Image resizing Loss of Quality

[timur]

Is it inevitable that there is loss of quality when resizing a large picture to a smaller one? I am a novice in photo editing...

I have read on this website couple of posts about this and followed there instructions but yet still I have loss of quality.

I have tried exporting the image and tried all the options on the pop-up box (yes including lanczos).

I have tried rasterizing also...

Is this a affinity issue          or        in general is it normal??

 

EDITED

the pictures is 5847x2535

I am trying to get it to 959x (whatever the equivalent of that is)

 

The picture is degraded so bad that I don't want to post it on the website anymore.

 

My question: will there be degregation when decreasing size of picture? 

I know R C-R has already confirmed this but I would like to have a few more, to be certain.

 

ANSWER

By pressing 1 + cmd you zoom to the actual size. 

This allowed me to figure out my source of confusion, my backend provider for my website is misinforming me. I have full width of screen picture on the website and on the backend it is requesting me to input 950 pixel wide picture which is not the actual size needed.

If you still have issue read the posts below.

[DWright]

Hi timur,

 

Welcome to the forum.

 

Can you please provide me with further details on the original photo format and dimensions and the resize dimensions and the format that you are exporting to so that I can investigate further.

 

Regards,

Darren

[deborah]

I would like to add to Timur's question. I am new to Affinity Photo and am also having problems with resampling. In Photoshop, when resampling, width, height and resolution were all linked. If I reduced the width of an image, the height would reduce proportionally and the resolution would increase in proportion. This is not happening in AP and I have followed the exact steps shown in the tutorial on resizing images. It is also frustrating that when I change a measurement or the resolution, not only do I not see a proportional change in the other properties, the window just closes when I click to apply the change and I see NOTHING! Is there something we are doing wrong or is this a shortcoming in AP?

[R C-R]

Any time you reduce the size of an image by resampling it, the number of pixels in the image will be reduced, causing its quality to be degraded. This has nothing to do with Affinity or the resampling method used, although the latter can minimize the visible effects, as does the content of the image -- purely rectilinear elements will show fewer jagged edges & look less burry than ones with diagonal elements when resampled.

 

If you are not resampling (the box is not checked) then you can only change the dpi, which will affect only printed output & related items like the ruler units. The Understanding DPI video tutorial explains this in more detail.

[timur]

I have put the dimensions of the picture on the original post on top. 

@R C-R: hmm I was informed (maybe misinformed) that on photoshop the quality of picture is degraded only if the picture is increased in size and not decreased.

[Oval]

Think, you only have to resharpen the small picture. But without knowing the picture it is difficult to know what you mean by “loss of quality” and “degraded so bad”.

 

 

(Yes, you have been totally misinformed. In most cases you lose sharpness, information etc.)

[accendare]

I too am having issues-taking a high res digital image , 360dpi 8x10 inch , want to make a web ready 72 dpi 6x8 inch image,

no matter what what re-sample choice it makes the image blurry....there should not be such a horrible quality , as a long time photoshop user and teacher (24 years) I have never had an issue making web images from higher res images. If this program cannot export a simple image and maintain image integrity then I do not see much hope for this program in the future. yes I understand that there will be some loss of the image resolution but it seems way to much in affinity. If I put photoshop and affinity  resize image side by side there is a visible difference in quality.

[R C-R]

timur,

 

Decreasing the size of an image is the same as reducing the number of pixels in it. Since a pixel is the smallest unit of color detail an image can have (each pixel can be only one color), when you reduce the number of pixels in an image, you unavoidably reduce the maximum amount of color detail it can contain.

 

Consider a hypothetical 10x10 px image. If you reduce its size by 10%, converting it to a 9x9 px image, you have reduced its 100 total pixels to just 81 pixels, reducing the maximum color detail it can contain to 81% of the original. If there was not much pixel to pixel color variation to begin with, the loss of detail might be unnoticeable, or almost so, but if there is then it will be.

[MEB]

Hi accendare,

Welcome to Affinity Forums :)

How are you resizing the document? Are you using the Crop Tool or Resize Document from the menu?

 

Hi timur,

Welcome to Affinity Forums :)

When you scale down an image there's always some loss of quality (sharpness) which you need to compensate with some unsharpening or with a high pass filter.

