Increase image size

DF
Posted By
David_Freed
Dec 22, 2006
Views
591
Replies
25
Status
Closed
I want to increase my image size, from about 2880 x 2160 px to about 5760 x 4320, a 200% increase.

I’m wondering if it’s better to do this from within the RAW converter, or is it better to use either Bicubic Smoother or a plugin such as Genuine Fractals?

And I’m especially interested if the RAW file size converter is the same as or better than Bicubic?

Best, David.

Must-have mockup pack for every graphic designer 🔥🔥🔥

Easy-to-use drag-n-drop Photoshop scene creator with more than 2800 items.

S
stevent
Dec 22, 2006
GF themselves claim better results with larger increases in size (5/600% or so).

What version of Photoshop are you using? As the Image Interpolation was overhauled in CS2 and is better than previous versions.
DF
David_Freed
Dec 22, 2006
I have CS 1. I’ve used GF for very large increases in size and like it.

David.
BL
Bill_Lamp
Dec 26, 2006
I find this works for me:

Use the largest available file size in ACR.

Run NoiseNinja at a light strength and smoothness setting.

Crop to ratio, NOT to size and DPI.

If the change image size shows the current file has over 300 DPI, resize with Bicubic Sharper. The final sharpening is on a duplicate layer with FocusMagic, normally 1-2 pixels LESS than the detect button shows and often at 50-60% opacity.

If it shows a DPI under 300, I save as TIFF and use PhotozoomPro, withOUT sharpening, for resizing. Final sharpening as per above.

This method is good enough to easily take a Fuji-S2 landscape and crop to an 11×14 portrait print using about 3/4 of the original frame height.

The main drawback is the need to make "pre-sharpen" files at all the print sizes you plan on using before any sharpening. At least for me it takes that.

I got to test the previous GF vrs PhotozoomPro on a 3 meg. pixel file (Nikon-990 highest quality JPEG) I had of a cypress tree with Spanish Moss that was back-lit. I had done a slight crop. Not too much to play with. GF gave a lot more fringe color and the twig & moss edges weren’t as sharp. PhotozoomPro had about 10% of the fringe with sharper edges. Final print is 12×18 Epson 2200 Enhanced Matte.

As I said in the beginning, this was the PREVIOUS GF and the latest may be a lot better.

Bill
DF
David_Freed
Dec 26, 2006
When you say: "Use the largest available file size in ACR" are you referring to the Adobe RAW converter?

And I don’t have all these programs/plugins you mention: FocusMagic, PhotozoomPro, NoiseNinja (though I have another noise reduction program), but I’ll check them out.

I assume PhotozoomPro is similar to Genuine Fractals (GF is on sale till end of the year for $80 US)?

I’m also sizing my images at 360 DPI for Epson printers. I used to use 300 DPI, but now use this since it matches the printer’s native DPI.

Best, David.
CF
Cecil_Fuselier
Dec 26, 2006
I am new to the forums (two days) and cannot tell you guys how much all of this has been helpful. I still cant install CS2 because of "Error 1933" message (under separate thread). Again, these threads are so helpful.
Cecil
DF
David_Freed
Dec 26, 2006
Cecil, I was going to suggest a separate thread for your CS2 error message but I see you’ve done that.

Best, David.
C
chrisjbirchall
Dec 26, 2006
I’m also sizing my images at 360 DPI for Epson printers. I used to use 300 DPI, but now use this since it matches the printer’s native DPI.

David. I’m afraid you are confusing DPI (the number of dots of ink per inch your printer lays on the paper) with PPI (the number of pixels per inch that define the resolution of your image file). Both are entirely different entities, and are not interchangeable.

The 360 DPI setting on your Epson printer is the lowest resolution setting and is generally used for printing text documents.

720 Dpi is better. But for photographic quality you need to be using 1440 or (if your printer supports it) 2880 DPI.

That does not mean you should prepare your files at 1440 PPI. Anything higher than 300 ppi is completely wasted and for an inkjet printer 240 ppi is generally more than acceptable.

You printer driver will interpolate the RGB image information "on the fly" as it converts it into the CMYK information required to control the application of the cyan, magenta, yellow and black inks.

To use your original logic (but with the correct mathematics) assuming your printer has six inks: two shades of cyan, two of magenta and one yellow plus, of course, one black. Each of those six inks are (potentially) used to lay one "dot" on the paper. 1440 divided by six equals 240. Therefore little or no further interpolation is required of the image data at this resolution.

In theory, increasing the ppi resolution of your file above this level could result in poorer quality. In practise, you’d never notice.

Season’s Greetings. 🙂

Chris.
C
chrisjbirchall
Dec 26, 2006
Just to go back to your original post:

I do remember reading in the ACR forum some time ago that is was considered marginally better to open the file in ACR at its generic size and use Photoshop’s Bicubic algorithms ("Sharper" and "Smoother") for downsizing and upsizing respectively.

