![]() ![]() In order to get as true a comparison as possible I exported each image from Lightroom as 800x533 jpgs at 100% quality. I added Gigapixel AI on the advice of Nawlins. So if your main goal is upsampling for printing, here is probably the best place to post about this. I'd just caution that techniques that look good for a print sometimes don't look that good on-screen, and vice versa. I wonder if this thread would be better over on Retouching. And I don't doubt that Qimage and On1 would also do a plenty good job. However, for the modest upsampling I'm likely to to, the Lightroom Print module's algorithms do a plenty good job for me. My personal suspicion is that currently Topaz Gigapixel AI is generally the best, and well ahead of anything else for most image. Probably the starting image, any control settings, and your taste / priorities will have a substantial effect on what you thing produces the best result. At this point it's a question of personal taste. The sharpening makes it look like a palette knife painting. It is a bit sharper, but also a bit cruder. And starting with effectively 30 ppi is very low.īut I'm not sure I like the ON1 result. I bet so: that's a 10x upsample, which is a ton. I cropped a 120x180 slice out of the photo and had ON1 resize it to 1200x1800 and then printed both on glossy 4圆 paper. The dramatic differences seem to show up in the 4x to 6x range. ![]() My impression is that in the 2x upsample range, methodological differences often produce relatively modest / subtle differences in results. Pixma printers are usually (always?) 300 ppi, so again depending on just how you printed, going from the 125 or 140 ppi you had to the 300 ppi the printer used was a 2.1x or 2.4x upsample. * In any case, the upsample was relatively modest. So to go from the pixels you fed the software to the limits of what you can see might have effectively been a modest 1.1x to 1.6x upsample, and anything beyond that was more than you could see. Depending on how good your vision is, the image colors and contrast, the viewing distance, and the paper (here, luster, which does not show as much fine detail as glossy), the visible limits might be in the 150 to 200 ppi range. Depending on whether you printed the 8x10 'full bleed' or with white space (the proportions of 8x10 don't match 1000x1400), you had a native 125 or 140 ppi (presumably high-quality pixels at that, given the 0.4x downsample at the outset), which is low but not extremely low. * The test had approaching enough pixels before the upsample. To the extent you believe the few samples you see online and the comments others make, the only consumer-oriented product that I'd expect to be appreciably better than Qimage in some situations is Topaz Gigapixel AI (which seems capable of being astoundingly better with some images). Although I have not used either Qimage or On1 (or Genuine Fractals, which I understand On1 bought and incorporated), by all accounts Qimage has some excellent upsampling capabilities. FWIW-maybe not a lot, given my lack of experience with some of this-your experience does not surprise me. Given the lack of more informed responses, I'll offer some thoughts. Is this not a large enough change to see a difference? Or is Qimage just so good at resizing on its own that ON1 is moot? Darned if I can tell even the slightest difference. I took a 1000x1400 image and blew it up to 2500x3500 and printed them both on 8x10 luster paper on my 3-color Pixma using Qimage. I downloaded the latest ON1 package because (1) it was on sale for $50 which was (2) less than the Resize module alone, which is all I am interested in. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |