| Author | Thread |
|
|
01/13/2007 10:12:45 PM · #26 |
Originally posted by Gordon:
Originally posted by wavelength:
Isn't that sort of analogous to telling a remedial kid, "It's okay Billy, you can't read and we're not going to suggest you try." ?? |
Only if you assume big = better. Why pretend something is what it isn't ? All you do is make it look bad. |
Well, agreed, but it's what the client wanted, not just bigger for no reason. Looks like Jason figgerd it out anyhoo.
|
|
|
|
01/13/2007 10:39:19 PM · #27 |
Originally posted by DrAchoo:
OK, here's evidence of why I was breathing faster before, but am breathing a bit easier now...
  |
Not to rain on your parade, but I'm not so sure the second image has that great of quality either. The good news though is that you are seeing it closer to how it is. Looking on the screen is a little deceptive since screens have poor resolution compared to prints. But you are seeing it at the level you need to for corrections that insure higher quality.
Looks over NeatImaged slightly to me and the big noise is still there. The battle between loss of detail and noise will be your biggest issue in printing.
This is a tough image to work with and I'm not exactly sure how you can do better with it. But you are experiencing similar issues to what I have with 16 X 20s.
As Gordon said, you need a high quality image to start with to get a high quality print. If you have to "save" the image then you've already lost the battle.
Another thing you will discover in some images, particularly with ones that have subtle tonal variations, is that you will eventually reach a pixel limit and will not be able to improve it. It will show up as pixelation in places like the sky. That isn't a camera limitation, it is a limitation of the color space of the image where there is a visible leap from one tone to another because the color space lacks a tone between the two. It can't be corrected in post processing. I know, I've tried. :) That is where "old fashioned" medium format film still holds the edge.
|
|
|
|
01/13/2007 10:44:42 PM · #28 |
| I agree with Steve that it looks over-NeatImaged. I know it's been upsampled, though. It would be interesting to see what this same area of the image looks like at 100% without noise reduction or upsampling... |
|
|
|
01/13/2007 10:46:37 PM · #29 |
Originally posted by jaysonmc:
Skip medium format all together. Go Large Format. You know, photography is just an arms race. :) Better yet, see if you can get with Polariod and use that 20x24 Camera that William Wegman used. If you ever see the prints from that thing, man are they big.
I guess in the end it all depends on how big you plan to go...
Your photos are fantastic as is, are you losing alot of quality with big prints? |
Screw it... go Gigapixel. |
|
|
|
01/13/2007 10:51:46 PM · #30 |
Whoever said it might have been joking, but if you are gonna go film, do go large format.
Your 5-D should compete pretty decently with medium format. I've seen 30d and even d70 prints that look almost as good as medium format.
The d70 looked completely photographic at 16x30 and the 30d looked really good with about 1 foot viewing distance at 30x40 inch print.
The 5d should be good enough to print that large, and I've seen examples in galleries from it already.
But if you are really concerned about noise (where film still wins at long exposures) You really should shoot 4x5, because medium format just isn't that much of an advantage over your 5d. 4x5 would be where it would get noticeable.
Both formats are really fun to shoot, but tons of good landscape shooters still use the view camera, but labs that process e-6 sheets are dwindling, I would check on that in your area to begin with, see how long you are talking for shipping. You also will need a light meter if you don't have one already, as those are exposures you can't screw up (unless you have dollars to throw around)
A good large format set up won't really cost you more than a medium format set up (possibly even less), as well.
|
|
|
|
01/13/2007 10:59:48 PM · #31 |
Overall, even at 20 X 30, I doubt upsampling will cause any problem whatsoever, it just isn't being upscaled all that much.
Noise reduction, however, is a big issue with this image. The problem is the large 'grainy' noise that cannot be reduced with noise reduction. That is always a big problem in underexposed areas of an image, like the basaltic overhang that is dark to begin with and is in shadow. It even gets hard to discern noise from real data. It is obvious it remains as a problem here because the image shows signs both of oversmoothing and grainy noise. If this were an important image to me I would spend hours hand smoothing the big noise. I'd also apply less NeatImage to start with. That was an issue with my Sony and always had to do that with large prints.
Message edited by author 2007-01-13 23:00:51.
|
|
|
|
01/13/2007 11:02:32 PM · #32 |
Originally posted by kirbic:
I agree with Steve that it looks over-NeatImaged. I know it's been upsampled, though. It would be interesting to see what this same area of the image looks like at 100% without noise reduction or upsampling... |
Your wish is my command. This is straight from the RAW after conversion...
|
|
|
|
01/13/2007 11:16:32 PM · #33 |
|
|
|
01/13/2007 11:17:19 PM · #34 |
Sorry bad advice. I've deleted.
Message edited by author 2007-01-14 10:46:07. |
|
|
|
01/13/2007 11:21:39 PM · #35 |
Originally posted by DrAchoo:
.........
as mentioned above, the gallery requires the ability to print at 30x40. |
Choosing the correct tool always makes the job easier. It appears something larger than 35 would be more appropriate for the task of the print size required.
|
|
|
|
01/13/2007 11:32:15 PM · #36 |
Doesn't it make a difference as to whether you use Adobe rgb or srgb? I heard that when printing you should use the adobe setting, but when that is looked at on screen, it looks pretty bad, so that is why on computer display we use srgb. Just wondering if printing it out will appear differently then what you are seeing on the screen. I did an experiment on printing an image 30 by 40 using adobe and just printing out an 8 by 10 portion of it, and there was little noise, though focus needed to be better. I will work on that one.
This is the image that I used for my experiment. I also saved it in
eps format before printing.
I think this is the right proportion for 30 by 40, at 150 dpi, though the print was in adobe, which this one isn't.