Chris.
DF
David_Freed
Dec 26, 2006
Sorry for DPI/PPI confusion.

Yes, I understand using 720 or 1440 DPI settings on Epson printer (which my Epson R2400 calls "photo" and "best photo" quality settings I believe); but I’ve been told that the image PPI (which I often call DPI by mistake) should be a factor of these, so 360 PPI is good minimum setting (e.g. 360 x 2 = 720) and is not ‘wasted’ on the printer. If you use 240 or 300 PPI–which are not factors of 720–then the printer does the interpolation, but it’s not as good a quality.

I found the basic info. about this at Inkjetart website newsletter: <http://www.inkjetart.com/news/archive/IJN_09-27-05.html#11>

Of course I’m open to all possibilities and to learning more, but I’ve found Inkjetart folks to be pretty consistent in the quality of their informaion. I’m getting them to do a printer profile right now.

Best, David.
B
Bernie
Dec 26, 2006
If you use 240 or 300 PPI–which are not factors of 720

Actually, 240 is a factor of 720 (3 x 240 = 720)

And the site you link to seems to assume that if the resolution of the image is not a factor of the printer’s resolution, then interpolation takes place. Which is probably a flawed premise since many RIPs do not interpolate images before processing them and I quite seriously doubt the built in RIPs of most inkjet printers do so.
DF
David_Freed
Dec 26, 2006
"and I quite seriously doubt the built in RIPs of most inkjet printers do so"

So, are you saying that the image is sent to the printer at it’s original PPI, and the printer only ‘sprays’ on ink at 720 or 1440 DPI?

Best, David.
BL
Bill_Lamp
Dec 26, 2006
YEP. You got it!!!!!
DF
David_Freed
Dec 26, 2006
Thanks Bill.

In truth, besides the 300 PPI that I was using, I’ve never tested or compared all of these other file resolutions (PPI / LPI): 240, 360, and now 288, but I’ll give them a go sometime.

Seems to be way more complicated than what they show us in the ads :>)
C
chrisjbirchall
Dec 26, 2006
…flawed premise since many RIPs do not interpolate images before processing them and I quite seriously doubt the built in RIPs of most inkjet printers do so

Oh but they do, Cyb, they have to!

Interpolation is a method of constructing new data points from an existing set of known data points.

A printer has to take the image data and interperate it (perhaps that is a better word in this instance) into signals to pass to the ink heads.

A printer has to take the 300 (or whatever) ppi from the image file and translate it into the 1440 (or how ever many) dots per inch laid down by the six (or more) ink jets.

Therefore it interpolates.

Bill: Using LPI will only confuse the situation further I’m afraid. digital images are made up of pixels (picture elements).

David: All this is pretty academic, you know.
When you do "give them a go sometime" I wouldn’t mind betting you’ll not be able to tell the difference between images printed from 180, 240, 300 or 360 ppi files.

Anyway – shouldn’t we all be eating mince pies and drinking wine? 🙂

Chris.
BL
Bill_Lamp
Dec 26, 2006
* goes back to drinking eggnog with brandy*
B
Bernie
Dec 26, 2006
Chris,

I agree with what you say, but:

I said: "many RIPs do not interpolate images before processing them"

I know some sort of resampling/interpolation needs to occur during the RIPing process, but this "interperating" (nice word BTW) would happen no matter what resolution your file is at.
MD
Michael_D_Sullivan
Dec 27, 2006
I just did a test that shows the Epson driver, at least, does interpolate. I made a grid of lines — in the first row, they are alternating 1px black, 1px white, in H, V, and crossed. Next row the lines are 1px black, 3px white; the third row has lines that are 2px black, 2px white. Here’s the grid:
< http://www.pixentral.com/show.php?picture=15Zk9TQsR4b2yWunMm qQmcOjM3J50>

Next, I set ppi to each of several resolutions in ppi and printed at 100% with the printer set to 1440dpi (photo resolution), then scanned in at 1200ppi. Here are the results (make sure to click on the thumbnails; the thumbnails have their own artifacts):
180ppi: < http://www.pixentral.com/show.php?picture=1MPzmWaV0nEYUXFbgs 6KDWNlsClGDC1>

200ppi: < http://www.pixentral.com/show.php?picture=1DN3Vsna3QFx9pVbZ3 CTzHzDWrWCH>

240ppi: < http://www.pixentral.com/show.php?picture=1fzDJkYZDpZkvn2pu5 DPewR7OMmVO1>

300ppi: < http://www.pixentral.com/show.php?picture=10DuSKERYflnZU3ZQA u6AOkkUabJqZ>

360ppi: < http://www.pixentral.com/show.php?picture=18jqnJS6feU0JfR3PE Rb4NCUTE2GT>

The images for which the dpi rating is an integral multiple of the ppi (180, 240, 360) are printed with fewer artifacts resulting from interpolation than those printed at a non-integral multiple (200, 300). Notice in particular the banding that appears to occur in the uppermost section. So, yes, the Epson driver is indeed interpolating in order to scale the ppi to its dpi.
DF
David_Freed
Dec 27, 2006
Michael, wow, thanks so much, and quite a bit of work you’ve done. I’m wondering, what would 288 DPI look like (since 1440 divided by 5 equals 288)? In other words, would it behave like 180, 240, 360 with fewer artifacts? And can you see any difference in image quality between these numbers? For example, does 360 DPI give me a better print than 240?

