NikonGear'23
Gear Talk => Processing & Publication => Topic started by: simsurace on May 19, 2016, 11:13:06
-
I have heard from a few people that they are experiencing suboptimal results when trying to downsize a high-res image -- say, from a D810 -- to a lower resolution (for web use). The downsized images are reported to have certain qualities which are not pleasant, but sometimes hard to describe.
It has been said that software has not kept up with this challenge and that it is hard to find the correct downsizing method for those images.
Another related problem is that high-res images are sometimes badly displayed by some software for previewing/culling. Since the software has to downsize the image to display it on a monitor with a much lower resolution than the image itself, the same downsizing challenge applies behind the scenes.
Of course, we have to realize that any digital image cannot properly represent detail which exceeds the Nyquist frequency, therefore we might have to reset our expectations accordingly. But we should not be getting worse results out of the high-res file compared to a lower-res file, e.g. the same scene shot with a D800 vs. a D700 should not look worse when downsized to web resolutions unless the downsizing was done in a suboptimal manner. What I understood from some people is that they are finding it hard to match the results obtained with lower-res cameras, and this is where this thread comes in.
I think it is important to thoroughly understand the problem first, for there might be several individual problems for which separate solutions have to be found. Personally, I have not experienced these problems (despite not having a 36mp cam, I have worked on panos that measure several hundreds of megapixels and I have downsized them quite successfully IMHO), so I might first need to be educated about the problems others are perceiving.
Here, my aim is to collect examples of downsizing to try and identify the problem and how it manifests itself. If you are experiencing problems when downsizing high-res files, please post an example. If possible, please make available the raw/full-res file (e.g. on dropbox or via some other hosting service) as well as your attempt at a downsized version (jpeg). If you can, please try to describe what you don't like about the downsized version.
When we have collected enough examples, we will try to find a workable solution to solve the problem. Other forum participants will be able to take your full-res file and come up with their downsize, and you will be able to judge whether an improvement has been made over your own attempt. The concept I have in mind is very similar to the RAW editing challenge, but more specifically geared towards downsizing.
-
I think a lot of it has to do with how the image is sharpened.
-
I'm very interested in this. I'll get back with a couple of examples this weekend. I'm still getting caught up on the domestic front after being gone 2 weeks on the Scotland NG Meetup.
-
I am experiencing the same issue with downsizing. Images don't appear sharp. 99% of the time I dont have to sharpen my images at all - (when processing the high res) so for me it can't be an issue due to how I sharpened
-
For my girlfriend (not able to post here herself due to lacking English skills, especially when it comes to technical explainations) the biggest problem is tha images that appear perfectly sharp on a pixel level on her computer, even when converted from NEF to JPG, appear completely out of focus when subsequently uploaded to facebook. Since she uses facebook as the main portal to portray her work we have been looking into a solution for this for quite some time.
It does not happen with all images, but a considerable amount.
To the tehcnical details, she uses a D800 to shoot lossless compressed 14 bit NEF files
I will make a file available later today. Need to upload it to dropbox which given my slow internet connection will take a while.
I will also post a link to the uploaded file on facebook.
I think this here will be a good example.
https://scontent-fra3-1.xx.fbcdn.net/t31.0-8/13115992_878889058903527_8707603759883841782_o.jpg
Any help in how to prepare the file for an upload on facebook and ending up with a representation of the file that appears sharp is greatly appreciated.
Here is the NEF to the picture
https://www.dropbox.com/s/ozr8i2zt04pwhy3/DSC_5551.NEF?dl=0
And the processed JPG
https://www.dropbox.com/s/04z2me0zyeng1yc/DSC_5551-Bearbeitet.jpg?dl=0
-
I learned long time ago, talking to Graphical designers and printer guys, to shoot/stitch at the desired size needed for the image reproduction, still holds true.
I use Photoshop - Image size 'Bicubic - Sharper' for my D810 images.
BTW do it in small steps - Several reductions instead of only one resize!
I think it does a great job,,,
-
Yes, agree Bicubic - Sharper can do great things.
However if I want to control contrast-sharpness for the final print-size I use the Image Size 'Bicubic - Smooth Gradients' in Photoshop and apply sharpness (if necessary) later at the desired size. Especially for images with a lot of bokeh (fast lenses) and lots of contrast.
-
Yes, sharpening is a big factor in the resizing process.
I was one of the people at the Meetup talking to Simone S. about resizing. I've been unhappy with my D810 resizes. I know the pro shooters have lots of experience already with resizing. For some of us however, a good experiment is in order which tests all the various resizing advice we have received over the years, don't you all think so?? ;D In particular for me - who only occasionally manages to sell a stray photo or two - this will be a very valuable experiment because I really would like to make better use of my D810. (I loved 16 MP. Got used to 24 MP. Never have quite adjusted to 36 MP.)
*******
FIRST TRIAL
For an initial trial, I went out this morning and made 3 photos from a 16 (Df), 24 (D750) and 32 (D810) MP camera. (The camera brand is irrelevant here, btw.) All 3 photos were made with the same lens (same field-of-view mostly) and the same aperture, ISO and other camera settings. Some of those camera settings are irrelevant to the raw file (depending on your choice of converter), but I will list them anyway. The field-of-view is a bit choppy between the 3 photos, so I hope it is acceptable.