Message edited by author 2007-01-13 23:43:30. |
|
|
|
01/14/2007 10:08:51 AM · #37 |
Originally posted by wavelength:
Only if you assume big = better. Why pretend something is what it isn't ? All you do is make it look bad. |
Well, agreed, but it's what the client wanted, not just bigger for no reason. Looks like Jason figgerd it out anyhoo. [/quote]
Yup, that seems to be what the gallery wants, but that doesn't necessarily mean that it is possible to provide it, at a level that will look as good as medium format. 20x30s will look excellent, with an extremely good starting shot, handled well. It starts a downward slide from there on. You can get to larger resolutions, but the further you go, the more perfect the starting shot has to be. Maybe 1 in 10 will work well at 20x30, assuming great technique, maybe 1 in 20 or 1 in 30 will go larger. Numbers made up, but I hope the point is clear - the further you want to stretch it beyond what it is comfortable capable of, the more exceptional the image has to be. Fine details and subtle features suffer the most and show up the worst compared to say medium format.
It isn't saying medium format or the 5D are better or worse than each other, just one is maybe more appropriate for huge enlargements (and 30x40 is pretty darn big)
What I've done in the past is make 8x10 sample prints of sections of the image, at the target resolution/ resampling/ sharpening. I'll process the file the way(s) I'm going to print, then cut sections from the middle and corners, that cover areas of concern (fine detail, potentially out of gamut colours etc) and collage those sections onto some 8x10s at 100% size, that I then get printed (I'll put comments on the page about particular processing - e.g., with different sharpening settings and so on)
Then get the prints back and look at them at the expected viewing distance. That way you can really see how the final large print will look. I find that's been the only really reliable way to find out what it'll look like, without spending $100 per sample.
Message edited by author 2007-01-14 10:12:19.
|
|
|
|
01/14/2007 10:32:29 AM · #38 |
Let's also not forget that what we are looking at here at 100% is not what the spectators in the gallery will see when printed. Image 2 is just 2.88" of the total image at 250ppi. On my screen, to get this sized so that I am seeing it at the same size as it will be when printed, I have to view it at 36%. When I view it at 36% on my 21" LCD running at 1600x1200 what I am looking at is about 2.8" of the overall 30"x22" image, what the folks will see in the printed photograph. I also printed on my color laser to see how this small portion of the pic would look printed and it looks very good. I did this because when viewed on an LCD you are again limited to the number of pixels you get per inch so printing gives you a more realistic feel of how it will look when it comes back from the lab.
I agree that when viewed at 100% it looks overly NI and plasticy but when viewed on screen or printed at the size it will actually be printed it looks very good. I have used used Genuine Fractals to up size 8.2MP images to 30x40 and they look great when printed printed at that size, even though they look odd or down right bad when viewed at 100% on screen. People even comment how clear they are and how much better they look than film. However, most people without photography knowledge compare to 35mm and even some to 110 or disk. LOL!!!
The true test will be to have this printed by a good lab and check out the results. I'm sure you will be happy. Good luck with the show!!! Let us know how it goes.
|
|
|
|
01/14/2007 10:42:43 AM · #39 |
I really like Gordon's suggestions on test printing portions of the image. That seems to be an ideal way of really cutting to the chase and finding out how a particular size, processed a certain way, will look on the wall.
I played with the 100% direct-from RAW crop, and based on that I'd recommend either ditching the NR altogether or running it gently on only the deeper shadows. Also, I'd back off some on the USM, which is also contributing to a blocky look. Just upsampling (nothing complex, just basic bicubic) the original and adjusting contrast and applying an R=0.6, Amt=100 USM pass seems to produce smoother, more natural detail rendition. Using a threshold of 3 keeps the noise from being enhanced. The result, IMO needs little or nothing in the way of noise reduction. |
|
|
|
01/14/2007 10:43:22 AM · #40 |
If I had to consistently produce 30x40 images, I would likely look into alternate methods of shooting it digitally. I have had EXCELLENT luck with my panoramas using Panorama Factory software. I have NOT experimented with stitching images that weren't in a sequential left-to-right collection. I think it would be worth considering though. I think the software was $69. It paid for itself on the first panorama I sold. If you are selling in a gallery, I would consider giving it a try and reshooting something in multiple images to try this method.
|
|
|
|
01/14/2007 11:06:47 AM · #41 |
| William makes a good point, screen resolution is way worse than printed. A print should always look better than a 100% view on the screen. Even 150dpi is twice as good printed! That being said on the 5D, if you take an original sized image and print it at 150dpi you're getting about a 29 x 19 print. Really going to 30 x 40 isn't much bigger than that! You can achieve that a 109dpi from an original file. Upsizing to 150dpi+ however should give you something that looks great! |
|
|
|
01/14/2007 11:16:29 AM · #42 |
Thanks guys, William is right, the screen cannot duplicate printing. The whole thing came up when I printed an 8x10 crop of it on the original process. The detail was just not there, but I think the second (and now third) attempt are much better.
I'm going to print some more sections on Monday and see how they look.
One thing to remember is that for every 30x40 ordered, you are going to have 25 or 50 prints at 16x20 or 8x10 or something similar. So while the owner wants the ability to print that large, it's not likely to happen too often.
Message edited by author 2007-01-14 11:17:48.
|
|
Home -
Challenges -
Community -
League -
Photos -
Cameras -
Lenses -
Learn -
Help -
Terms of Use -
Privacy -
Top ^
DPChallenge, and website content and design, Copyright © 2001-2025 Challenging Technologies, LLC.
All digital photo copyrights belong to the photographers and may not be used without permission.
Current Server Time: 09/09/2025 05:46:41 PM EDT.
