Mike Chaney's Tech Corner

A question re the 'soft proof' screen.

Ctrl/space bar.
The 'quality' of the results is very poor/low resolution. Is this normal or have I simply not noticed it before?
Jaggies and 'noise/artifacts'. The colours are ok but the poor quality is affecting what I see. Print results are just fine.
Is there something I've done or a method I can use to give me a hi res view.

(I've been on the road for 3/4 months and haven't used Qimage for sometime so my memory may be a bit 'off'!!!).

Regards

Curious! What version Qimage are you using?

Ctrl Spacebar is a Hi Resolution image.  If you select a low resolution image,  let's say, 200 x 100 you will get a screen image that size. It is not stretched or expanded.
You get a High Rez image in the image editor also. In the page Editor screen, depending on your setting, you might have to click the HQ button to see Hi Rez,

Low Rez images are for speed... so thumbs are low rez, and the upper right preview screen is low rez.

Need more information... what is the resolution of the image or images in question?  I assume they all look this way to you????

Fred

Hi Fred,

Qimage ver 2014 131.

Image is 6000x4000 tiff. So that's not an issue!!

OK here's what my work flow is. Has been forever!
NEF's out of a D7100. Converted to 16bit tiffs in Capture NX-D.
Fine tuned in PSPx6.
Brackets merged/tonemapped in Photomatix.
Photomatix output fined tuned in PSPx6. As PSPImage. Numerous layers.
Converted/flattened back to 16bit tiff. 160 meg files.
Open in Qimage for printing.

I've run some tests. Compared current assignment work with my last assignment.
All the old files have excellent results. They are either jpegs or older tiff files from a D7000 from previous assignment.
All the 'new' files have the same issue. The thumbnails are, well thumbnails with little resolution issues. The 'preview' screen in the upper right now gives a poor quality image. The edit screen also is giving a less than great resolution. The soft proof screen is very poor. Almost as if there is a huge amount of white noise.

Now the interesting part!!!! I merge 3 files to get one file. That file has numerous layers. Ultimately merged into one layered 16 bit tiff. This file gives the problem.
If I load just one of the D7100 unedited 16 bit tiffs into Qimage every thing is ok.
So the issue is with the final output of my work flow.

I have taken the 6000 x 4000 image and resized it down to 4200 x 3000 as per the D7000 output. But it makes no difference. It's in the files and not a resolution issue.

Everything works fine. My only issue is the change, since my last edit session some weeks ago, that give poor quality previews.
Printing output is fine.

I had a commission for prints. A4. 50 prints. All different. Nikon D7000. Everything was fine.
I replaced the D7000 with a D7100. I also now use Capture NX-D instead of AfterShotPro.
Other than that I have made no other changes.
I now have a repeat order of new photos for the same customer. Printed results, as I said, are fine. Just that I can't use the soft proof to judge the results.

So in a nut shell the workflow using the old D7000 files are fine. A single unedited file from the D7100 is fine. But anything I work on from the D7100 is giving me heartache. Issue seems to be with the output from the D7100. ??????????? Could it be a problem with the high resolution output of the D7100 files??

Any clues? It's not the end of the world but I liked Qimages soft proofing.

regards

Sounds like a messed-up TIFF where only the thumbnail embedded in the TIFF can be loaded.  Can you put the TIFF online somewhere and provide a link to it?

Regards,
Mike

OK it's the next day and I've had some sleep!!

I run some more tests.

Same work flow for each test.

Started with 3 nefs from a D7000. (Converted to 16 bit tiffs into Photomatix fine tuned in PSPx6 saved as 16 bit tiffs. Into Qimage).
Same but with 3 nefs from D7100. (Same as above).

Soft proof in Qimage for the files from the D7000 are fine.
Soft proof in Qimage for the files from the D7100 are rubbish.

In edit mode.
Files from the D7000 are fine. Zoom in and they remain fine.
Files from the D7100 are rubbish. But zoom in and they are fine.