Lens: Nikon 28/1.8G
Tripod: Yes
Picture Control: Standard 0 with only Sharpening set to +3.
White Balance: Auto1
Metering: Matrix
Mode: Manual
ADL: Low
AF: AF-S, focused in Live View on same spot as best I could.
Aperture: f/8
ISO: 100
EV: 0
Speed: got one @ 1/25" & two @ 1/30". I accepted the cameras suggested matrix-metered speed. The Df really should be at 1/30" though as it is a tad too bright compared to the other two. Oh well.
Subject: I photographed a green shrub in morning light for LOTS of detail. Boring, but it will perhaps serve the purpose here. I am happy to supply later some other subjects for resizing tests if suggestions are offered. Fences? Flowers? People? Landscapes? Close-ups? Cats?? <lol>
Now let me get organized to work on this...... :P ::) ???
Simone S, a question about conversion: I'm thinking that in this first trial, I'll use the conversion software supplied by the manufacturer because it preserves in-camera settings. There will be no edits. The output will initially be a TIF so that different resizer tools can be tested. I'll add the converted TIFs to the Drobox cache with the NEFs. Does this sound OK?
Simone S, a question about resize size: What size shall we agree on for posting a resized photo? 800, 1000, 1200 pixels width??
TESTS to be performed:
(we can perhaps make this more precise?)
- Well resizing of course. That's why we're here in this thread. But some details would be useful.
- Tests like sharpening before resizing or sharpening after resizing. Or both. Or none.
- Different types of resizing. [bicubic, next door neighbor, bicoastal flyover, random dice roll, etc. Yes, I jest. Do see: https://en.wikipedia.org/wiki/Image_scaling (https://en.wikipedia.org/wiki/Image_scaling)]
- Blur before resizing? How much?
- Lo-pass filter before resizing? How much? Is this already in some sharpeners?
- Upsize then downsize? Up-res then down-res?
DROPBOX LINK: https://www.dropbox.com/s/emtjail7mbru4bg/Resize_Challenge.zip?dl=0
The field-of-view is a bit choppy between the 3 photos, so I hope it is acceptable.
Note that the camera was leveled, but the fence is not particularly straight in actuality.
If these photos are not suitable for the Resize Challenge, then I'm happy to reshoot.
Andrea Resize Test One: Resized NEFs to 800 pixels width in Capture NX-D and saved as JPGs. No edits before or after.
Ok, that simple resize test drive took a little Detour to Software H3ll.
Check out this attached screenshot.
What filetype do you suppose that Capture NX-D can open??
Inquiring minds wish to know.
I'm now out of time and will return later after I've run mountains of errands, cleaned the kitchen, spoken to the long-suffering SigOth and cooled down from dealing with the World's Stupidest Converter Software.
-
Andrea Resize Test One:
Conversion of NEF and 800-pixel resize of Df, D750, D810 raw photos in Capture NX-D with no edits.
There was a bit of preserved in-camera sharpening in this baseline view.
1. Df (16MP)
2. D750 (24MP)
3. D810 (36MP)
-
I have found the best results yielded by upsizing the image using the smoother method, then downsizing incrementally with the sharper method.
-
Three files await application & display of your method. See Dropbox link above.
And what software are you referencing with the "smoother method" and the "sharper method" please??
(Not all of use use the same apps.)
-
I use a "one click" method ;) (see picture below)
The final sharpening is (has to be) done on the final image size (the downsized picture)
Downsizing with bicubic method in one step
Sharpening of the final picture with a high pass filter 1pixel (may be 0.8 or 1.2 pixels depending on final size) overlay soft light mode
OOC jpgs need to be done without (or only little) sharpening, else the downsized will look bad
All in all very simple and effective according to my experience
-
Andrea, the link says no file available (404)
-
Andrea Test Two: Here is a conversion and resize of the converted TIF using Photo Mechanic with no additional sharpening.
Do these look better, worse or the same as the Capture NX-D resized photos above?
1. Df (16MP)
2. D750 (24MP)
3. D810 (36MP)
-
Werner, thanks. I will repair that link right away.
ADDED: dropbox link repaired
Here it is again.
DROPBOX LINK: https://www.dropbox.com/s/emtjail7mbru4bg/Resize_Challenge.zip?dl=0 (https://www.dropbox.com/s/emtjail7mbru4bg/Resize_Challenge.zip?dl=0)
-
Andrea Test Three: Here is a conversion and resize of the converted TIF using Photo Mechanic's default sharpening.
I usually think that Photo Mechanic's sharpening is a bit harsh.
Do these look better or worse than the preceding?
-
So, let's see your refinement of Resize & Sharpening using your favorite methods,
now that we've seen my first three "basic" tests:
- conversion of NEF & resizing to 800 pix with no sharpening from Capture NX-D
- resizing of converted TIF to 800 pix with no sharpening using Photo Mechanic
- resizing of converted TIFs to 800 pix with "default" sharpening using Photo Mechanic
You can perform your own NEF conversion or just use the converted TIFs.
-
I have similar downsizing issues with files from my D800 and A7RII - my solution is to do light capture sharpening in ACR, then at full size make another sharpening pass with smart sharpen (very light), downsize in Photoshop (bicubic), then make another light sharpening pass viewing the image at the final size, adjusting the sharpening parameters to just make the image "crisp" w/o going overboard.
