Resampling vs scaling


#1

I avoid the possibility of introducing resampling artifacts by unchecking that option in PS/Image/Image Size. This means that when choosing a resolution of 360 dpi, the print size (from an uncropped image) will be slightly larger than 10" x 15". This size satisfies much or my needs - but not all. Not-resampled images printed at higher resolution result in a smaller print (because one is matching dpi and ppi.) So, is it possible to get a larger print, at a higher resolution, while avoiding the artifacts that might be introduced by the resampling algorithm?

Jon, in his manual, states, “In fact, anyone will tell you that anything more than 360 ppi is wasted on the Epson printer driver. But, printing at 1000 ppi with Piezography actually results in significantly more detail. You would not want to add this resolution to the image file by resizing. Resizing is actually less effective as a way to increase the printed image size than simply scaling in the printer driver.”

That sounds like I can have my cake and eat it too. But I’d appreciate an elaboration of Jon’s statement. How does scaling differ from resampling (resizing*)?

Harry

*I’m assuming that resizing=resampling.


#2

resampling adds pixels.
during printing when you wish to scale the image up without increasing pixels prior (which necessarily reduces optical quality), the image is reproduced by using more extremely tiny dots of ink - rather than averaging more pixels.

when Epson prints they use a dithering algorithm which necessarily uses the white paper to simulate having lighter inks than they really do. With Piezography we are only printing image data that you can see at the points in the dithering process in which white paper (space between dots) is not used.

So where Epson does not print any information (the space between dots of ink), Piezography is printing information.

complex to be sure. always try printing your native resolution before resampling more pixels… compare for yourself and choose which you prefer! it does defy the “experts” advice on printing - but they are mostly regurgitating information from the 2000s from one book to another. very few experts who print have enough time to write books on printing! same was always said of drum scanner operators. and no guide book was ever produced as a result.


#3

I use the QTR Print Tool and load the native file and change the dimensions to the size I want to print inside QTR Print. Or is it better to change the image in PS (PS>Image>Image size) without resampling and print that unchanged with QTR Print Tool?


#4

I do not believe that you will see a difference. Neither way resamples. Make a test and let us know!


#5

Before I go and run some tests, :-), I am curious if anyone has run the test Jon / Michael M have suggested? If not, I guess I will run some tests… (And as I write this QTR Print is now my standard print workflow on MacOS. And I bring my master image file into QTR at 100% native resolution from PS CC at 360 ppi. And, so far, I have never printed below 360 ppi. Normally, for proofs, on 8-1/2 x 11 paper, I scale my image using QTR Print and the ppi resolution increases above 360 ppi. But since I am thinking about printing larger then 100% native resolution I want to figure out, for my own prints, how low of ppi I can go and still be satisfied with the print.)

Jon: I know that you have previously posted printing large piezography prints (24x24 or 24x30) from iPhone capture with 180 dpi or even 120 dpi and being very satisfied. Have you ever run any tests to determine the lower limit of dpi for piezography prints, i.e., where there are obvious visible signs of pixelation or other artifacts? Or is the answer really up to the photographer to determine what they are looking for in their final print?


#6

I have tried this test once or twice, but I haven’t formed a firm conclusion. I think it needs the right image and you also have to print fairly large to fully see the effect. I’ve got an image to print larger in the next few days, and may try this little test on that one.

This thread may be of interest in relation to the theory of what QTR does or doesn’t do in terms of resampling:
http://www.inkjetmall.com/tech/showthread.php?1833-1000dpi
Note especially the post by Roy Harrington that I linked to and quoted. I regard the conversation in this thread as incomplete.


#7

[QUOTE=Michael;8943]Jon: I know that you have previously posted printing large piezography prints (24x24 or 24x30) from iPhone capture with 180 dpi or even 120 dpi and being very satisfied. Have you ever run any tests to determine the lower limit of dpi for piezography prints, i.e., where there are obvious visible signs of pixelation or other artifacts? Or is the answer really up to the photographer to determine what they are looking for in their final print?[/QUOTE]

We’ve been fortunate to work with some very smooth low resolution files. And we have printed to 90dpi on some images that were destined for exhibition.

On the other hand, we have been sent 500+MB files that were horribly prepared by over-sharpening and with poor masking lines and defects from resampling filters that absolutely fell apart when printed.

The moral is that with images where the histogram is not jumpy - you can get away with lower res. And with high res files in which less skill was used to image them - they can fall apart quickly.

