Jump to content
crookedhalo10

Image quality suffers after resizing document

Recommended Posts

Often I will need to send images to my boss, and to get them to through email I simply reduce the file dimensions, to reduce the file size, so they can go through. In some cases after making the images smaller, their is a noticeable drop in image quality. I wouldn't call it noise but things don't look as sharp or there are jagged edges. This happens with images I take or that I find on the web. Also happens regardless of other filters or effects added. I will try to post some examples when I can, but just wanted to see if anyone else is experiencing this or if I'm just lucky.

 

Share this post


Link to post
Share on other sites

i just tested that as I am preparing some samples. It is not as bad when using the other sampling methods, but still seems like an obvious reduction in image quality. And its not a matter of being zoomed in too close at the newer size. When viewing at actual size or exporting the file, the image issues are still there, though depending on how small one goes, it becomes less noticeable. I'll post some samples soon.

Share this post


Link to post
Share on other sites

Here is a sample of what I'm experiencing. The original file is 4672px by 2628px. DPI of 72. So low DPI but huge in dimensions. Should be able to work with this, especially if the end result is a smaller dimension.

 

So I change it to 792px by 446px which is still pretty big at this point. I kept the Resample: at "Nearest Neighbour" but I experience similar results with other sample options.

 

To me its seems obvious where the strange artifacts and jagged edges appear, but if you look closely at the inside edge of the washer and the overall texture of the washer looks rough or has noise compared to the original picture. As if a sharpen filter was applied and someone over did it.

 

Any thoughts?

Post1.tiff

Pre1.tiff

Share this post


Link to post
Share on other sites

I don't know how you're getting these results but we do better than that.

The post1.tiff file seems more like a screen capture at a higher zoom level (to match the pre1.tiff original pixel size) than just a simple downsample.

You can only compare overal image quality at 100% pixel view.

 

Can you please attach the original file (4672px by 2628px) too (or PM me a link if you prefer) just to make some tests please?

I will delete the file as soon as i finish.

Share this post


Link to post
Share on other sites

We think the results are OK, because the original is not perfect.

 

(You cannot get "similar results with other sample options” if not using unusual photos”)

Share this post


Link to post
Share on other sites

Don't use nearest neighbor. It's the most simplistic resample algorithm. Use Bicubic or Lanzcos as MEB says. The result will be MUCH better.

 

But this in NOT a bug. You are reducing the size A LOT.

 

Let me put it to you this way....

If you reduce the pixel size of an image by 50% in both dimensions, the resulting image contains only 25% of the original information.

You've reduced the dimensions of your image by over 83%, so (according to the exact pixel dimensions given) your second image contains just 2.87% of the original information.

You can't expect the same sharpness as the original when doing that.. no matter what resampling you're using. It would be the same in any photo manipulation application.

 

That being said..... after such a big reduction I would use an Unsharp Mask filter to (I should say a least try to) tighten things up a bit.

Share this post


Link to post
Share on other sites

JoePoe, forgive my ignorance because I may be comparing apples to oranges but I often get images from clients that are very large, dimensionally, but only 72 dpi in resolution. To use the image I need it at 300 dpi. In Photoshop I would resize the image to 300 dpi. with "Resample" turned off. This significantly reduces the dimensions of the image, usually by about half,  but I often need it to be even smaller. I then resize again with "Resample" turned on, to the dimensions I need, maintaining the 300 dpi resolution. The images are sharp and print perfectly. I'm not sure what I am asking here other than that surely this can be accomplished in Affinity Photo, correct?

Share this post


Link to post
Share on other sites

Changing the dpi will change the default print size of the image but it will not change its resolution (the number of pixels that are in the image) or its file size.

 

To oversimplify a bit, dpi stands for dots per inch, where dots refers to the ink dots a printer deposits on a printed page. It is different from ppi (pixels per inch), which refers to the resolution of the rasterized image itself. Changing the dpi doesn't change the number of pixels in the image; it just changes how close together they are on printed output.