It's not as serious as when you enlarge images but it's still noticeable. This is expected and generally happens with any software. There may be some differences between them mostly due to the algorithms used to down sample the images.

[Oval]

there should not be such a horrible quality

 

Cannot believe that horrible quality. AP makes a great job. Never had such an issue. The workflow is not the same in PS, so perhaps you missed a step in AP. Please upload the image (or just a little stripe) and we can show you how to get the quality you need.

[R C-R]

I too am having issues-taking a high res digital image , 360dpi 8x10 inch , want to make a web ready 72 dpi 6x8 inch image ...

If you are changing from 8x10 inch to 6x8 inch (whatever the dpi, which as I'm sure you know is not relevant to a web page display) you must be changing the aspect ratio of the image. How are you doing that?

[Dale]

Another way to look at these resizing issues... for starters I usually throw away DPI, ignore it, just think about the pixels. After all, a 600 x 300 pixel image has the exact same 180,000 pixels whether it's saved at 72 or 1200 DPI. So how many pixels do you start with and how many do you want to end up with?

 

An image "10x8 inches at 360 DPI" is 3600 pixels by 2880. In camera-friendly terms that's a 10.4 megapixel image. Resampling to make the image "6x4 inches at 72 DPI" would make it 432 pixels by 288. That's a 0.12 megapixel image, over 80 times smaller than the original. When you say it looks blurry, where is that? On a web page mockup on a laptop? On a 5K retina iMac webpage mockup? Or when scaled to fit the workspace area in Affinity Photo? Those 0.12 megapixels won't look good except when viewed taking up a small space.

 

A website won't usually be able to display the same 432 x 288 image at 6"x4" on all screens, and if it did the appearance would vary wildly. Visitors' screens (and browser settings) are all different, so it's usually necessary to make images at multiple sizes for retina, regular, tablet and mobile visitors. For instance, variants of a small image should vary from 480 pixels wide to 960 to 1440 to look good on different iPhones (with regular, x2 retina, and x3 retina screens).

 

There's always going to be reduced quality when resampling images though, the algorithms cannot usually create detail, but they can try not to throw too much away when redefining the image at a new size. And that first tip of ignoring DPI and focussing on pixels can help with some work, maybe not all, but it can help — DPI sometimes adds complexity just for the sake of having been around for a long time. It can be like a confusing "bank compound interest rate" middle-man in resampling, whereas original and desired pixel dimensions are like the much more obvious "loan amount" and "amount to pay back" monetary values. You know where you stand with pixels.

[R C-R]

In my opinion, "just think about the pixels" is good advice for anybody concerned with preserving image quality. Even if you are printing, you still have to include enough pixels in the image to preserve every detail you want the print to include. If you are not printing, DPI is irrelevant; if you are it sets the print's dimensions but not the amount of detail it contains.

[accendare]

First-thanks for everyone who responds -so here the image quality issue,- using a different image but same resultst

 

the photo is a RGB  2336 X 3505 pixels at 72 DPI

 

I go to resize document and change the size to 144 x 216 pixels-at the same DPI  (72) and the image comes out pixelated using lanzocs 3 (both options) - now how should making the image smaller have such a reduction in quality-and this is before exporting as a jpeg.

 

the purpose of the small images (170 x170 px)  for thumbnails on a website

[MEB]

Hi accendare,

What's probably happening is that after you resize your image, Affinity is displaying it enlarged to fit the screen. To see it as it will appear on a webpage (that is, at its real pixel size) change the zoom to 100% or press ⌘ (cmd) +1. As you can see the image is much smaller, contains obviously less details and is slightly less sharpen because you are reducing its pixel size/resampling considerably. Don't forget to change the Resample algorithm to Lanczos when resizing it, then you may need to apply an Unsharp Mask Filter or a High Pass Filter technique to recover some of the sharpness.

[accendare]

maybe i am being too picky but at you mentioned even at 100% the image is jaggy

i am not familiar on how to upload an image or would include it, anyway thanks for taking

the time to respond. i will do more experimentation and see if I can obtain a cleaner image

[MEB]

Hi accendare,

Can you upload the image directly to my Dropbox account using this link please?

[JimmyJack]

As per your numbers above.... 2336 X 3505 to 144 x 216. Lanczos. 

