RSS/Atom feed Twitter
Site is read-only, email is disabled
  1. Chat
  2. Useful links
  3. Polls

Featured content

image quality degraded due to scaling

This discussion is connected to the gimp-developer-list.gnome.org mailing list which is provided by the GIMP developers and not related to gimpusers.com.

This is a read-only list on gimpusers.com so this discussion thread is read-only, too.

2 of 2 messages available
Toggle history

Please log in to manage your subscriptions.
image quality degraded due to scaling smit sanghavi 10 Feb 10:57
  image quality degraded due to scaling Burnie West 10 Feb 16:08
smit sanghavi
2012-02-10 10:57:37 UTC (almost 13 years ago)

image quality degraded due to scaling

Hi,
While using GIMP, I noticed that when an image is scaled initially, and then rescaled after making some changes, the quality of the image reduces. It is very significant when the image is scaled by a large extent. I understand that this is due to the data loss that occurs during scaling. But i was wondering if there was any alternative approach to avoid this problem.
I wonder if the original image can be cached or stored somewhere and when the image is rescaled, the changes that have been done, mapped to the original image and then displaying the final image with respect to that. Is this approach implementable? Or is there already some provision to handle such things?
I am new to open source development. So any help on how to go about solving this would be great.

Burnie West
2012-02-10 16:08:46 UTC (almost 13 years ago)

image quality degraded due to scaling

On 02/10/2012 02:57 AM, smit sanghavi wrote:

Hi,
While using GIMP, I noticed that when an image is scaled initially, and then rescaled after making some changes, the quality of the image reduces. It is very significant when the image is scaled by a large extent. I understand that this is due to the data loss that occurs during scaling. But i was wondering if there was any alternative approach to avoid this problem.

My approach to avoiding this problem is to scale up in binary steps (2,4,8) a lot first, then modify the image, then scale down to the final target size. This sort of works around the limited resolution by providing a couple of extra bits to average with on the down-scale step. It seems to work better if the scaling mode is set to Sinc (Lanczos 3) rather than the default cubic mode. Haven't really explored it much, though.

-- Burnie