Straight down-sampling/down-sizing always leaves the image looking soft . .
-
Here are the pics with mine standards
(first RAW (not standard): ACR: reduce the high lights by 100, reduce Loca, nothing else)
-convert to sRGB
-Then: first sharpening (via high pass filter) with 0.8 pixel soft blending
-Downsize bicubic (soft edges) to 1200
-second sharpening (via high pass filter) with 1 pixel and soft blending
-safe jpg at highest level (12) (usually I reduce quality till size is below 500k, it seemed to me to cause to many jpg artefacts here, but even with 12, most leaves look like little squares)
df seems a bit brighter, 750 and 810 seem very similar
-
Andrea Test Three: Here is a conversion and resize of the converted TIF using Photo Mechanic's default sharpening.
I usually think that Photo Mechanic's sharpening is a bit harsh.
Do these look better or worse than the preceding?
Too much sharpening for this subject with this method in my opinion. Doing that with one of our dog pictures owuld make the dog appear to have pigs bristle instead of hair.
I was amazed by the results from your first test though
-
Any help in how to prepare the file for an upload on facebook and ending up with a representation of the file that appears sharp is greatly appreciated.
I think Facebook does a lot of compression, so the safest bet is probably to give them files which are already at the correct size and compressed. This approach gives you control over the compression and ensures more predictable results. But I think the interaction with Facebook algorithms makes this a somewhat more challenging problem than just resizing for your own hosting.
-
I am experiencing the same issue with downsizing. Images don't appear sharp. 99% of the time I dont have to sharpen my images at all - (when processing the high res) so for me it can't be an issue due to how I sharpened
Could you please elaborate? How do you know that you don't have to sharpen if your result is unsharp? This seems a contradiction in itself.
-
Simone S, a question about conversion: I'm thinking that in this first trial, I'll use the conversion software supplied by the manufacturer because it preserves in-camera settings. There will be no edits. The output will initially be a TIF so that different resizer tools can be tested. I'll add the converted TIFs to the Drobox cache with the NEFs. Does this sound OK?
Whatever you use, please give us the full-res image just before resizing happens, so everyone has the same starting point. Raw conversion is another can of worms, so let's leave that out of the equation. I think for massive resizes the small details will not matter anyway (e.g. how exactly the image has been de-mosaicked, these things happen at very large spacial frequencies, none of which can be represented in the final image).
Simone S, a question about resize size: What size shall we agree on for posting a resized photo? 800, 1000, 1200 pixels width??
I think any of these should be OK. The person uploading the picture can state his/her desired target size so we have a healthy variety of target sizes.
-
I have similar downsizing issues with files from my D800 and A7RII - my solution is to do light capture sharpening in ACR, then at full size make another sharpening pass with smart sharpen (very light), downsize in Photoshop (bicubic), then make another light sharpening pass viewing the image at the final size, adjusting the sharpening parameters to just make the image "crisp" w/o going overboard.
Straight down-sampling/down-sizing always leaves the image looking soft . .
Thanks! Do you find that the result changes if you leave out the capture sharpening in ACR and the full size Smart Sharpen pass?
-
Andrea, thanks for participating!
- Which one of your resizes do you find best?
- Are you seeing the differences between the D810 file vs. the two others that you were talking about in Scotland? If yes, could you try to describe them?
I find almost no difference between the resizeNoEdit and PhoMechTifResizeNoShrp version for either of the cameras, whereas the sharpened Photo Mechanic version seems oversharpened to me. I don't find any big difference between the D810 vs. the other two. But maybe I will after you tell me where to look.
-
I think Facebook does a lot of compression, so the safest bet is probably to give them files which are already at the correct size and compressed. This approach gives you control over the compression and ensures more predictable results. But I think the interaction with Facebook algorithms makes this a somewhat more challenging problem than just resizing for your own hosting.
Yes. Never a make a file too large for Facebook. 1200pix is max in my experience, otherwise you seem to lose control.
-
One observation which may be relevant to some of this discussion. Sharpening, and especially evaluating sharpness, got more difficult for Mac users when Apple put 'retina' screens on their laptops (and now I think one of their iMacs). The high-res screens seem to me to make almost all photos look more punchy - sharper and higher in contrast at small scales - than they would on screens of lower pixel density. Presumably the intention is to make cell-phone photos look as good on your Mac as they do on the bright, sharp and contrasty screen of your cell phone. However I find that I cannot reliably evaluate sharpness on such a screen. Photos that looked great on my retina laptop screen often look noticeably unsharp on conventional screens, and remain off-sharp even when downsized. I would also add that trying to get around the problem by "pixel doubling" - enlarging to 200% so that each pixel in the photo covers 4 pixels on the retina screen - does not work for me. I don't find the 200% retina screen image equivalent to viewing at 100% on a conventional screen. When I updated my laptop about a year ago this forced me to do all photo sorting, evaluation and processing on an external monitor. For some this may be affecting assessment of sharpness in the full-sized image and in the down-sizing procedure.
Andrea, your 3 initial photos (in reply #8) look fine to me. The 3 downsized by Photo Mechanic look to me unpleasantly over-sharpened - i.e. distracting, over-contrasty detail at small scales, which prevents me seeing the photo as a whole. My eyes dance around the photos trying to see what's behind the distracting detail.
Werner (Alaun), your 3 look more natural. Perhaps they could even take a little more unsharp masking (especially the middle one, 1200_810_8171), but it's a matter of taste. With your procedure nothing about them looks unnatural or distracting on my screen.
I agree with John and previous posters, that if you know the size at which your output will be displayed, downsize to those exact pixel dimensions. Otherwise your photo will be re-rasterized to those dimensions by someone else's rendering algorithm, over which you have no control.
Cheers, John
-
Any help in how to prepare the file for an upload on facebook and ending up with a representation of the file that appears sharp is greatly appreciated.
I've read PNG's have less of a chance of suffering compression artifacts than JPG's on the facebook, and uploading at the max size, 2048px. But I think you sort of take what you can get with them. Your girlfriend is not alone in experiencing random low quality images on that site.
-
Charlie, Simone, John and John, thank you for the advise given so far. Downsizing to 2048 and providing a JPG at 80% or a PNG will be the next things we will try.
Since my girlfriend uploads quite a few pictures I am not sure whether downsizing in small steps in Photoshop is a viable option because that would slow down the workflow considerably.
Does anyone have experience with Irfanview for batch downsizing?
-
Charlie, Simone, John and John, thank you for the advise given so far. Downsizing to 2048 and providing a JPG at 80% or a PNG will be the next things we will try.
Facebook's max pixel size is 1.600 or something (they change it regulary) I would suggest to make the file smaller around 1200-1400 region when using it for FB. I never mess with the percentages (because I don't know what actually is 'changed' and always keep it at max.
Goodold 'View NX2' but also 'Faststone' and DxO are good in resizing, but I prefer CC for most of the resizes because of the control you have when downsizing.
-
Fred Miranda makes an excellent tool for downsizing.
http://www.fredmiranda.com/WP_Pro_Plugin/
It's available for Mac and PC.
-
Thank you for all the comments. And also to Werner for his good examples.
My initial two tests were simply to show what a baseline photo would look like with only resizing and no additional sharpening. We need an unsharpened Resize for comparison purposes.
The third test was to show that crude "aftersharpening" is not desireable.
****
Andrea Test Four: Converter Plays No Role in My Resizing Problems
Now I'd like to begin to turn to a finished result from my typical Photo Ninja conversion and enhancement procedures so that I can show you my resizing difficulties. I'll just stick to the D810 photo on this or we will get bogged down in too many comparisons.
First of all, I want to say that the actual converter used plays no particular role in any resizing problems. So let's show that quickly.
Here is a comparison of a Nikon NX-D conversion and a Photo Ninja conversion. NO EDITS were applied in either of these conversions except for proper colour balance. When using Nx-D, any in-camera sharpening would typically be retained, but I removed it here in order to keep the two conversions as equal as possible. When using Nx-D, distortion control will be applied to any recognized Nikon lens. I removed that also for the sake of equal comparisons. You will see some very small differences in how the two converters handle highlights and colour. This is expected between any two converters - especially if one of them cannot fully read any in-camera settings.
(1) Nx-D Conversion with No Sharpening
(2) Photo Ninja Conversion with No Sharpening
-
To further show that both converters render details equally, here are some unresized crops ("100% crops").
Yeah, stuff looks terrible with no sharpening when viewed at full size.
-
Andrea Test Five: Finally!! The D810 Resizing Problems
I edited my photo in Photo Ninja and applied my typical enhancement procedures. If it is of interest, I can later describe my PN settings in detail.
- Highlights were pulled back a bit but not fully because it would have looked wrong. The strong morning sun to the left combined with some cloud scrim would naturally cause the upper left portion of the sky to be very bright. Midtones were pulled up fairly strongly to show the morning light as I saw it. Shadows were opened up a small amount for no particular reason except that I like that "look".
- The Detail slider was set to 15. The Sharpening slider was set to 75/.6.
Here are how the two detail excerpts above looked after the various corrections & enhancements. Looks better, doesn't it now?
BTW, IMHO - there is a "art" involved in viewing unresized photos at full size. They never look as good to me as the resized versions do. Viewing distance becomes very important (circle of confusion and all that) and I typically fail to compensate and look too closely!! :P ::)
-
[Note from Andrea: I have to break up the presentation because photos get clumped together at the bottom of a post which makes it difficult to write about them and keep the writing near the photo.]
Andrea Test Five - Continued
Photo Ninja offers a 1/8 resize option (among others) which resizes the original D810 7360 x 4912 px photo to a postable 922 x 616 px. Thus I am going to use a 922 pixel width for these next 3 resizes.
(1) Photo Ninja 1/8 Resize of the Edited, Sharpened Photo.
Problem: looks quite crunchy and blocky. It would seem that PNinja applies further, quite crude sharpening and perhaps contranst to the resized photo. Not good. So, to compensate for that in the next resize attempt, I removed the original sharpening.
(2) Photo Ninja 1/8 Resize of the Edited, UNsharpened Photo.
I removed both the PNinja Detail setting and the Sharpening setting and then made the same 1/8 resize in PNinja. Problem: Not as bad as the preceding, but I think that PNinja's resizing "enhancements", whatever they are, have ruined some of the brighter areas. And the photo doesn't look sharp to me. Where did all that lovely detail go?
CONCLUSION 1: Do not resize in Photo Ninja ??
Too bad, because I really do love the way it converts.
(3) Photo MECHANIC 922px Resize of the Edited, Sharpened Photo - with NO additional sharpening.
Problem: This resize might be a bit better than the two preceding, but does not seem optimal. Or particularly sharp.
(4) Photo MECHANIC 922px Resize of the Edited, Sharpened Photo - with PMechanic default sharpening.
So this one has a double dose of sharpening. The PNinja sharpening edits were not removed and the PMechanic default sharpening was used.
Problem: Not sure. My eyes have gotten tired, but this one seems somewhat better than the preceding 3 efforts. I'm sure it is not optimal.
What do you think?
CONCLUSION 2: I need to refine my Resize & Resharpening techniques & apps for the D810.
(5) Capture Nx-D Conversion, Edits and 922px Resize
Problem: Everything. I am simply unable to achieve the edits I want in this app. I think it has actually made the D810 raw conversion look worse than the in-camera JPG whose edits Nx-D supposedly preserves!! But here is my effort anyway just for the record. And so that we can compare the Nx-D resize to the other resizes.
CONCLUSION 2: I need to refine my Resize & Resharpening techniques & apps for the D810.
(6) Capture Nx-D Conversion, Edits and 922px Resize - followed by resharpening in Nx-D using USM10-5-5
I was not sure how much to push the resharpening? I hope this looks better.
-
Well, I'm still pretty much a beginner when it comes to raw converters. However, I personally like to resize in NX-D after doing my usual enhancements. The results are pretty good, I think, and I also quite like NX-D in its latest incarnation. I always remove any in-camera sharpening before resizing, convert to a jpg of the size I need, and then sharpen, often just using USM in NX-D.
If it's a shot that really matters I use other software for sharpening, usually Gimp.
Pretty simple workflow, but it's fast and straightforward.
Hermann
-
Hermann, I'm relieved to hear that Nx-D is working for someone. My effort above was a disaster!! I'm not sure why I am so bad at using Nx-D. I've tried but I just do not do well with it.
-
Facebook's max pixel size is 1.600 or something (they change it regulary) I would suggest to make the file smaller around 1200-1400 region when using it for FB. I never mess with the percentages (because I don't know what actually is 'changed' and always keep it at max.
Charlie, Simone, John and John, thank you for the advise given so far. Downsizing to 2048 and providing a JPG at 80% or a PNG will be the next things we will try.
Perhaps the facebook specific sub topic should be its own thread so we don't convolute this thread to much?
That said I just found this on facebooks website: https://www.facebook.com/help/118731871603814/
How helpful did you find this answer?
How do I upload high-resolution photos?
For better quality photos, check the High Quality box when you create an album.... then it goes on to say how to do it...
How can I make sure that my photos display in the highest possible quality?
We automatically resize and format your photos when you add them to Facebook. To help make sure your photos appear in the highest possible quality, try these tips:
Resize your photo to one of the following supported sizes:
Regular photos: 720px, 960px, 2048px (width)
Cover photos: 851px by 315px
To avoid compression when you upload your cover photo, make sure the file size is less than 100KB
Save your image as a JPEG with an sRGB color profile
You can also set it so that your photos are uploaded in HD by default.
-
Andrea Test Six: Converted (PNinja), edited (PNinja, includes sharpening) photo resized in Photoshop Elements.
No resharpening is added after the resize because this test is simply to judge the basic qualities of these PS tools.
Later I'll run a test on the PS resizer using the UN-sharpened version of the photo converted in PNinja.
We already know that a couple of the 5 possible resizing choices are not going to look good, but I'll include them for the sake of completeness.
(1) PSE Nearest Neighbor
(2) PSE Bilinear
(3) PSE Bicubic
(4) PSE Bicubic Smoother
(5) PSE Bicubic Sharper
CONCLUSION: I think it is obvious that the (5) PSE Bicubic Sharper choice is the correct one for this landscape photo. This is, of course, absolutely no surprise. 8)
-
Andrea Test Seven: Using Photoshop Elements Bicubic Sharper resizing tool, apply it to both a pre-sharpened and an un-sharpened version of the photo. Show results first with no resharpening, then with some light resharpening.
I'll use the same PNinja conversion as before. In that conversion I applied Detail = 15 and Sharpening = 75/.60.
(1) PSE Bicubic Sharper resize applied to PNinja converted, UNsharpened photo. No further re-sharpening.
(2) PSE Bicubic Sharper resize applied to PNinja converted, already sharpened photo. No further re-sharpening.
I think that (2) looks best. But that's not to say that (1) might not look good after it gets some re-sharpening. Given that I'm not sure what kind of resharpening to use and that I do not have Big Photoshop, I might not get this quite correctly done the first time.
(3) Resized #1 sharpened photo with resharpening: Adjust Sharpening 25/1.
(4) Resized #1 sharpened photo with resharpening: High Pass Overlay 0.5 px.
From this set I think that I slightly prefer the High Pass Overlay sharpening in (4). But then an adjustment of the other sharpening tool might be called for? The problem I see is that some colour artifacts have been introduced in both. Look on the fence on the left side. Perhaps I need to sharpen only on the Luminance layer? This is harder to do in PSE, but can be done.
(5) Resized #2 UNsharpened photo with resharpening: Adjust Sharpening
(6) Resized #2 UNsharpened photo with resharpening: High Pass Overlay 0.5 px.
-
Suggestion: pick ONE version from each post which looks best. The compare your choices.
I think I like this one but I'm no longer sure because I have stared at this photo all day. I'll look again tomorrow.
Thanks to everyone who is participating. I am reading all your comments and suggestions about various ways to downsize.
-
Thanks Andrea. This is a lot of data to look at. I will need more time to come back with any meaningful feedback.
-
Simone S: So far, using a D810 landscape, I have illustrated downsizing using PS(E), Photo Ninja and Photo Mechanic and included downsizing with no resharpening before and/or after.
Maybe the conclusion here is that such detailed D810 landscapes cannot be downsized in a way that preserves as much detail as we would like to see?
I need to evaluate how much my Retina screen is or is not "getting in the way" of evaluating these D810 photos. I currently have it set to "Best for Retina" display.
-
Andrea, Here's a version downsized from the (4912 x 7960) .tif file -- no adjustments. I used the WPP tool with no sharpening and then added a trace of smart sharpening (18% 0.2px). File saved at 600 x 900, Jpeg quality 7 (240 kb). The WPP tool uses a stair-step reduction process but also uses a custom (linear) gamma during the reduction process which preserves the look.
-
Thank you Bill. We'll add this to the evaluation.
-
From the various downsizes I made, I've selected a strip. Looking at the strips side-by-side might give some insight as to which downsize is "best". It will take me a minute to paste together the strips and post them. I just realized this forum software does not permit display of photos side-by-side.
It is very obvious this way which downsizes are not good.
-
I was one of the "complainers" stating that my 12MP a7S files looked better on the web than the ones from the 42MP a7RII.
As there is more detail in the "in focus" area with 42MP files I was expecting an improvement in separation from the OOF background, instead my a7RII files look less sharp overal when exported from Capture One. The general idea was that there's too much downsizing needed with the high res files they start to lose details.
Looking at my own situation this is probably caused by my unfamiliarity with Capture One and the apparent lack of additional sharpening options when exporting a 1200pix JPG, something which I used to do with CNX2.
That said I don't add any sharpening besides the default C1 sharpening settings, when working with these files there seems no need to as a little nudge on the Clarity, Structure and Contrast sliders usually does the trick while looking at the RAW files at roughly the same resolution as the exported JPG (1440 vs 1200).
-
The general idea was that there's too much downsizing needed with the high res files they start to lose details.
Yes!! But I'm not sure that resharpening after downsizing helps give the illusion of detail. Hence all these experiments.
It should be that oversampling helps produce a better smaller file. I'm just not seeing it yet.
-
Here are some two more excerpts from my experiments. This time look at the bark on the tree trunk to see if it is still detailed. The pink flowers are hopeless. Too much pink hides the details.
It is easy to see the really bad resizes. It is easy to see the soft-ish resizes. It is not so easy to pick out the best of the remaining resizes.
-
I tried to narrow it down by selecting the best 2 from each panel.
Turns out that I chose the following downsizes.
Shrub & grass composite, panel 1.
- Left: Photo Ninja conversion with no sharpening or detail slider.
Resize in PSE using Bicubic Sharper followed by PSE Adjust Sharpen 25-1. - Right: Photo Ninja conversion with no sharpening or detail slider.
Resize in PSE using Bicubic Sharper followed by High Pass Overlay 1-20.
Shrub & grass composite, panel 2.
- Left: Photo Ninja conversion with Sharp=75, Detail=15.
Resize in PSE using Bicubic Sharper followed by PSE Adjust Sharpen 25-1. - Right: Photo Ninja conversion with Sharp=75, Detail=15.
Resize in PSE using Bicubic Sharper with no further sharpening.
Trunk & flower composite, panel 1.- Top: Nx-D conversion with no sharpening.
Resize in Nx-D followed by Nx-D sharpening USM 10-5-5. - Bottom: Photo Ninja conversion with Sharp=75, Detail=15.
Resize in Photo Mechanic with no further sharpening.
Trunk & flower composite, panel 2.- Top: Photo Ninja conversion with no sharpening or detail slider.
Resize in PSE using Bicubic Sharper followed by High Pass Overlay 1-20. - Bottom: Photo Ninja conversion with Sharp=75, Detail=15.
Resize in PSE using Bicubic Sharper followed by High Pass Overlay 1-20.
[ADDED: I don't like this one anymore.]
Do we have a winner here yet??- 7 out of 8 votes for Photo Ninja conversion.
- 6 out of 8 votes for resizing in PSE using the Bicubic Sharper algorithm.
- 6 out of 8 votes for resharpening after resizing.B
- And it is an even split for sharpening during conversion or not.
- Of the 6 resharpened versions, 3 used HiPass Overlay.
- 4 out of 8 votes for the No-sharpening-during-conversion followed by resharpening after resizing.
- But also 4 votes for Sharpening-during-conversion.
As I am the only "voter", this may not be a fair election. ;D
-
I actually made progress in this quest today!!!
Test EIGHT: Pixel peep for differences in resizes as described below.
I used 4 methods of resizing. It turns out that there is little difference between certain pairs. I'll try to show you this in the next 3 posts.
Test EIGHT: PART 1
The first part investigates some Resizes which have no addition sharpening after the resize.
The subject is a Peritoma arborea shrub in full bloom found in the Mohave Desert.