Affinity Photo 1.7.3, Affinity Designer 1.7.3, Affinity Publisher 1.7.3; macOS High Sierra 10.13.6 iMac (27-inch, Late 2012); 2.9GHz i5 CPU; NVIDIA GeForce GTX 660M; 8GB RAM
Affinity Photo 1.7.3.155 & Affinity Designer 1.7.3.1 for iPad; 6th Generation iPad 32 GB; Apple Pencil; iPadOS 13.1.2

Share this post


Link to post
Share on other sites

R C-R, Thanks for the response. I didn't do a very good job explaining my question. I do understand the difference between "dpi" and "pip". It gets confusing because the two terms have come to be used interchangeably even though they mean different things and I made that mistake in my post. I'm really trying to understand why crookedhalo10 is getting poor results resizing in Affinity Photo and I am not when resizing in Photoshop. When I do transition to Affinity I want to be sure I am resizing correctly to avoid this. I know using "nearest neighbor" is the wrong setting. I really don't understand why it is the default but that is a different question. So given a large size, low res image, if I set a high ppi I get an image reduced in size dimensionally but the same number of pixels. If I then reduce the dimensions to the desired size should still get a sharp image in Affinity. Do I have that correct? In Photoshop if you reduce the dimensions of a large image with a resolution of 72 pip, with resampling turned off, the image resolution increases. If you reduce the dimensions with resampling turned on, the resolution remains at 72 ppi. Is that what is happening for crookedhalo10? He is reducing the dimensions without increasing resolution?

Share this post


Link to post
Share on other sites

I think you've got it Jmac.

(although I think you got dpi and ppi mixed up again on your second post...)

 

The key thing here is the resample box and pixels. Pixels = photo information.

If you take pixels away (or add them) you are resampling. (btw: a change in DPI setting does affect pixels.... if resample is on).

If resample is off then the pixel information is unaffected, and you're just telling the program to compact/spread ALL the pixels into a certain space for printing (as R C-R said). Notice when resample is OFF your pixel fields are greyed out.... they're untouchable.

 

So when, you "resized" from 72 to 300 with resample off all of your initial information was retained

(don't believe me ;) ? Check file size at this point, they will be exactly the same. AND they will be the exact same size on screen. The screen doesn't care about dpi. )

Then, your secondary/final resample was small enough to not have (much of) a visual impact.

 

CrookedH was resampling at the same dpi setting, so basically all he was doing was throwing out original data..... and, as a result, the image was smaller in terms screen real estate and file size. (remember the dpi setting is meaningless on screen in terms of pixels.)

 

...But don't get me wrong, Resampling is a great (and necessary) tool. It just can't be pushed too far. I personally start getting nervous around 30/40% (different images hold together better than others.)

 

So bottom line. Do exactly as you did in PShop and you'll be fine :).

 

(please ask if any of that needs clarification)

Share this post


Link to post
Share on other sites

Thanks, JoePoe! I think I am finally seeing the light. I've been nervous about making the switch to Affinity because of export issues some have been experiencing and needing to make sure I can get a client the file and quality they need. Not that the software can't do it but that I wouldn't get the most from the software. I keep sneaking up on the idea of making the switch and today I think I got a little closer.

Share this post


Link to post
Share on other sites

Keep in mind that what a client needs will depend on how they intend to use the image, & that there is no one export setting (or even file format) that is best for everything. There are many helpful articles on the web about this, for example this one, which I think covers the basics fairly well.


Affinity Photo 1.7.3, Affinity Designer 1.7.3, Affinity Publisher 1.7.3; macOS High Sierra 10.13.6 iMac (27-inch, Late 2012); 2.9GHz i5 CPU; NVIDIA GeForce GTX 660M; 8GB RAM
Affinity Photo 1.7.3.155 & Affinity Designer 1.7.3.1 for iPad; 6th Generation iPad 32 GB; Apple Pencil; iPadOS 13.1.2

Share this post


Link to post
Share on other sites

I've read through everyone's post and here are few my thoughts.

