Pages: [1]   Go Down

Author Topic: Scaling up advice  (Read 1790 times)

stephen23

  • Newbie
  • *
  • Offline Offline
  • Posts: 5
Scaling up advice
« on: July 11, 2007, 11:37:14 pm »

Hey everyone,

I'm shooting RAW on a Nikon D80, and scaling the images up slightly to get the size and dpi that the print shop I am using recommends. I'm not using any third party programs to scale up... just bumping up to the size and dpi I need in photoshop (CS2).

My question is, is there any benefit to scaling up by an even percentage as opposed to a percentage that might not adjust the pixels evenly? I think I remember reading somewhere that when scaling up or down, you'll get a better image if you scale up by, say 50% rather than 48%. Obviously the resulting image size is going to be very similar, but does photoshop handle the addition of pixels in a way that will make one image better than the other?

As an example, the D80 creates an image that is 3872x2592 which translates to roughly 8.5" x 13" at 300dpi. I'd like to make that a bit larger, so I scale up by 150%, giving me an image size of roughly 13" x 20" at the same dpi, the minimum for the lightjet printers I'm using. The end result is an odd size.... 12.96" x 19.36",  but that doesn't bother me, and the number of pixels added is a straight forward 50%. In this example, I guess the question is, is this 50% scale up going to give me a better image than if I'd just plugged 13" into the window and let it scale it to 51% or whatever that difference would be? On the one hand, it makes a certain logic to me, adding one pixel for every two results in a pretty simple process, but on the other hand, were talking about millions of pixels, and there is some heavy duty processing going on anyway, so what does photoshop care if it's increasing by 150% or 151%?

Also, I am wondering about the advice the print shop gave me.. Everything I've read on here says that when scaling up, it's better to go in small jumps, so the first couple tests I did, I used a couple of 120% increases rather than one at 150%. Perhaps even three, with the last one being a small step up, to get to a size I wanted. As per some advice, I also did some sharpening between those steps, but found that the resulting image had way too much noise. When discussing this with the shop, their "printing expert" who has a very good reputation around town, after asking what kind of camera I'm using, said "Oh... that camera has plenty of bit depth. You can just scale up in one step, no need to do it the hard way... then just sharpen a little after that, if you think you need it. I probably wouldn't even bother.." Then he also recommended that I might want to bump the images up as high as 400 dpi. I came to digital from a medium format background, and I would love to get that kind of detail, but doubling the number of pixels makes me nervous... I'd prefer to be a bit less detailed if it means adding 50% less artifacts or noise.

So... now I'm wondering, does this guy know what he's talking about, or is he just blowing smoke? And what's with the "bit depth" thing... The pixels have a depth of 8 bits, same as every other SLR, right? How in the world would one camera's 8 bit pixels have more "depth" than another to scale up better or worse?

I went into this thinking I'd done enough reading to have at least an idea of what I was doing, and would be able to learn the rest as I went, but now I'm wondering if I even have a clue! or maybe if I should just find another printer...

Thanks for the help!

Stephen
Logged

jbrembat

  • Full Member
  • ***
  • Offline Offline
  • Posts: 177
Scaling up advice
« Reply #1 on: July 12, 2007, 06:19:28 am »

Quote
My question is, is there any benefit to scaling up by an even percentage as opposed to a percentage that might not adjust the pixels evenly?

The answer is: it depends on resampling algorithm. For good algorithms, it is not true. Poor algorithms, like nearest neighbor , have a benefit using even percentage.
Quote
Everything I've read on here says that when scaling up, it's better to go in small jumps
That is wrong.
Interpolation creates pixels starting from image pixels.
If you resample in one step, every added pixel is created from original pixels.
If you resample in incremental steps, added pixels are created from previously created pixel too. Worse quality.
The myth of incremental resampling is due to a poor interpolation algorithm that is partially compensated by other interpolation steps. But the final quality is poor anyway.
Quote
Then he also recommended that I might want to bump the images up as high as 400 dpi
I think they mean PPI. For lightjet printers (contone printers), the DPI value is equal to the PPI value.

Jacopo
Is 400 the PPI value expected for printer driver? May be yes, may be no.
If your OS is Windows, you can go to: PhotoResampling
and download PrinterData. This utility can show the PPI, the printable area and margins exposed by printer driver. You can see these values and verify the changes induced by modifying the printer settings.
I suggest to read the article on DPI/PPI.
Quote
Oh... that camera has plenty of bit depth. You can just scale up in one step,
This is a quote from a review:
Quote
The processor also offers colour-independent analogue pre-conditioning and high-precision 12-bit digital image-processing algorithms. Nikon claims that this reproduces faithful colour and tones, for natural images.

12 bit raw is standard, and as you said jpg is 8 bpp.
Of course, starting from a better image is better, for resampling too. But bit depth do not make visible difference in resampling and, as you stated jpg is always 8 bpp.
Logged

stephen23

  • Newbie
  • *
  • Offline Offline
  • Posts: 5
Scaling up advice
« Reply #2 on: July 12, 2007, 08:41:46 am »

Okay... last night after posting this I did some more searching and I found several good posts on interpolation, especially one leading to an article testing bicubic smoother, genuine fractals and a few other programs and processes. I think I've got that covered.

As for the bit-depth question, I should have been more clear, sorry. When processing images, I am opening them in ACR as 16-bit and doing all my retouches and editing there. Only in the final step, when I am creating an actual print file am I saving the file as a flattened tif file and taking it down to 8 bits, which is what all the print shops here in Austin have recommended. At no point am I using jpeg in any way. In any case, I think that question is probably moot, since the above mentioned articles have given me good information on how far I can scale up and what kind of results I can get.

I'm not sure I'm clear on the answer you gave about the lightjet PPI values, but again, I think I was using too much shorthand. Yes, on a lightjet PPI = DPI, so I was just using that term instead. I will do some more looking around. I am working on a Mac Pro, but I'm not sure if the print shop is PC or Mac. I haven't read anything about the lightjets having a native target to shoot for, but I hadn't thought to ask what driver they are using.

It sounds like a targeted search on the lightjet specifics is in order.

Thanks.

Stephen
Logged
Pages: [1]   Go Up