Let r be the downsampling ratio (approximately equal to the output and input widths in pixels).
I would not expect NoHalo to give great results if r * OFFSET_0 is much smaller than 1, and LoHalo if r * OFFSET_0 is much smaller than 2. Specifically, if OFFSET_0 is 16, I would not expect any of the two methods to give great results if downsizing to 1% original size (r * 16 = .16 which is much smaller than 1). Tolerable results? Maybe. Great? Probably not. This limitation used to be overcome by relying on mipmaps. For speed + stability reasons, they've been taken out. (I do wonder if using one mipmap level could make enough of a difference.) Short version: Big time downsampling most likely will not be as good as it used to be. _______________________________________________ gimp-developer-list mailing list List address: gimp-developer-list@gnome.org List membership: https://mail.gnome.org/mailman/listinfo/gimp-developer-list List archives: https://mail.gnome.org/archives/gimp-developer-list