  • One of the first things I did was to make sure the image quality was not a result of it being zoomed in too close. Even at actual size there is noticeable drop in image quality. It's just a little bit less noticeable but still there. The samples I provided were captured at a zoom level so that people wouldn't have trouble seeing the issues.
  • There must be something wrong because there is a workaround I forgot to mention in my original post. I wanted to add this detail, but for some reason I was limited in how many post I could make. The workaround is simple enough. I do the normal changing of the document size, because usually I want to know the image size in inches vs pixels. Then I undo everything and work on the image at the original dimensions. When I am ready to export the image, at the export dialog box, I make the size changes there instead. And the image comes out just fine. If I make the dimension changes through the Document menu I get the issues. Why would changing the size in one dialog box (Document>Resize Document) produce issues but not another (File>Export)
  • I hear what everyone is saying about the resolution being a 72ppi. But remember if we start at a very large image dimensions wise, even with low ppi, shrinking it down shouldn't degrade the quality of the image. If anything it should help it covering up any artifacts. Just imagine an image the size of a billboard, with some noticeable flaws. Now shrink it down to the size of a postage stamp. The flaws are still there, but at the smaller size any jagged edges would appear smoother. Does anybody remember Quark? They had a term for this in their software. Relative Resolution. As you take an original image, and shrink it down, the resolution (ppi) technically increases. If you have a 50inch by inch image at 72ppi and reduce it to 10inch by 10inch, it still 72ppi as far the software is concered, but Quark would also compare the new size to the old and give you a Relative Resolution. So now maybe your 10x10 is equal to having a ppi of 180ppi or something. In any case reducing the image should not degrade the quality of the image. It should help it appear more sharp if anything. 

Share this post


Link to post
Share on other sites

I think you've got it Jmac.

(although I think you got dpi and ppi mixed up again on your second post...)

 

The key thing here is the resample box and pixels. Pixels = photo information.

If you take pixels away (or add them) you are resampling. (btw: a change in DPI setting does affect pixels.... if resample is on).

If resample is off then the pixel information is unaffected, and you're just telling the program to compact/spread ALL the pixels into a certain space for printing (as R C-R said). Notice when resample is OFF your pixel fields are greyed out.... they're untouchable.

 

So when, you "resized" from 72 to 300 with resample off all of your initial information was retained

(don't believe me ;) ? Check file size at this point, they will be exactly the same. AND they will be the exact same size on screen. The screen doesn't care about dpi. )

Then, your secondary/final resample was small enough to not have (much of) a visual impact.

 

CrookedH was resampling at the same dpi setting, so basically all he was doing was throwing out original data..... and, as a result, the image was smaller in terms screen real estate and file size. (remember the dpi setting is meaningless on screen in terms of pixels.)

 

...But don't get me wrong, Resampling is a great (and necessary) tool. It just can't be pushed too far. I personally start getting nervous around 30/40% (different images hold together better than others.)

 

So bottom line. Do exactly as you did in PShop and you'll be fine :).

 

(please ask if any of that needs clarification)

I'll give this a try, but I don't remember having to do so many steps to reduce an image and keep it looking good in Photshop. I just reduced the size of the image, never even cared about the re sampling options, and it worked out just fine. Also a good point someone made, if Nearest Neighbor is such a poor choice, then why is that the default? I notice in the Export dialog box, Billinear is the default. I thought that might be the issue, but I tried all the re sample options in the other dialog box, and still had issues. But yes Neighbor did the poorest job. So why have that as a default?

Share this post


Link to post
Share on other sites

Just a heads up. I originally had some issues signing up for the forum, so I had to create another profile to post. But Affinity fixed it so I'm switching to the account I wanted to use.

 

OP crookedhalo10 

Share this post


Link to post
Share on other sites

Also a good point someone made, if Nearest Neighbor is such a poor choice, then why is that the default? I notice in the Export dialog box, Billinear is the default. I thought that might be the issue, but I tried all the re sample options in the other dialog box, and still had issues. But yes Neighbor did the poorest job. So why have that as a default?

There is no one resampling or resizing option that is the best (or the worst) for everything. If you have not already done so, I suggest you open Affinity Photo Help (from the Help menu), type "resample" in the search box, & read the first item in the list of topics that produces, titled "Changing image size."

 

Note in particular the text in the tinted box near the bottom titled "Resampling methods." Nearest Neighbor is recommended for hard-edged images. Because it uses a simple but very fast resampling method, it is also often used for quick checks when high quality isn't important, particularly when working with very large images on slower, less powerful hardware.

