for example, I often use a little Unsharp Mask sharpening before uploading. How much to use is always a matter of conjecture; but I tend to go with something around: amount 50 radius 0.5 and threshold 1. That is just a suggested starting point though.

The amount of resizing and method of uploading can also cause problems. When using the Tinypics option, which is usually simplest and best, I work around 1,000 to 1,200 pixels on the long edge and 96 ppi although the ppi is automatically adjustable anyway. And a Jpeg compression of 80% quality (level 10).

Going too large in image size can result in auto resizing during uploading which may cause some softness. For example, the CinC Gallery method has a smaller maximum size; which I think is around 700 pixels and 150 kb. Anything over this limit is auto resized.

[rpcrowe]

I have been using Perfect Resize (part of the Perfect Photo Suite) to resize my images. Since I often need to work with very small and very bad images of our rescue dogs, sent to me by our volunteer foster families, I investigated the Genuine Fractals resizing program which was once a standard of the industry.

I learned that Genuine Fractals had been purchased by OnOne and incorporated into their Perfect Photo Suite as Perfect Resize. I found that I could buy the Genuine Resize alone or purchase it as a part of the larger Perfect Photo Suite at not too much more money that the Perfect Resize alone.

OnOne had a sale for photographers that were using NIK Software and that puit the icing on the cake for me and I purchased the Perfect Photo Suite.

The Perfect Resize is a very simple program which has a lot of customization built into it. The basic workspace looks like this.



You can adjust the size and PPI for various types of end products including: monitor viewing and various types oof printing, including for Canon, HP or Epson printers.

The settings include those for low res JPEG, Portrait, Landscape and high detail.

You can customize any of these choices.

I am quite happy with Perfect Resize and I also like the control I have with various other portions of the Perfect Photo Suite including. There are some duplication of capabilities with NIK Software but, these two programs enhance each other.

The Perfect Photo Suite can be used as a Plug-in with Photoshop, Photoshop Elements and with Lightroom (it provides layer capabilities in Lightroom). It can also be used as a stand alone editing program. The combination of Lightroom and Perfect Photo Suite would (IMO) be quite a good full service editing program at a price greatly reduced from Photoshop.

Perfect Resize: https://www.youtube.com/watch?v=S9hdJXahuSw
Perfect Photo Suite: https://www.youtube.com/watch?v=oOyxPeF88WQ

Perfect Mask is a super easy way to mask...
Perfect Mask: https://www.youtube.com/watch?v=IIj-BHENt9Y

[Manfred M]

Richard - a few years ago I would have done exactly what you have recommended, and in fact bought the OnOne suite precisely for the same reason as you did, improved resizing of images.

Much to my chagrin, I found I wasted my money. At one point, a number of years ago, Photoshop's resize engine was not particularly good and the only way to get a "clean" resize was to enlarge the original image, no more than 10% at a time up to 3 or if one was pushing things, 4 times before the quality degraded too mych.

Somewhere around CS4 or CS5, Adobe seems to have re-written that algorithm and the current version CS6 and CC) is every bit as good as Perfect Resize. I did some extensive testing and actually did a pixel by pixel overlay using both approaches and found no appreciable difference in the results from either software package. Save yourself some time and just use Photoshop; the quality is there.

[Manfred M]

Manfred-What I understand is that 1- I don't normally get a softer photo by downsampling and losing some pixels but on the contrary I might get a sharper image if I do this operation on my computer 2-I might get a softer photo while uploading to CIC by choosing one of the upload options with smaller dimensions ,because files are compressed by the hosting service and losing pixels in this way may cause a softer image. Is this what you mean?

Yes

[bnnrcn]

Thank you Manfred,If I resize an image on my computer instead of letting CIC autoresize it,will my image file still be compressed by the hosting service?Does hosting servive compress each file even if the files are already in smaller sizes?

Yes

[Manfred M]

Thank you Manfred,If I resize an image on my computer instead of letting CIC autoresize it,will my image file still be compressed by the hosting service?Does hosting servive compress each file even if the files are already in smaller sizes?

It depends on the hosting service, and unfortunately, none of the ones I know tell you what is happening behind the scenes, so it is very difficult to predict outcomes. What I would suggest you do is play around until you get what you want, and if then the service is not giving you the quality you want, change services.

[bnnrcn]

Thank you Manfred,do unsharp mask and output sharpening mean the same thing?Geoff mentioned unsharp mask in his post and you mentioned output sharpening,so which one I have to use after resizing?

It depends on the hosting service, and unfortunately, none of the ones I know tell you what is happening behind the scenes, so it is very difficult to predict outcomes. What I would suggest you do is play around until you get what you want, and if then the service is not giving you the quality you want, change services.

[Manfred M]

The unsharp mask is a technique for sharpening. Output sharpening is a stage of sharpening.

There are three levels of sharpening that are discussed whenever people write about sharpening:

1. Input sharpening - a small amount of sharpening applied at RAW conversion to counteract the effects of your camera's anti-aliasing (AA) filter;

2. In process sharpening, where localized sharpening is applied to areas of the image that need it; and

3. Output sharpening - adjusting image sharpness based on the size you will be displaying the image at. I will finish processing my image and save it and will do the output sharpening prior to posting it to a website or to printing. Both of these operations will have a different amount of sharpening, because the image size varies so much.

I find output sharpening for print easier that one for posting because there are less variable to consider (after all, the final size of the print is known; different sizes of prints each get their own output sharpening done).

