Photoshop Image Resize

donaldejose

I have been working with Image Resize in Photoshop Elements 13 and am wondering what it does. You can set the size of the final image in inches (or centimeters) and the resolution you want (in pixels per inch) and then "somehow" using "bicubic" something or other Photoshop translates your existing image into the size and resolution you want. I don't see a difference in the image when I do this and notice the final megapixel count of the final image increases. I am wondering if this is a good way to "gain" pixels in an image. Perhaps someone here is knowledge about the pro and con of using this Photoshop feature? I really have not been able to detect any significant degradation of my image but maybe I don't know what I am looking for.

heartyfisher

Its all too confusing for me .... I just keep with what i understand and just work in pixels .. and forget about the inches(or centimetres)

donaldejose

You can work with pixels also. The reason to work in inches is when you are printing and want to make a 24x36 inch print at 300dpi. You can input those numbers and the software generates exactly the number of pixels you need to make that print. Just tell the printer to print the file at 300dpi and you will get a 24x36 inch print. Or just tell the printer to print the file at 24x36 inches and you will automatically get a print at 300dpi. If you don't "upsize" first you can tell the printer to print the image at 24x36 inches but it will be printed at 150dpi, not at 300dpi because there are not enough pixels in the image do do so. If you want to work in pixels you can tell the software how many pixels you want in one dimension and it will automatically generate them with the proper ratio of pixels in the other dimension. The software is generating pixels you didn't have in your original file. There are also some options in how it does this. I have been doing this and I don't see what I am losing from the original image. It seems like magically being able to increase the megapixels of your original sensor. For example, if you have a 24 mp sensor that is 6,000 x 4,000 pixels. 6,000 x 4,000 pixels prints out to about 13x20 inches at 300dpi. But what if you want to print that image from your 24mp sensor at 24x36 inches? It would print out at 150 dpi. The image resize software will generate a 10,500 x 7,000 pixels image from your file which prints a 24x36 inch image at 300dpi or pixels per inch. 10,500 x 7,000 is the same as having a 73 megapixel sensor. I am not trying to claim software can turn a 24 megapixel sensor into a 73 megapixel sensor but I am saying I don't see what is being lost in my images. However the software is generating those additional pixels is seems to be doing so just fine. I am suggesting that an image upsized in this way and printed very large at 300 dpi may be better than an image not upsized and printed out at 150 dpi. Does anyone have any experience or knowledge with using photoshop to upsize the number of pixels in your image prior to printing it very large?

ggbutcher

Simply put, some rescale algorithms handle edges better than others. Bicubic is good in that regard, but you might compare it to catmull-rom, if available. I came by this insight programming with the FreeImage library; the documentation has an appendix that concisely discusses and illustrates the six algorithms included in the library.

FreezeAction

Costly but if anyone needs really large, I've seen first hand at a trade show what can be done with the add in in CorelDraw's suite. I've yet to try it. At the trade show was an 8x10" print displayed by an 8'x10' mosaic. I got right up close and saw no nasty pixels. Photoshop tells you in the tool which is best for upsizing and downsizing. I took the example photo of Nikon's D800E and enlarged to 40x60 (the library of congress with the 14-24 lens) and got some friends to examine it. No one could find any thing bad with it. I know that Nikon had pulled out all stops before they put it up for try outs. Still with one click it was enlarged to not even the maybe max.

donaldejose

Right, it seems to me to be the cheapest way to add pixels to an image as long as you are starting with a very clean image in the first place.

FreezeAction

This is the link to the image I enlarged to 40x60 and then printed the lower 24x60 as a test. I was wrong about the D800E as it was from the D800. http://imgsv.imaging.nikon.com/lineup/dslr/d800/img/sample01/img_01_l.jpg
Here 36MP does more than IMHO the mass majority of shooters ever will need. I think the majority of shooters need very clean high ISO so they can hand hold at a faster shutter speed and hope the form of image stabilization in use will make up for a dead still MUP image from a tripod. Then at least a modest 24x36 poster from 20MP should be doable with civilian forms of interpolation software. Ten years and so ago we used the 10% rule with Photoshop. I've tried that in PS6 and it is not as good as one quick move. I tried a 1% action for interpolation and it is useless.