Jump to content
You must now use your email address to sign in [click for more info] ×

Shrinking Images Makes them Jagged/Fuzzy


Recommended Posts

Since the compression in Affinity is different than Photoshop, what would be a comparable option in Affinity to a setting of 72 in Photoshop? 

 

That said, just to test it myself, I cropped the original image of the car (from the onedrive link above), and exported it as jpg in affinity and photoshop elements using the max options for optimization, bicubic resampling, and the results are here.. personally, I think the PSE file still looks better. 

 

I'm not sure about Photoshop Elements as I only have Photoshop, however I've just used Photoshop to resize your car pic and these were the settings I used to get more-or-less the same image quality/file size.

 

Affinity Photo

Size: 1333 X 1000

Resample: Lanczos 3 Separable

Quality: 85

Embed Metadata Off

File Size: 334KB (342,383 bytes)

Pic (Direct Image Link):

post-29590-0-28763500-1484692927_thumb.jpg

 

 

---------------------------------------

Adobe Photoshop CS4

Size: 1333 X 1000

Resample: Bicubic Sharper

Quality: 65

Embed Metadata already off in 'Save For Web'

File Size: 334KB (342,786 bytes)

Pic (Direct Image Link):

post-29590-0-70744300-1484692947_thumb.jpg

 

---------------------------------------

 

For the record I also tried Affinity Photo Bilinear with a quality setting of 88 (which made a 352KB file), but it's noticeably softer than the above two, so I'm not entirely sure why that's the default Affinity resampling method rather than say Bicubic or Lanczos 3 Separable.

 

Also, although it shouldn't affect your smaller photos, it may be worth going to Edit > Preferences > Performance, and changing the View Quality setting from 'Nearest Neighbour' to 'Bilinear (Best Quality)'.  It's set to 'Nearest Neighbour' by default, which displays jaggy edges in Affinity Photo when a larger image is displayed smaller than 100% (it doesn't affect the export quality, only viewing in Affinity).

Link to comment
Share on other sites

Sima, thanks for the linked images. The two of them are very similar, I don't know that I could tell a difference.  

 

Yeah, they're pretty close.  Here's the 1000px X 1000px straight crops as well in case you're interested.  I also added a note to the end of the previous post about the view quality settings just as you posted your reply.

 

Affinity Photo

Size: 1000 X 1000 (straight crop)

Resample: Lanczos 3 Separable

Quality: 85

Embed Metadata Off

File Size: 205KB (210,451 bytes)

Pic (Direct Image Link):

post-29590-0-12147300-1484701801_thumb.jpg

 

---------------------------------------

Adobe Photoshop CS4

Size: 1000 X 1000 (straight crop)

Resample: Bicubic Sharper

Quality: 65

Embed Metadata already off in 'Save For Web'

File Size: 206KB (211,211 bytes)

Pic (Direct Image Link):

post-29590-0-78707400-1484701821_thumb.jpg

 

---------------------------------------

Link to comment
Share on other sites

  • 2 months later...

Hi JimmyJack,

Yes, if you don't change the dimensions when exporting, there's no resampling.

 

I noticed that in some cases the resampling algorithms are applied, even if not changing dimensions at export:

  • Open any image in AP
  • Place another image via File > Place… (you get a new Image layer)
  • Downsample via Document > Resize Document…
  • Export via File > Export… to TIFF; don’t change dimensions, choose Bilinear
  • Export again via File > Export… to TIFF; don’t change dimensions, choose Lanczos Non-Separable
While steps 4 and 5 do not further resize the images, the downsample algorithms are applied nonetheless --> the exported images have the same dimensions, but quality (and file size) are quite different due to the different algorithms. (Bilinear: blurred, Lanczos: sharper)

 

Shouldn’t the exported images be identical, since I don’t change dimensions? (Note that the whole document has already been resized before.)

 

 

Alternatively:

  • Open any image in AP
  • Duplicate the Background layer with ⌘J (you get a new Pixel layer)
  • Resize the duplicated layer by dragging its handles
  • Optional: Downsample via Document > Resize Document…
  • Export via File > Export… to TIFF; don’t change dimensions, choose Bilinear
  • Export again via File > Export… to TIFF; don’t change dimensions, choose Lanczos Non-Separable
The result is similar to above.

 

AP 1.5.2, macOS 10.12.4

 

 

– Tom

Link to comment
Share on other sites

I prefer the Catmull-Rom resampling for all my down-sampling, since it:

  • will not add additional noise
  • will not add fringes/halos along the edges (looking at you, Lanczos!
  • keeps photographic details remarkably intact
  • keeps sharp highlights
  • keeps the fidelity of vector artwork and results in crisp and clean edges

The main issue with Lanczos and down-sampling are the edge halos that Lanczos introduces - almost as if a sharpening method is applied. Lanczos is good for up-scaling, though.

 

Because Photoshop and Affinity Photo are very limited in their choice of resampling options, I used ColorQuantizer to scale the photo with the road sign to 800px width. No additional processing performed. (download ColorQuantizer here: http://x128.ho.ua/color-quantizer.html)

 

Download the results for comparison:

https://drive.google.com/open?id=0B03AC_RA8Ch1VmpXeE1jdnd0Um8

 

Notice that Lanczos results in a larger file size caused by the extra information (edge halos) that was introduced while downsampling. This continues down in the pipeline: the down-sampled Catmull-Rom version is "cleaner" and results in smaller JPG files at the identical compression settings compared to the Lanczos down-sampled version. Those extra halos may also become more problematic to process in jpg, resulting in increased artifacting (slightly increased noise).

 

For example, the exact same image processed with Catmull-Rom and saved as a jpg results in a 123,449 bytes jpg. Compare this with the Lanczos version that produces a 127,808bytes large file. And the Catmull-Rom has less noise around the edges

(RIOT jpg settings: 85%, Progressive, Chroma sub-sampling set to Low (4:2:2).

 

All this is somewhat academic, of course. For most people Lanczos will suffice, but Catmull-Rom is the better choice for down-sampling images nonetheless - both in quality and final file size. It has been proven in other tests online as well.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
×
×
  • Create New...

Important Information

Terms of Use | 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.