 

As the help topic says, Bilinear is a good choice for shrinking images, Bicubic a good one for enlarging them, & the two Lanczos 3 ones will produce the highest quality results, but at the cost of significantly increased processing time.

 

It also may be helpful to read what it says about the difference between resampling & scaling an image, & how each affects the image's pixel dimensions.


Affinity Photo 1.7.3, Affinity Designer 1.7.3, Affinity Publisher 1.7.3; macOS High Sierra 10.13.6 iMac (27-inch, Late 2012); 2.9GHz i5 CPU; NVIDIA GeForce GTX 660M; 8GB RAM
Affinity Photo 1.7.3.155 & Affinity Designer 1.7.3.1 for iPad; 6th Generation iPad 32 GB; Apple Pencil; iPadOS 13.1.2

Share this post


Link to post
Share on other sites

 

I've read through everyone's post and here are few my thoughts.

  • One of the first things I did was to make sure the image quality was not a result of it being zoomed in too close. Even at actual size there is noticeable drop in image quality. It's just a little bit less noticeable but still there. The samples I provided were captured at a zoom level so that people wouldn't have trouble seeing the issues.
  • There must be something wrong because there is a workaround I forgot to mention in my original post. I wanted to add this detail, but for some reason I was limited in how many post I could make. The workaround is simple enough. I do the normal changing of the document size, because usually I want to know the image size in inches vs pixels. Then I undo everything and work on the image at the original dimensions. When I am ready to export the image, at the export dialog box, I make the size changes there instead. And the image comes out just fine. If I make the dimension changes through the Document menu I get the issues. Why would changing the size in one dialog box (Document>Resize Document) produce issues but not another (File>Export)
  • I hear what everyone is saying about the resolution being a 72ppi. But remember if we start at a very large image dimensions wise, even with low ppi, shrinking it down shouldn't degrade the quality of the image. If anything it should help it covering up any artifacts. Just imagine an image the size of a billboard, with some noticeable flaws. Now shrink it down to the size of a postage stamp. The flaws are still there, but at the smaller size any jagged edges would appear smoother. Does anybody remember Quark? They had a term for this in their software. Relative Resolution. As you take an original image, and shrink it down, the resolution (ppi) technically increases. If you have a 50inch by inch image at 72ppi and reduce it to 10inch by 10inch, it still 72ppi as far the software is concered, but Quark would also compare the new size to the old and give you a Relative Resolution. So now maybe your 10x10 is equal to having a ppi of 180ppi or something. In any case reducing the image should not degrade the quality of the image. It should help it appear more sharp if anything. 

 

 

Which part to start with? :)

 

Bullet #3:

First, I sincerely hope you are meaning DPI when you say PPI,... yes? If so, I believe this is what we are saying here. Without resampling, "reducing" an image in DPI does not change the # of pixels in the image.. just how crammed together they are for printing. And without resampling, the higher the DPI the smaller the print image, but the amount of information stays the same and the image will display the same on screen. I'm not sure I follow the billboard to postage stamp analogy because other factors are at play. Billboards are printed at fairly low DPI because they are (and have to be) viewed from greater distances.  Anyway, we can discuss that later....

 

Bullet #1:

Same here. I am posting images at zoom level to see issues .... and lack of issues.

The first attachment is a straight up comparison to PShop. Nearest neighbor and Bicubic. They are exactly the same.

Oh, and from your response a couple of posts earlier regarding # of steps.... It can be just one step, I was just explaining what was happening at each part of jmac's process. It can absolutely all happen at the same time!

 

Bullet #2:

Saved the best for last. THERE IS AN ISSUE. I SEE WHAT YOU'RE SAYING!!

The second attachment is an output comparison AP resize document (then saved) vs AP export (nearest neighbor). Very different. If my first attachment comparing output to PShop output is any indication, the resize version is the correct one. The, for whatever reason, "cleaner" export version seems to be the anomaly.

 

I think there should be a new thread reporting this issue, but with a different, more to the point title:

"Different output results with identical settings.... Resize doc vs Export."

post-12544-0-17457300-1453871424_thumb.png

post-12544-0-72283800-1453871436_thumb.png

Share this post


Link to post
Share on other sites

JoePoe,

 

In your Bullet #2 example, I'm not quite sure what you are comparing. If it is the exported JPEG file to the working document in memory, they won't ever be quite the same bit-for-bit because JPEG file compression is always at least slightly lossy, even at the 100% quality setting. (This is due to rounding errors in the DCT algorithm used to compress the image data.)

