Posted by Steve Foss on January 1, 2008 at 10:18 AM
When opening a RAW image in CS2, I can ask the program to interpolate and enlarge the image to approximately 13x19 at 300ppi. Or I can open it to its standard size and increase size later, also then asking CS2 to interpolate. My question is whether there's any difference in quality between the two options. Or in other words, does CS2 interpolate any differently one way or the other?
With a somewhat older, slower G4, it's much more time-smart for me to do full toning and color correction with the smaller file and interpolate at the last step, but if interpolating upon opening is better, that's what I'll do.
I'm using CS3, so I'm not sure if what I have to say will apply to CS2.
According to Michael Reichmann and Jeff Schewe over at the Luminous Landscape the difference between the Camera Raw interpolation and that of Bicubic Smoother in Photoshop is very slight.
Supposedly there is some advantage to doing interpolation in Camera Raw as it is earlier in the "demosaicing pipeline" (their words, not mine).
If you're looking to uprez just a little bit for printing you might be interested it getting their "From Camera to Print" video tutorial. It is a great tutorial, especially if you're new to the world of printing. In it they state that it is actually better to keep your photos in their "native resolution" (again their words, not mine) to print and not resample.
For example, if you've got a 10 x 16 300ppi image you're better off printing a 13 x 19 240ppi image than uprezing. They recommend not resampling for anything between 180 (although this is a bit low) to 460 (or something in the 400's). Interesting stuff.
I'm still using CS2 Steve, but my techniques have not been verified against any "official" pathways. One of those works-for-me circumstances, so it may be specific to my images from Nikon D2X. I do all my image adjusting first, but uprez before my final sharpen. That allows me to make any final small corrections needed after uprez before I sharpen.
If it's a low-contrast shot I may vary the process by gently applying Reichman's "high pass" sharpening technique or whatever it's called (I think the tutorial is called "local contrast control" or some such at his site). It's the one where you create a dupe layer, apply the high pass filter, then switch to hard light in the layers pallete, and use the slider to adjust the effect downward. Done down at 10-20% opacity, this will often seem to carry low contrast detial through the uprez better. Even after that little bit of high pass sharpening, I'll go ahead and do my normal sharpening after the uprez.
Again- That's nothing "official" and will probably drive the purists nuts, but it works well for me when applied. Try it, and if you don't like the effect you can go back into the history pallet and readjust or eliminate the effect.
Your uprez isn't all that extreme, so you may not suffer many ill effects with either approach, and my extra step may be superfluous.
Posted by Nat Coalson on January 9, 2008 at 11:48 PM
I believe it's always better to work on your Master at native resolution.
Only do resampling when it's necessary, such as making a print or preparing for the web, and save derivative files at those sizes if you will use them again.
You'll benefit from smaller file sizes and more flexibility in the future. For example, if you upsample when converting to raw, and then you need to downsample later you're losing more data than if you left it alone during raw conversion.
Posted by Steve Foss on January 10, 2008 at 12:29 AM
"You'll benefit from smaller file sizes and more flexibility in the future. For example, if you upsample when converting to raw, and then you need to downsample later you're losing more data than if you left it alone during raw conversion."
Thanks, Nat, I had that part figured out. What I'd like to know, however, is whether the interpolation process is different (potentially better) upon opening the file through the preview screen than it is when upsampling later in CS2 proper. Do you know if there is a difference?
Discussion
4 repliesWith a somewhat older, slower G4, it's much more time-smart for me to do full toning and color correction with the smaller file and interpolate at the last step, but if interpolating upon opening is better, that's what I'll do.
According to Michael Reichmann and Jeff Schewe over at the Luminous Landscape the difference between the Camera Raw interpolation and that of Bicubic Smoother in Photoshop is very slight.
Supposedly there is some advantage to doing interpolation in Camera Raw as it is earlier in the "demosaicing pipeline" (their words, not mine).
If you're looking to uprez just a little bit for printing you might be interested it getting their "From Camera to Print" video tutorial. It is a great tutorial, especially if you're new to the world of printing. In it they state that it is actually better to keep your photos in their "native resolution" (again their words, not mine) to print and not resample.
For example, if you've got a 10 x 16 300ppi image you're better off printing a 13 x 19 240ppi image than uprezing. They recommend not resampling for anything between 180 (although this is a bit low) to 460 (or something in the 400's). Interesting stuff.
Hope this helps.
Alex
If it's a low-contrast shot I may vary the process by gently applying Reichman's "high pass" sharpening technique or whatever it's called (I think the tutorial is called "local contrast control" or some such at his site). It's the one where you create a dupe layer, apply the high pass filter, then switch to hard light in the layers pallete, and use the slider to adjust the effect downward. Done down at 10-20% opacity, this will often seem to carry low contrast detial through the uprez better. Even after that little bit of high pass sharpening, I'll go ahead and do my normal sharpening after the uprez.
Again- That's nothing "official" and will probably drive the purists nuts, but it works well for me when applied. Try it, and if you don't like the effect you can go back into the history pallet and readjust or eliminate the effect.
Your uprez isn't all that extreme, so you may not suffer many ill effects with either approach, and my extra step may be superfluous.
Only do resampling when it's necessary, such as making a print or preparing for the web, and save derivative files at those sizes if you will use them again.
You'll benefit from smaller file sizes and more flexibility in the future. For example, if you upsample when converting to raw, and then you need to downsample later you're losing more data than if you left it alone during raw conversion.
Thanks, Nat, I had that part figured out. What I'd like to know, however, is whether the interpolation process is different (potentially better) upon opening the file through the preview screen than it is when upsampling later in CS2 proper. Do you know if there is a difference?