The iPhone apps do imaging just after the time of exposure. Much of it is just at the limit - but not over. So Hipstamatic filtering survives the test at their maximum of 2400 or so pixels.

A lot really depends on how you image and adjust your images. If you interrupt the smooth tone and make it jumpy it will print jumpy. If you do not - it will be difficult to see the pixels until sub 90 - maybe even lower.

Make sense?


#8

Jon: makes perfect sense. And that leads to the answer to my question, the inevitable: “it depends” :slight_smile: – (And I forgot to mention previously another factor is viewer distance from the print. Pixel peeping a 30" x 40" print meant to view at 6 feet or greater certainly doesn’t qualify in my book as a valid test case for this discussion.)


#9

[QUOTE=Michael;8943]Before I go and run some tests, :-), I am curious if anyone has run the test Jon / Michael M have suggested?[/QUOTE]

I’m not a Mac user, so I can’t attempt the exact same proposed test. But I’ve done a similar test in Windows. I have an image that I just printed at 13"x19’ for an group exhibition. Taking Jon’s usual advice I sent the optical resolution as-is to QTR at 251.5 DPI. Based on what Roy Harrington has said publicly, QTR would have resampled this to 720dpi prior to printing (on a MAC this is done by the OS as I understand it).

Next I resized a copy of the image to 720dpi using Qimage, which in the Windows world has a reputation for having the best possible upsizing algorithm. In order to make this a valid comparison, I had to turn off CM and any sharpening in Qimage. I printed from Qimage to a TIFF file and then printed using QTR.

And the result? There was more detail, i.e. better resolved detail in the Qimage version compared to the version at optical resolution. Was there a lot of difference? No, not at all. You really had to look for it but it was there. In this instance the resampling to 720dpi that Qimage performed was better than that performed by QTR. Was it worth the extra trouble? No, probably not, but if you want to wring out every last ounce of detail then maybe.


#10

One of the things that we have done at Cone Editions Press is to freeze all of our production Mac Operating Systems to 10.4.7 This was the last version prior to the Apple experiments of image manipulation in regards to color management and printing that began on 10.5 and haven’t ceased since…

So, we benefit even more by having lots of optical resolution when we can take advantage of it. Also in our Classroom we keep the five 3880s on older systems…

You can duplicate this by buying a $150 Mac-Mini on eBay that has not been tampered with and has an older Mac operating system. Then you NEVER FREAKING UPGRADE IT! Resist every evil temptation to connect better to your iCal, iPhone, iTunes and the iCloud by upgrading. Every upgrade from Apple moves you just a little further away from your having control.

Me highly recommend putting any printer on a Print Server that is separate from the device that controls your (i)Life and your (i)Entertainment and your (i)Connection to (i)Facebook. The (i) is really an acronym for (They). And you know what They say? Right?


#11

As you know, I’m not on a Mac. But my understanding is that on a Mac there are two issues. One is what happens to CM post 10.6.8, and the other is the resampling to 720dpi. My understanding from what Roy has posted is that the resampling is done by MacOS and happens no matter what version of MacOS you are on. Based on what Roy has written, I don’t think you’re immune from it on 10.4.7, even though you are immune from the CM issues. But then I’m not a Mac expert. Perhaps I don’t fully understand all the MacOS changes, but that statement of Roys didn’t suggest that this aspect of Mac behaviour had changed.

On Windows the resampling is done by QTR rather than the OS. I haven’t tested whether I can resample down to 720dpi better than QTR does, but what I reported is whether, on Windows, I can resample up to 720dpi better. I think yes, but only a little.


#12

Jon, since QTR Print does not work with 10.4.7 what application do you print from if setting up a Mac Mini? Neither does PS CC or LR CC or… I feel like Alice down the rabbit hole…


#13

[QUOTE=Michael;9020]Jon, since QTR Print does not work with 10.4.7 what application do you print from if setting up a Mac Mini? Neither does PS CC or LR CC or… I feel like Alice down the rabbit hole…[/QUOTE]

That’s the beautiful thing about it. We print directly from Photoshop.

You can buy a 10.4 Mac-Mini with PS 4, 5 or 6 already on it.

Right now on eBay there is a 10.4 mac-mini at $89 or a lot of six of them for $129.95 which comes out to about $21.50 each. Cheap enough to look for a copy of PS 4, 5, or 6.