Posted by Jamie Botdorf on February 16, 2009 at 10:54 PM
I'm trying to create a 12x36 pano print from an smaller RAW file and looking for suggestions on PS steps to take (btw - I'm using PSE 6).
I've considered two processes and not sure if one is better than the other or if they are pretty much the same.
The first would be: open a copy of the RAW file, do the crop but don't set any parameters (size, ppi, etc), then use Image>Resize Image to enlarge the cropped image to 12x36 at 300ppi and select the bi-cubic adjustment.
The second approach is to open a copy of the RAW file and just use the crop tool and set the size to 12x36 and 300 ppi and let PS take care of it in one step.
Or...is there a better way? I'm open to suggestions. Thanks!
Posted by David Bostedo on February 16, 2009 at 11:22 PM
I guess they'd come out about the same Jamie. Re-sizing isn't really done on the raw data I don't think, even if you're in ACR. It would still be applied to the final, de-mosaiced, data - almost as if you'd saved the file, the did the resize in PS. Wouldn't hurt to try it both ways and see.
I think the more important thing would be the adjustments made to the file after resizing - particularly sharpening. There are lots of sharpening flows out there - I personally would do the re-size, then do all my sharpening, but YMMV.
Posted by Peter Langman on February 17, 2009 at 12:18 AM
from my understanding it sounds like you want to increase the file size.
you could try the 110% method where you increase the size by 110% (that is actually 10% overall size increase ) repeat this process untill you reach the desired size less chance of pixel damage this way will still get some but not as much.
. . . I second Peter's suggestion, and suggest that you sharpen with a layer mask for ultimate control (this is possible in Elements--you can check this out).
Posted by David Bostedo on February 17, 2009 at 9:37 AM
Adobe has acknowledged that changing the file size up incrementally no longer offers any advantage to doing it in one step. (It used to, but that was several PS versions ago.) I'd upsize it all in one go.
Posted by John Mumaw on February 17, 2009 at 9:38 AM
I agree with David if Elements offers the 'Bicubic Smoother' option of resampling.
Posted by Jamie Botdorf on February 17, 2009 at 11:40 AM
Thanks for the suggestions. They help quite a bit. I'll do all the processing after cropping and resizing. I can use Bicubic Smoother and do some layers with PSE. It's actually pretty good.
Posted by Steve Foss on February 17, 2009 at 11:42 AM
David, I'm still on CS2, and upsize 5 or 10 percent at a time. Do you know if Adobe's acknowledgment applies to CS2 or to the two later versions?
Posted by David Bostedo on February 17, 2009 at 1:55 PM
Steve, Bicubic Smoother was created to give the same result as the incremental steps. If you have that, you're probably good to go. As further reference for some semi-random techniques, this is interesting.
Oh...I just found the article (not the one from Adobe, but a good reference) ...seems that CS and newer gets rid of the need to incrementally upsize. Also found references to Scott Kelby and Tim Grey both recommending a single step flow. I, personally, have gotten some good results from oversizing, oversharpening, and then reducing slightly (sort of like the Adamus web prep flow).
Posted by Steve Foss on February 17, 2009 at 2:08 PM
Thanks, David. CS2 has bicubic smoother, so it sounds like that's good to go. I've been using bicubic for smaller enlargements (how's that for an apparent oxymoron?) for quite awhile now.
Posted by David Bostedo on February 17, 2009 at 3:42 PM
Bicubic is probably fine for most things Steve....sometimes I try things a couple of different ways. After all, there's no "right" way, only different ways that give different results. Sometimes results are clearly better, but a lot of times it's just personal preference.
Actually, a lot of times anymore I just let Qimage handle it, since I typically print from there now. It uses a variation of the Lanczos algorithm, which is pretty nice. That same feature is available in the free Irfanview program if anyone is interested.
Posted by Brad Kargus on February 17, 2009 at 11:26 PM
I recently compared Qimage and CS3 bicubic smoother for a file enlarged in one shot with each. I thought the Qimage result was ever so slightly more refined but the difference was so small (viewing at 100%) that I may have been imagining things. I did this for enlarging some 5D files prior to printing up to 36x24. I was very pleased with the results.
Posted by Sasha Gregory on March 14, 2009 at 7:20 PM
If you are using a full version of Photoshop, not elements I would use the bicubic upsizing method but set up an automation to increase the size just 10% (you will set the image size to percentage and set it to 110%) then do the automation again and again. doing that three times will produce better result than doing 30% once, not sure why but it does make a difference.
Posted by David Bostedo on March 14, 2009 at 10:27 PM
Sasha - Please read the earlier part of the thread, RE : Incremental upsizing
Discussion
13 repliesI've considered two processes and not sure if one is better than the other or if they are pretty much the same.
The first would be: open a copy of the RAW file, do the crop but don't set any parameters (size, ppi, etc), then use Image>Resize Image to enlarge the cropped image to 12x36 at 300ppi and select the bi-cubic adjustment.
The second approach is to open a copy of the RAW file and just use the crop tool and set the size to 12x36 and 300 ppi and let PS take care of it in one step.
Or...is there a better way? I'm open to suggestions. Thanks!
I think the more important thing would be the adjustments made to the file after resizing - particularly sharpening. There are lots of sharpening flows out there - I personally would do the re-size, then do all my sharpening, but YMMV.
you could try the 110% method where you increase the size by 110% (that is actually 10% overall size increase ) repeat this process untill you reach the desired size less chance of pixel damage this way will still get some but not as much.
Oh...I just found the article (not the one from Adobe, but a good reference) ...seems that CS and newer gets rid of the need to incrementally upsize. Also found references to Scott Kelby and Tim Grey both recommending a single step flow. I, personally, have gotten some good results from oversizing, oversharpening, and then reducing slightly (sort of like the Adamus web prep flow).
Actually, a lot of times anymore I just let Qimage handle it, since I typically print from there now. It uses a variation of the Lanczos algorithm, which is pretty nice. That same feature is available in the free Irfanview program if anyone is interested.