This is what I get. 100% side by side comparison. (Simple shape to see edge better... rasterized before sizing)

[Oval]

As per your numbers above.... 2336 X 3505 to 144 x 216. Lanczos. 

This is what I get. 100% side by side comparison. (Simple shape to see edge better... rasterized before sizing)

 

This is not a normal photo and it is not resharpened and cannot explain his “loss of quality” and “degraded so bad” and the issues of accendare, right?

[R C-R]

the photo is a RGB  2336 X 3505 pixels at 72 DPI

 

I go to resize document and change the size to 144 x 216 pixels-at the same DPI  (72) and the image comes out pixelated using lanzocs 3 (both options) - now how should making the image smaller have such a reduction in quality-and this is before exporting as a jpeg.

Maybe thinking of this way will help:

 

You are trying to reduce all the details in a large 2336 x 3505 pixel grid into a tiny one that is only 144 x 216 pixels. No matter what resampling algorithm you use, you cannot avoid throwing away massive amounts of detail -- there just are not enough pixels left for that. Remember, a pixel can only be one color -- there is no way to split it into smaller, differently colored sections.

 

So the best the algorithm can do is to sample variously sized adjacent groups of the original image's pixels & reduce them to much, much smaller groups of pixels representing some kind of weighted average of the colors of the larger groups. This averaging process is what causes blurriness & pixelated edges.

[R C-R]

As per your numbers above.... 2336 X 3505 to 144 x 216. Lanczos. 

This is what I get. 100% side by side comparison. (Simple shape to see edge better... rasterized before sizing)

After resampling, try setting the canvas to the Pixel view mode, then set the view to Zoom to Fit. It won't matter if you rasterize the ellipse or turn on Force Pixel Alignment or whatever. You still only have 144 x 216 pixels after resampling to fit the shape into, & the results will be a stair-stepped, anti-aliased edge.

[JimmyJack]

After resampling, try setting the canvas to the Pixel view mode, then set the view to Zoom to Fit. It won't matter if you rasterize the ellipse or turn on Force Pixel Alignment or whatever. You still only have 144 x 216 pixels after resampling to fit the shape into, & the results will be a stair-stepped, anti-aliased edge.

 

Zoom to fit? Why in the world would I do that? Of course that will be pixelated. 

I'm showing that at actual pixel size (the way it'll be seen on a website) the degredation shouldn't be so bad.

(unless nearest neighbor is being used....   ;) ).

It is a massive drop down, so a little something has to be expected. 

 

Oval.... here's a better example(s).

Lanczos sep, non sep.

[R C-R]

Zoom to fit? Why in the world would I do that? Of course that will be pixelated. 

I'm showing that at actual pixel size (the way it'll be seen on a website)...

Because it shows you what the edges really look like at the pixel level.

 

And don't forget that at "actual pixel size" is not the only size a browser might display images on a web page. Depending on <img> tags, style sheets, & frames (among other things) the actual display size may be something other than a 1:1 mapping of the image's pixels to those of the monitor (or HDTV or smartphone or whatever).

[JimmyJack]

Because it shows you what the edges really look like at the pixel level.

 

And don't forget that at "actual pixel size" is not the only size a browser might display images on a web page. Depending on <img> tags, style sheets, & frames (among other things) the actual display size may be something other than a 1:1 mapping of the image's pixels to those of the monitor (or HDTV or smartphone or whatever).

 

Mmmm, it's an exaggerated view of what it looks like at pixel level. Pixel level is pixel level.

 

1:1

Yeah remember the days when all the apps  weren't caught up to retina screens? Temporarily, while they resized everything properly, they just showed everything at 2:1 ..... and that's why it all looked like crap  ;) . If a builder is pushing that ratio too far, it's wrong. IMHO.

[deborah]

Hi R-C-R, You made a comment about my post above and it was my bad when I said resampling. It IS when resampling is UNCHECKED that width, height and resolution are linked. I watched the tutorial you suggested and understand but am experiencing many of the same issues as Accendare and am wondering why Affinity has made a complicated issue out of what seems fairly simple in Photoshop. If I understand correctly, when you cram a large number of pixels into a smaller area than its original area, the detail and resolution increases (and in the opposite case, decreases). Is that correct? Please help if I am not understanding correctly.

 

I work with DPI a lot because of printing but work also with web images that need to be high quality for our online store. Thanks for any help.