When it comes to sharpening for display, the size of the screen the image will be viewed on comes into play, as well as the amount of compression. An image that looks fine on my large computer screen will look oversharpened on my tablet.

[bnnrcn]

Thank you Manfred,when do you do resizing and output sharpening,before saving as JPG or after saving as JPG?It sounds sensible to me that it is better to do after saving as JPG because when an image is saved as JPG some information is lost and it is better to sharpen the image again.What do you think?

Btw I have found out why my image's dimensions were 639x578 in the metadata although I chose 640x480 in CIC's uploading screen.My real dimensions (I mean after cropping) were 947x856 and CIC's system keeps the ratio between length and height while downsizing (947/856 = 639/578)

The unsharp mask is a technique for sharpening. Output sharpening is a stage of sharpening.

There are three levels of sharpening that are discussed whenever people write about sharpening:

1. Input sharpening - a small amount of sharpening applied at RAW conversion to counteract the effects of your camera's anti-aliasing (AA) filter;

2. In process sharpening, where localized sharpening is applied to areas of the image that need it; and

3. Output sharpening - adjusting image sharpness based on the size you will be displaying the image at. I will finish processing my image and save it and will do the output sharpening prior to posting it to a website or to printing. Both of these operations will have a different amount of sharpening, because the image size varies so much.

I find output sharpening for print easier that one for posting because there are less variable to consider (after all, the final size of the print is known; different sizes of prints each get their own output sharpening done).

When it comes to sharpening for display, the size of the screen the image will be viewed on comes into play, as well as the amount of compression. An image that looks fine on my large computer screen will look oversharpened on my tablet.

[Manfred M]

My workflow is as follows when posting to the internet:

1. Edit the image (usually using the RAW data).

2. Convert from the working colour space (usually ProPhoto or AdobeRGB to sRGB)

3. Resize the image to final size;

4. Do output sharpening (looking at the image at 100%; after downsizing the image usually fits on the screen);

5. Convert to jpeg.

I have found that so long as I stick with a higher quality setting (usually 10 to 12 setting in Photoshop, although at times I will go as low as 8) , I don't see any appreciable quality difference between the sharpness as long as I do this.

Sharpening a jpeg will give you two different levels of compression; once when you create the original jpeg and once when you resharpen, so in theory, you are applying data loss twice. I don't know if there is a practical visible difference, as I have never worked this way.

[bnnrcn]

I'm clear about the matter now,thank you very much Manfred.Btw my photoshop book arrived 2 days ago and I downloaded some files with the code given in the book.Now I can start learning with the help of the book

[Geoff F]

This thread has quickly moved on and covered some potentially tricky subjects, which can be difficult to understand, but everything seems to have been explained well.

Binnur. In answer to your earlier question about what actual sizes to use for resizing.

I always find it is best to find out the required dimensions and file sizes beforehand then resize to exactly those sizes on my computer before attempting to upload.

But make sure you are checking the correct method. For example, the Tinypics option, which for many people will be the best method, allows for substantially larger files than, for example, using the CinC Album method.

When selecting a downwards resize interpolation option it is often recommended to use Bicubic Sharpener (other names in different software) which automatically sharpens the image when resizing. But I have found this to give a somewhat crude sharpening effect which often isn't to my liking.

So I simply use the basic Bicubic Interpolation method then add a little sharpening to suit my tastes afterwards - but only if needed!

[bnnrcn]

Thank you very much Geoff,I don't know very much about resizing and sharpening techniques because I'm new into PP but I will learn with the help of the photoshop book (I have subscribed for Photoshop CC very recently) and I will keep in mind your comments

This thread has quickly moved on and covered some potentially tricky subjects, which can be difficult to understand, but everything seems to have been explained well.

Binnur. In answer to your earlier question about what actual sizes to use for resizing.

I always find it is best to find out the required dimensions and file sizes beforehand then resize to exactly those sizes on my computer before attempting to upload.

But make sure you are checking the correct method. For example, the Tinypics option, which for many people will be the best method, allows for substantially larger files than, for example, using the CinC Album method.

When selecting a downwards resize interpolation option it is often recommended to use Bicubic Sharpener (other names in different software) which automatically sharpens the image when resizing. But I have found this to give a somewhat crude sharpening effect which often isn't to my liking.

So I simply use the basic Bicubic Interpolation method then add a little sharpening to suit my tastes afterwards - but only if needed!

[bnnrcn]

Hi Manfred,while converting working color space from Adobe RGB to sRGB,there are so many sRGBs such as sRGB IEC61966-2.1,wscRGB,wsRGB,e-sRGB.I'm confused about which one to choose I might as well choose sRGB IEC61966-2.1 but I want to make sure.(BTW this is in ACR part,while downsizing and saving.)

My workflow is as follows when posting to the internet:

1. Edit the image (usually using the RAW data).

2. Convert from the working colour space (usually ProPhoto or AdobeRGB to sRGB)

3. Resize the image to final size;

4. Do output sharpening (looking at the image at 100%; after downsizing the image usually fits on the screen);

5. Convert to jpeg.

I have found that so long as I stick with a higher quality setting (usually 10 to 12 setting in Photoshop, although at times I will go as low as 8) , I don't see any appreciable quality difference between the sharpness as long as I do this.

Sharpening a jpeg will give you two different levels of compression; once when you create the original jpeg and once when you resharpen, so in theory, you are applying data loss twice. I don't know if there is a practical visible difference, as I have never worked this way.