Can enlarge w/ Photoshop 110% with great results?

Discussion in 'Digital Darkroom' started by nancy_moore, Jul 30, 2003.

  1. Hi,
    I was told I could enlarge a 5X7 image (must be at least 3 mega-
    pixels)to poster size with crisp clear results in Photoshop if I only
    enlarge it 10% at a time. Anyone else aware of this nifty trick, and
    could you enlighten me on how to do this? Thanks!
  2. The results are slightly better this way than increasing the size to the final output in one shot. Go to Image>Image Size, then increase the length and width by 5-10% until you reach your output size. It will take a while this way, but you can increase your final image to be printed. I don't believe you will be able to reach crisp poster sized images, but they will be acceptable.
  3. I donot know where this rumor came from, but it is not true. I have included a JPEG, one which I increased the size to 250% in 10% steps and the other in one step of 250%. As you can see, there is no difference, and I did not include which is which on purpose. Chances are someone got a decent interpolation on one image using stepping, got excited and started imaging that the result were better in all images, but never bothered to make a comparison. You cannot pull data from nothing.
  4. This "myth" comes up occasionally, and only 2 weeks ago there was a thread asking if you can improve results when *down-sizing* by using this method (similar arguments apply)
    The consensus is that the results may perceived to be "better", but there really isn't much difference. The topic is complicated by the fact that there are many different methods and filters used for resampling (covered in that thread).
  5. <<fact that there are many different methods and filters used for resampling>>

    I have seen RealFractals ect, and I have not been impressed AT ALL.

    Like I said in another post, pulling information from thin air is the alchemy of
    digital photography.
  6. I very much agree, Genuine fractals is highly overrated. However different methods do seem to perform better with different images, although the quality is still very similar.
  7. <<The consensus is that the results may perceived to be "better", but there really isn't much difference.>>

    Maybe after stepping 20 times at 5% increments to get a 1,000% enlargement you wish so much that you did not waste the afternoon doing it you start to hallucinate improvements?
  8. The way to do it is with a script, eg. Photoshop action (Fred Miranda has one for sale, with all kinds of small improvements to the basic idea) - if you have the time to do it by hand, you have too much time on your hands!

    I think the method is overrated - a good interpolation algorithm (eg. Lanczos) will produce comparable results with one step - saves time, even with scripts. There's no silver bullet; if you want a large, sharp image, you need a big file to begin with.
  9. Shawn asserts, "as you can see, there is no difference." I can see better than that. The top has more detail, the bottom is smoother, and here is the ImageMagick "diff" to prove it. Only black pixels are identical, and there aren't many of them!
  10. It does not suprise me that there is a difference on an analylitical level. I did
    after all say "as you can see". The original is different in each iteration, so the
    placement of interpolated pixels might differ. I am quite suprised of the
    amount of difference. I think that this is a matter of placement rather than a
    matter of retained useable detail.

    However, I fail to see your point on the "bottom one being smoother" atleast
    not significantly. Also, is smoother always "better" in regard to enlargements?
    I *may* see very slight increase in edge deffinition on the lower one (which i
    think is the stepped version) and your analysis would suggest this as well I do
    not think it is enough of a *visual* difference to spend even 10 cents on a
    script, which, btw, I could write in about 5 minutes and will do it for free with
    greater flexability thant the ones available. If you are really interested and use
    a Macintosh... just a matter of finding five minutes to write the macro might be
    a little bit tough.

    Call me old fashioned (at 22) but I really think that if you want a bigger print,
    just get a bigger file. If you're using film, get a bigger negative. If you are using
    digital get a bigger CCD. It's a lot easier and the results are considerably
  11. Also- is it impossible that the differences are in JPEG compression or in
    registration errors on your part?
  12. It's late, nevermind what i said about smoothness. I see no difference in detail.

    Also upon another examination, I think the "edge deffinition" may have been
    chromatic abberation in my eye glasses.
  13. Could be JPEG artifacting, but I don't think it's registration,
    otherwise the horizon-line diffs would be greater. The areas where
    I see big differences in detail are the puffy clouds, and the sand
    drift lines on lower left. So if you're correct that the bottom one
    is stepped-up, stepping gains smoothness at the expense of detail.
  14. There is a SIGNIFICANT difference in the photos above. The bottom one CLEARLY has a big white blur spot on the top-right corner! Are you guys BLIND?!

    Oh wait - I had something in my eye. They're identical.

Share This Page