 

There are completely lossless JPEG schemes like JPEG 2000, but I don't think Affinity supports it.


Affinity Photo 1.7.3, Affinity Designer 1.7.3, Affinity Publisher 1.7.3; macOS High Sierra 10.13.6 iMac (27-inch, Late 2012); 2.9GHz i5 CPU; NVIDIA GeForce GTX 660M; 8GB RAM
Affinity Photo 1.7.3.155 & Affinity Designer 1.7.3.1 for iPad; 6th Generation iPad 32 GB; Apple Pencil; iPadOS 13.1.2

Share this post


Link to post
Share on other sites

Both are saved images (I went back and edited to make that clearer). But even if that weren't the case.... if that were to produce such a pronounced difference I would expect to see a large difference in the single image between the pre-saved resized photo and the saved version. But there isn't.

 

FWIW Same results in TIFF. That's the closest we have to Lossless right? But I'm not so sure that's what's going on.

Share this post


Link to post
Share on other sites

Please forgive the confusion of an old man but do you mean the image on the left (the one labeled "straight export") was resized in AP & then saved in the "native" AP format (a file with the .afphoto extension) prior to exporting it? If so, how did you resize the image in AP (using the "Resize Document" menu item, the Transform panel, some other method)?

 

Either way, how did you generate your comparison image? From what you uploaded it looks like both were opened in Apple's Preview app, but that app can't open native AP formatted files, so I assume you exported both files as JPEGs using the same Nearest Neighbor setting. If so, the comparison is between a two step process of first resizing & then resampling during export vs. (from a user perspective) the one step resizing, resampling, & exporting one.

 

If I got that right, I would not be surprised to see differences if the one step process internally applies the resizing & Nearest Neighbor resampling in a different order from the two step one, both because Nearest Neighbor resampling would have more "neighbors" to sample prior to resizing & because JPEG conversion breaks the image down into 8x8 blocks during the compression process that would be different at different image sizes.

 

It might be more revealing to do the comparison in AP, comparing the native format file to the JPEG export(s), & possibly to make sure the images are constrained to a multiple of 8 pixels in both width & height to produce the best possible "fit" for the compression.

 

But like you said, I'm not sure any of this will tell us what is really going on. JPEG compression is used in several different file formats & all of them are commonly called "JPEG image files," so it isn't even clear if they are all processed & displayed exactly the same way by different apps or within a single app that might handle them slightly differently.


Affinity Photo 1.7.3, Affinity Designer 1.7.3, Affinity Publisher 1.7.3; macOS High Sierra 10.13.6 iMac (27-inch, Late 2012); 2.9GHz i5 CPU; NVIDIA GeForce GTX 660M; 8GB RAM
Affinity Photo 1.7.3.155 & Affinity Designer 1.7.3.1 for iPad; 6th Generation iPad 32 GB; Apple Pencil; iPadOS 13.1.2

Share this post


Link to post
Share on other sites

I'm enjoying this thread and have a reply to all of your comments... but I am unable to upload any images at the moment.

 

Rather than reply and then post the relevant images later, I'll wait until I can figure it out (or, *cough*, the forum is fixed) to do it all together. (really don't want to do the post elsewhere and redirect thing)

 

.... Spoiler alert.... I still think there's an issue.

Share this post


Link to post
Share on other sites

I'm enjoying this thread and have a reply to all of your comments... but I am unable to upload any images at the moment.

 

Rather than reply and then post the relevant images later, I'll wait until I can figure it out (or, *cough*, the forum is fixed) to do it all together. (really don't want to do the post elsewhere and redirect thing)

 

.... Spoiler alert.... I still think there's an issue.

Well I'm glad that people are seeing that there is some kind of issue, even if pinpointing the exact cause is confusing. I've tried different sampling methods, and compared them when reducing the image from the Document menu versus the export menu and going through the export menu is always noticeably better. 

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×

Important Information

These are the Terms of Use you will be asked to agree to if you join the forum. | Privacy Policy | Guidelines | We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.