Note: The photos were resized to 1200 pixels width. Please expand your browser or click up the photo to best see it.
Photo 1) Convert photo in Photo Ninja. Add initial Sharpening = 75 and Detail = 15.
Resize in Capture Nx2 with NO resharpening.
Photo 2) Convert photo in Photo Ninja with no initial Sharpening nor Detail.
Resize in Capture Nx2 with NO resharpening.
Methods 1 and 2 produce resized photos which are very nearly identical when you pixel peep. It must be that the resizing drowns out any initial sharpening that was done. When you layer photo #1 as a difference layer over photo #2, nothing shows up even when blown up to 3200%. The actual pixels (next) show how minor the differences are.
Detail 3) 3200% blowup of an area from Method 1 resize which had some initial sharpening of the raw photo. This blowup has a very very very small amount more contrast (apparently from the initial sharpening?) than does the Method 2 blowup shown next.
Detail 4) 3200% blowup of an area from Method 2 resize which had no initial resharpening.
-
TEST 8: PART 2
Part 2 investigates some Resizes to which resharpening was applied after the resize in the form of USM 15-4-4. Later we can figure out the best kind of resharpening and the necessary amount. That was not my goal today.
Note: The photos were resized to 1200 pixels width. Please expand your browser or click up the photo to best see it.
Photo 5) Convert photo in Photo Ninja. Add initial Sharpening = 75 and Detail = 15.
Resize in Capture Nx2 followed by application of USM-15-4-4.
Photo 6) Convert photo in Photo Ninja with no initial Sharpening nor Detail.
Resize in Capture Nx2 followed by application of USM-15-4-4.
Just as before, methods 3 and 4 produce resized photos which are very nearly identical when you pixel peep. Again I suppose the resizing drowns out any initial sharpening that was done. When you layer photo #3 as a difference layer over photo #4, nothing shows up even when blown up to 3200%. The actual pixels (next) show how minor the differences are.
Detail 7) 3200% blowup of an area from Method 1 resize which had some initial sharpening of the raw photo followed by USM resharpening after the resize. This blowup has a very very very small amount more contrast from the initial sharpening than does the companion area next.
Detail 8 ) 3200% blowup of an area from Method 2 resize which had no initial resharpening but did have USM resharpening after the resize.
-
TEST EIGHT: PART 3
If you've manage to sort out the preceding, I hope that you have seen that it is beginning to look like that the only real factor we need to look at is whether the resizing of an image is followed by additional sharpening - or not - because any initial sharpening or detail enhancement seems to be almost all lost during the resize.
So I'm now going to compare one photo from each of the preceding two posts.
Photo 9) PN converted photo with no inital sharpening and with no sharpening after the Nx2 resize.
Shown above as Photo #2.
Photo 10) PN converted photo with no initial sharpening but followed by USM15-4-4 after the Nx2 resize.
Shown above as Photo #6.
There are - finally - some differences between these two photos. Not as much as you would think, but certainly enough to be able to distinguish between the two resizes.
Detail 11) This is an excerpt from the difference layer between the two photos used simply to show where the differences are between the sharpened and the unsharpened photos. It is interesting there are few differences due to the sharpening, but it is noticeable when you look at the actual photos. The human eye is very sensitive to contrast. [Note: 1000 pixel width]
Composite 12) I layered the photos and cut windows in the top layer. Inside the windows is the unsharpened version. It is (to me anyway) clear which is the sharpened version.
Composite 13) Flip-flop. It is sharp inside the windows in this version. These are both a little silly but I had fun making them.
-
Some actual pixels blown up to 3200% from the preceding test.
Detail 14) No sharpening during conversion nor after resize.
Detail 15) No sharpening during conversion but USM 15-4-4 applied after resize. No surprise that these pixels are brighter and more contrasty.
-
I propose the following method to evaluate the resizing/re-sharpening objectively:
You can download this image of a Siemens star here: https://www.dropbox.com/s/hijmrelhpaesijt/Siemens.png?dl=0 (https://www.dropbox.com/s/hijmrelhpaesijt/Siemens.png?dl=0)
I prepared the file to have the same resolution as a D800 file.
The red circles show equal increments of spacial frequencies.
The innermost circle is at Nyquist, the next circle at 1/2 Nyquist, the next at 1/3 Nyquist etc.
The stuff inside the innermost circle can be ignored. It looks nice, but is just false detail. The ideal pattern that is being represented would have infinitely thin rays as you approach the center, but because the file has a finite number of pixels, the rays cannot be resolved inside the circle.
There is a little bit of false details even outside the inner circle (but inside the second), which is due to the fact that the pixel grid is vertical/horizontal, but the rays are slanted. This is an issue that is very relevant since we rarely align all fine detail horizontally or vertically.
It would therefore be wise to go a bit below Nyquist with regards to our expectations of the finest representable detail.
You can test your algorithms on this file. You can test what happens to the detail during resizing. Let's say your resize is to 920px width (a factor of 8 ). The Nyquist frequency is now at circle no. 8. Look at the detail outside circle no.8. Depending on the resizing method, this detail will be more or less contrasty. After re-sharpening, the contrast should be higher (as high as further out) but not as high to give additional false patterns.
This should give you a way to fine-tune your sharpening strength/radius independently of image content.