Best, David.
C
chrisjbirchall
Dec 27, 2006
can you see any difference in image quality between these numbers? For example, does 360 DPI give me a better print than 240?

Can you see any difference? Just examine Michael’s results. They concur precisely with my argument back in post #7 that in the case of a six ink printer working at 1440 DPI the generic "input" resolution is 240ppi, not the 360ppi quoted on the inkjetart site. Also, that feeding it with higher resolution file can actually reduce quality due to the interpolation that takes place. Ironically, the 180ppi result is pretty damn good too!.

All that said (and an extra portion of plum pudding to Michael for his visuals 😉 ), you really would be hard pressed to see the difference with a normal photograph.

What this does, however, is reinforce the argument that for ultimate quality you should only ever resize an image once. That this should be done as the last step in the editing process (prior to sharpening of course) and that if possible the target resolution should be the generic input resolution of the printer.

Chris.
B
Bernie
Dec 27, 2006
I agree with Chris.

My conclusion is that what we are seeing is not the effects of interpolation, but the effects of applying an algorythm for continuous tone images to line art.

If you use a laser printer with a traditional halftone pattern, the effect becomes even more obvious.
DF
David_Freed
Dec 27, 2006
Can you see any difference?

Yes, but mainly I noticed the images are different sizes, which makes comparision difficult. Also–and not to discount the findings–we’ve got an image, then printed, then scanned, then I assume scaled, then posted on the web at different sizes?

That said, I believe that I have much of the info. I needed from posting my question originally. Giving these options a test on my printer is the next step.

Thanks to all.

David.
MD
Michael_D_Sullivan
Dec 27, 2006
No scaling involved. I printed the same pixels in 180-360 ppi resolution, which makes the printed images different sizes, then scanned them all in at a much higher resolution (1200ppi). The posted versions are not scaled at all; I simply cropped the 1200 ppi scans by reducing the canvas size by 40% H and V. So, yes, the scans are different sizes, because the printed grids were different sizes. The scans are all at the same scale.
DF
David_Freed
Dec 27, 2006
No scaling involved. I printed the same pixels in 180-360 ppi resolution, which makes the printed images different sizes . . . .

Okay, but wouldn’t it be a better comparision to print at same output size (for example 2 x 2 inches, or whatever) but different PPIs? For example, a 2×2 inch image at 240 ppi (480x 480 px), at 288 ppi (576×576 px), at 300 ppi (600×600 px), at 360 PPI (720×720 px), and so on.

That’s the comparision I’m imagining doing myself to test out all the PPI options. Again, thanks, David.
C
chrisjbirchall
Dec 27, 2006
but wouldn’t it be a better comparison to print at same output size…

No, because that would involve interpolating the image in Photoshop before sending the file to the printer at different resolutions. That in itself would introduce yet another variable, which is what Michael has striven to do in the interest of a meaningful test.

The very best test would be to make your own test file (or download Michael’s) and view the results first hand thus eliminating the scanning process.
MD
Michael_D_Sullivan
Dec 28, 2006
Another experiment:
I took a standard test image, set its resolution to an odd figure that isn’t a multiple of any of the resolutions it would be converted to, and then resampled it to 180, 200, 240, 300, and 360 ppi, in each case with a width of 3.5" so it could be printed on 4×6 paper. Scanned all of those in at 1200 dpi, then selected identical areas from each for comparison. The results are here:
< http://www.pixentral.com/show.php?picture=1epZWumYwon45GZ2Wv m6jUsDtm0IC1>

All I can say is, anyone who maintains 180-200 ppi can’t be distinguished from 300 or better hasn’t looked up close. (The scan is 1200 ppi, and it’s being displayed about 75-100 ppi on your monitor, so it’s enlarged about 15x or so.) The pictures all looked virtually indistinguishable in quality at a normal 12-18" viewing distance, however.

Apropos of our earlier discussion in this topic, it seems to me that the 200 ppi images are no better than the 180 ppi images in most respects, and are considerably poorer than the 240 ppi images. All in all, however, it seems that for print resolution there is continued improvement as resolution approaches 360 ppi on a printer producing 1440 dpi. 300 is not an integral divisor of 1440, but it is noticeably better than 240 in some parts of the image. So I’ll eat my own words; you don’t need to print at a ppi resolution that is an integral divisor of the dpi.

How to Master Sharpening in Photoshop

Give your photos a professional finish with sharpening in Photoshop. Learn to enhance details, create contrast, and prepare your images for print, web, and social media.

Related Discussion Topics

Nice and short text about related topics in discussion sections