NEF files Resized 2X Linear then .jpg'd?
I recently resized some NEF files from my D40 to 4000 x 6016 (24 MP) NEF files. They were all enhanced with the four standard options in the resize menu then saved as max quality .jpg files. Results looked much sharper and cleaner than the orignial 6 MP file sizes in both NEF and .jpg types.
Has anyone done any tests or experiments doing this?
DSLR's: D40, D700
FSLR's: FG, N80, F100
DX Lenses: Nikkor 18-55 VR, 18-70, 55-200 VR
FX Lenses: Nikkor 18-35, 24-85, 28-80, 28-200, two 50's, 70-300 VR; Series E 28, 100 & 135.
#1. "RE: NEF files Resized 2X Linear then .jpg'd?" | In response to Reply # 0blw Nikonian since 18th Jun 2004Sun 26-Sep-10 11:21 PM
You did this with the D40's on-board retouching menu, or with some software in the computer? When you looked at the results, was that on the LCD of the camera, or on a screen? (I'm pretty sure you didn't make an 8x12 print...)
Brian... a bicoastal Nikonian and Team Member
My gallery is online. Comments and critique welcomed any time!
#2. "RE: NEF files Resized 2X Linear then .jpg'd?" | In response to Reply # 1Mon 27-Sep-10 09:59 AM | edited Mon 27-Sep-10 10:40 AM by Bart B
Here's my process to get images for comparison:
1. Open original NEF file with Photoshop Elements 8 (PSE8) then convert it to 8 bits/pixel.
2a. Resize NEF file to 2000 x 3008 and 100 dpi for a 20 x 30 inch print then do step 3.
2b. Resize NEF file to 4000 x 6016 and 200 dpi for a 20 x 30 inch print then do step 3.
3. Enhance resized NEF file with "Auto Levels/Contrast" & "Unsharp Mask" set to 2.0 pixels.
4. Converted the resized and enhanced NEF files to .tiff and .jpg max quality then save them.
Compared images by opening them with two different graphics programs then sizing them to acutal print size on a 27 inch 1080 x 1920 monitor. Switching between the 24MP .jpg and .tif images exactly overlapping each other shows me the 24MP .jpg ones are noticably better. The .tif ones are about the same.
I'm going to try the same thing with my D700 an original 12MP then one resized to 48MP.....just for the heck of it. Then maybe post some samples in these forums for all to see.
#3. "RE: NEF files Resized 2X Linear then .jpg'd?" | In response to Reply # 2elec164 Nikonian since 15th Jan 2009Mon 27-Sep-10 01:13 PM
I believe there are several issues with your testing method.
First is that judging images on screen is very different then the results you will get from when they are printed out.
Second is that you are dealing with two resolutions which are inter-related but separate. They are spatial resolution which is fixed at the time of capture and pixel resolution which is changeable through interpolation. You can throw away spatial resolution by down sampling an image but you cannot add spatial resolution with up sampling an image. The benefit of up-sampling is when printing. Trying to print an image below 150 PPI will most likely show artifacts such as pixilation. By up sampling you can avoid those artifacts, but you cannot add the fine detail needed to maintain sharpness. So if you compare an actual image captured at 6MP interpolated to 24MP to one captured at 24MP then printed at 20x30, the actual captured 24MP print should blow away the 6MP print. But both may look perfectly acceptable when viewed on their own.
Third, I believe applying the same USM settings to both resolutions does not apply the same amount of sharpening. Sharpening should be done with the output devices and resolution in mind. It gets more complicated then this brief explanation but in ‘Image Sharpening’ by Fraser and Schewe they recommend when printing creating halos of about 1/100 of an inch. By that definition and your examples you should use a pixel radius of about 1 for the 6MP image and 2 for the 24MP image. But as I said it is more complicated than that. Along with the resolution when considering the amount of sharpening needed, you would use different amount tailored to the output device. You would use a different amount for screen display than when printing on an inkjet printer, and the amount of sharpening for glossy paper would be different than for matte paper.
Fourth, when viewing images on screen the PPI setting in the image resize dialogue box is meaningless for monitors only show images at actual pixels. Today’s LCD monitors generally range from 86 to 100 PPI. That would mean that at actual pixel size (100% view) your two examples would equal a print size of about 23x35 to 20x30 for the 6MP image and 46.5x70 to 40x60 for the 24MP image on the average monitors. Using the “view at print size” option in PSE8 would automatically scale the images in accordance with its resolution. In your case you monitor is displaying at about 81 PPI which means that you are viewing the two files at different percent views (about 81% for the 6 MP and 40% for the 24MP). That means you are altering the sharpness results through on the fly interpolation. So in that case it is no surprise to me that the 6MP image at near actual pixel viewing does not look as sharp as the 24MP image shown at less than 50 % view. Of course that is assuming you changed the default 72 PPI in the ‘units and rulers’ preferences in PSE otherwise your print size representation will be inaccurate and the percentages different than what I stated.
Just my thoughts on this for you to consider, as well as to put them out there to verify that my understanding of this complicated subject is correct.
Visit my Nikonians gallery.
#4. "RE: NEF files Resized 2X Linear then .jpg'd?" | In response to Reply # 3Mon 27-Sep-10 01:39 PM
Pete, you've made some good points; some of which I hadn't thought of.
I'm going to crop 8 x 10 inch samples from each 20 x 30 inch image then print them out on photo paper with the printer set to 600 dpi. There'll be some interpolation between the source file's resolution to the printers. It'll be interesting to see the differences.