I haven't given this much thought, but I would guess that a sharpening radius of 0.5 will be close to ideal, since it will operate at the Nyquist frequency of the resized image.
-
Based on everything in this thread so far, my conclusions are (i) downsize to the exact pixel dimensions, if known, (ii) sharpen to the desired degree after downsizing, (iii) different people seem to be happy with different levels of sharpness in the downsized images. Not too surprising, I think, and echo comments on other NG threads about sharpening - defer sharpening until the final size and intent are known. Andrea - I also discovered that I personally have a hard time judging the appropriate level of sharpening unless I can see a complete image. I can tell that some of the test strips you produced are sharper than others, but I can't tell how much detail and local contrast is going to be too much in the final image without seeing it as a whole. For me this is really true for the "pixel peeping" crops. I really can't discern anything from these except pixel-by-pixel contrast, hence can't tell whether the 100% image will look under- or over-sharpened (do "pixel peepers" really base their arguments about lens sharpness on staring at 3200% enlargements?). Amusingly, the 9 x 7 px crops in the final post would correspond to a 100% image 104 m wide x 81 m high (assuming a 24 MP sensor). That would be impressive!
Cheers, John
-
Well, the purpose of the 3200% enlargements was certainly not to judge sharpness. It was merely to illustrate that there were only the most minor of differences between two different conversion/downsizing/sharpening methods. And that there is definitely some difference between resized images one of which was subsequently sharpened and one not.
Nothing else can be deduced from such ridiculously enlarged samples. ;D
The complete images from which the strips are taken are shown in earlier posts.
The whole exercise was undertaken to explore the commonly accepted wisdom about resizing by performing the experiments myself because I have been unhappy with my D810 downsized photos.
Some of the commonly accepted wisdom about resizing seems to be invalid.
Example: It appears not to matter whether you do or do not sharpen the raw before downsizing it. Both downsizes were very nearly identical - and will still need sharpening after the downsize.. But only tested that on landscapes so far. May not hold for other types of shots.
I'm not sure what it means to "downsize to the exact pixel dimensions, if known". Please explain, thanks.
-
>I'm not sure what it means to "downsize to the exact pixel dimensions, if known". Please explain, thanks.
Some of the earlier discussion on the thread was about re-sizing photos for posting on web sites such as facebook. In such cases you might know in advance the pixel dimensions at which your image will be displayed. If so, it seems to be good advice to downsize to those exact pixel dimensions, and sharpen to your taste. Otherwise someone else's code will re-size your image, and may apply sharpening or other rendering adjustments beyond your control.
Cheers again, John
-
Oh yes, I understand. And certainly agree! It's best to do our own resizing to meet requirements rather than be at the mercy of sometimes crude forum software resizing.
**********
Simone, thank you for the Star link. I'm going to explore that next. It will be interesting to play with the Nyquist frequency thing we had discussed in Scotland. And it is time to resize something other than a landscape. ;D
-
Some of the commonly accepted wisdom about resizing seems to be invalid.
Example: It appears not to matter whether you do or do not sharpen the raw before downsizing it. Both downsizes were very nearly identical - and will still need sharpening after the downsize.. But only tested that on landscapes so far. May not hold for other types of shots.
I don't know about how commonly accepted it is, but the sharpening before downsizing is unsurprisingly not going to matter a lot in massive resizes as discussed here.
With the exception of regular patterns such as architectural or textile textures where interference patterns could show up more if the full-size image has very high contrast at spacial frequencies that are going to end up way above Nyquist in the final image, I do not expect a lot of visible differences, and your examples confirm that.
What could be interesting to test out, particularly with textiles or such, is blurring before downsizing. An efffective blurring is implicit in most resizing algorithms, but in cases where excessive interference patterns are obtained, additional blurring to suppress high-frequency detail and make the final resize look smoother might be in order. The final re-sharpen will thus only increase contrast below Nyquist and not accentuate any false patterns that have emerged due to the fine pattern of the cloth interferring with the pixel grid of the resized image.
-
In the olden days, we were taught to hold off on resharpening until resizing is complete. But I like to clean up my raws at full size - including sharpening - so I had been saving the cleaned up raws with sharpening included. That necessitated removing sharpening before resizing - in theory. However, in practice - me being the lazy person that I am - I usually did not remove sharpening before resizing. So it was nice to learn that that for landscapey photos it doesn't matter.
-
A lot depends on how spatial frequencies are distributed in the image. Natural scenes have a spectral density very close to a power law, i.e. they have a quasi-fractal nature where if you zoom in, you see an image that has again a similar distribution of spacial frequencies.
By contrast, a facade of a building might have a few very low frequencies and one very high frequency. A vast expanse of the frame might be occupied by this very regular pattern. Applying too much sharpening at that scale before downsizing will therefore increase the likelihood of getting a weird interference pattern.
Another thought: the anti-aliasing filters in some cameras are weaker than they probably should be. When using a very highly resolving lens, the image appears sharper than from a comparable sensor with strong AA filter not because there is more information, but because the contrast at very high spatial frequencies is high. Actually you are just baking false detail into the capture. This might explain some of the crunchiness. Moderately blurring the image prior to down-sampling might improve the visual appearance.
-
Please GOTO Part 2 now because this thread has gotten rather long. Thank you.
Part Two: http://nikongear.net/revival/index.php/topic,3690.0.html