I got the D800 instead of the D800E to avoid issues with moire. And yet, with the first event that I shot using my new D800 I ended up with a nasty case of it. See the attached section of a photo from my Wednesday night shoot.
I only once saw this problem with my D700, and it was subtle. Was the D700 less prone to the problem, or was I just very lucky to not have photographed the "wrong" subjects?
Anyone know how to get this fellow's suit straightened out? I'd like to fix this before giving him the picture. Incidentally, the problem was greatly after I downsized the file to post here, so maybe there is a hint in that? It was WILD before I downsized. Thanks!
To me this verifies what I am finding. The AA filter in the D800 is rather weak. I have not seen a need for the D800E, even in Landscapes, so far. It is very, very sharp.
I found moiré the first day I was shooting with my D800. I took a picture of a pair of black pants lying across the room (you know one of those important test shots) and I had false color all over the place in the pant's pattern.
Moral of the story. I don't need the D800E. My camera is plenty sharp and quite capable of making moiré. I shoot weddings and can't take any chances.
============================================== Darrell Young (DigitalDarrell) www.pictureandpen.com "Better too many words than not enough understanding." ==============================================
Sat 05-May-12 04:04 PM | edited Sat 05-May-12 04:16 PM by MotoMannequin
I'm not seeing any obvious moiré here, but I guess we're not seeing in this sample resolution what you're seeing in print. Is it possible to show a 100% crop which exhibits the moiré?
edited to add: at 200% I do see some color banding in the lower right, and maybe some diagonals in the upper left. Probably some color artifacts throughout. I'd still like to see a 100% crop to isolate things that may be done by my viewer's inerpolation algorithm. The color stuff is pretty easy to clean up, but I can certainly understand why you wouldn't want to run the chance of going through this on every shoot.
Does this show viewing at 100% on screen, or is it just a problem for the print? To show us your 100% view, you need to take a small crop out, and be sure it doesn't re-interpolate. In PS just crop and leave the DPI blank. Don't know about LR4 but I'd think it's similar.
I'm also curious if you could convert using the Nikon converter, either with trial of Capture NX2 or the free conveter in View NX2.
Another thing of note - certainly this image does some really bizarre things when I view it on screen at 150% or 50%, so interpolation can really hose this up (or de-hose it, as the case may be in your sample aboev). Of course I'd normally think the 100% view is the one that matters, but thinking about this further, your print driver is going to re-interpolate to its native dpi. For Epson, that's 360dpi, and AFAIK 300dpi for most others. I wonder if you improve your print by re-sizing in LR to the printer's native DPI first?
>This is good advice ... I have actually been wondering if >there was a tool specifically made for re sampling images >better than LR or Photoshop. > >Anyone ?
Back when I was first getting into printing big, I did a bit of experimentation and found 3 methods that worked pretty well: - PS upsize using Bicubic Sharper - PS upsize using Bicubic Smoother, then apply sharpening - Upsize using Genuine Fractals
Genuine Fractals worked the best in most cases, but the results weren't necessarily conclusive in some cases, and highly image-dependent. Some images worked better with other methods. At the time I was playing with a trial version of GF and I remember thinking the rather steep price ($600+?) wasn't worth the occasional incremental step up in quality.
It has been suggested to me recently that a product called Perfect Resize is worth checking out, and then by someone else that Genuine Fractals is now called Perfect Resize. I just looked further into this and turns out it's true, and it's now quite a lot cheaper than it used to be. I haven't tried it so don't consider this a recommendation, except that I'd say the trial version is worth investigating.
Sun 06-May-12 03:04 PM | edited Sun 06-May-12 03:05 PM by walkerr
It does a nice job for really large print sizes (I've owned it through several incarnations leading up to and including it being called Perfect Resize; I have the current version). Its strengths are really in making really large prints (rather than downsizing) where it does some clever things and those strengths really start showing up at around 2x the native resolution in my experience. For a D800, that would be an absolutely enormous print. Having said all that, I'd be happy to take a look at the original NEF and try a variety of things using LR4, Photoshop CS6, Perfect Resize, etc. I'm kind of curious what's driving the issue, and it would be good info for me to to have, too.
> Does this show viewing at 100% on screen, or is it just a >problem for the print?
If you view the file at the pixel level (cntrl-alt-0 or cntrl-1 in Photoshop) you don't see the pattern. In fact, that's the only way to view it on my monitor where you don't see ANY moire.
> >Another thing of note - certainly this image does some really >bizarre things when I view it on screen at 150% or 50%, so >interpolation can really hose this up (or de-hose it, as the >case may be in your sample aboev). Of course I'd normally >think the 100% view is the one that matters, but thinking >about this further, your print driver is going to >re-interpolate to its native dpi. For Epson, that's 360dpi, >and AFAIK 300dpi for most others. I wonder if you improve your >print by re-sizing in LR to the printer's native DPI first?
I will look into that. Perhaps that's a solution. Thanks!
> >> >>If you view the file at the pixel level (cntrl-alt-0 or >>cntrl-1 in Photoshop) you don't see the pattern. In >fact, >>that's the only way to view it on my monitor where you >don't >>see ANY moire. >> > > >That means the image doesn't have moire. It's the interaction >of the image and your monitor and not a camera problem. >
But what will I do about the horrible looking print? Is it a matter or resizing the image to get rid of the "not moire" that looks so terrible. As I said in my prior posts, that's the big issue. People expect a print from me, and they don't expect wild squiggly patterns--whatever they are properly labeled. Thanks!
I'm sorry that I couldn't get back with you yesterday, but I was off shooting!
I'm using LR 4.0 to create JPG files that I print using Photoshop CS 5. I had LR producing 240 ppi files at the time (I've since changed to 300 ppi). I generally leave the file size alone and let PS rescale the image when I go to print.
Sun 06-May-12 02:14 PM | edited Sun 06-May-12 02:15 PM by MotoMannequin
>Larry, > >You're a genius!
I know, my mom had me tested
>I resized the image to 240 and 300 dpi, then printed at 100% >scaling, and it looked much better (though there was still >somewhat of a problem). > >I'm going to play around with this and see if I can totally >eliminate the issue.
I think that you can play around with the different resampling methods (bicubic, bicubic sharper, bicubic smoother, nearest neighbor, etc. these are avialble in PS, I guess they are in LR?) that gives you the best result, I think you'll find a solution. Also remember if you're printing Epson, native dpi is 360.
What LR does is more sophisticated than that. It automatically adapts the resampling method to whatever you're doing, centered around the bicubic method. Downsizing heads toward bicubic sharper the more you decrease the size, whereas when you increase the size of an image, it applies something that more and more resembles bicubic smoother. You also have the option of not resampling, and I suspect that could be the problem, especially given the D800's substantial resolution. It might be causing a problem with the printer driver.
I can't see the slightest evidence of Moire in your test image which looks quite wonderful to me. Note, however, my monitor is 2550 x 1460 (28 inches diagonal)and configured to work in native mode. Interaction between the image sampling interval and the monitor resolution can lead to a "false Moire" which is not actually present in the original image. Have you investigated this possibility?
I agree in general with folks who think the 800E is not going to prove much of an advantage when it comes to sharpness. It is always asking for trouble to avoid using an anti-aliasing filter. I am using the 600 and it is exceptionally sharp with the right lenses. Remember too, you can always recover most of the detail lost to the filter by using good sharpening software (*not* Unsharp Mask).