?????????????????????

I repeated the above test on the D7100 files with some noise reduction. Still made no difference. Even heavy noise reduction resulted in a rubbish soft proof view.

Unedited 16 bit tiffs from the D7100 are fine. It's only when I start to work on them, which I do for every file, that the problem begins.
It's as if Qimage can't handle an edited D7100 file. I've looked at the same files in a number of other programs. PSPx6, ASP, etc and only Qimage shows any sign of problems. If I convert to jpeg they are fine.
As I said it's only the soft proofing that worries me. Everything else in Qimage is ok. Prints are ok. Now for this assignment I have to provide files. Not prints. So I'm now worried something may be amiss in the files!

I don't know where I can upload a tiff. They are 160+ meg files. Here's my web page. Not a good web site so no laughing please. http://www.cre8ivephotography.com/ The Home page is a photo from my last assignment. So it will give you an idea of my current work. All old motorbike for a museum.

Regards

WeTransfer is a free site the lets you upload large files. Pick an e-mail address that you want to send them to, select the files and upload them. You can "send" them to your own e-mail. You will get an e-mail with a link in it that you can post here if you want. See link to WeTransfer:

https://www.wetransfer.com/users/sign_in

I've all but exhausted my ideas. The quality of the 'new' files are not useable. This issue has arisen since my purchase of a D7100 and the return home with 5500 files. Old nefs out of the D7000 and an identical work flow give no problems.

If I go into Edit mode. Then click on the edit preview to bring up the 'zoom' window. At 1x all the edges have dotted lines around them!! Like marching ants that don't march. Or selections in some programs.

So old files work just fine. New unedited tiffs work fine. (With a very small amount of quality issues if I really look hard). Any new files from the D7100 that have been edited in anyway fall to pieces.

I tried those same files in other programs and don't get any problems.

So??

I'll continue to use Qimage for printing. But the soft proofing is all but useless.

Yell if you can give me any insight as to what may be going on.

Bruce

Bruce-If you send Mike the file, as he requested, I'll bet he can sort it out. And someone else told you how to send large files, if it's too big for email. Try that, and I bet it will be fixed in 2 days max, if it is a Qimage problem!

Thanks jeffjessee but sending a 160 meg files isn't as easy as you think. I'm on a 130 meg a day contract!! I'm also on a laptop traveling and making use of suspect bandwidth. So? So I'll see how I can manage but it's not that simple.

regards

Hi Mike. I sent two files few days back. I was wondering if you had time to figure out my problem.

regards

I've guessed by now my problem is not an issue as I can't seem to get a response to my queries. Sad because I believed QImage had an excellent reputation for user support. I'll just revert back to my old work flow.

Regards
Bruce

I've received nothing to date.  How did you send them?

Edit: I did a global search based on your registered QU email address (which includes all spam filtered messages for at least 6 months) and the only emails I've ever received from you were some from December 2013 where you requested (and I sent) the manual activation.  Might be a finicky mail server or hub somewhere so I would suggest sending to Fred (below).

Mike

Hi Bruce,
To avoid any address anomalies, how about sending the files to me?
I can make sure Mike gets them immediately.

My email is   wathree.ssz@verizon.net

I am home all day.

Thanks,
Fred

Thanks Fred n Mike. I sent two files to mchaney@ddisoftware.com via WeTransfer on the 28-06-2014.

Bruce

I use WETRANSFER all the time!

When you send the file, you get a confirmation email from them that your file has been uploaded.

After the recipient downloads the file, you get a confirmation email  that your file has been downloaded by......
Did you get either or both of those.?

I have sent stuff to Terry and I typoed the email address, and he never got it.

Regardless, senf them again.

Fred

I sent the files via WeTransfer and received their confirmation email on the 28th of June. No other emails from them. Unfortunately time has now run out for them to hold the files.
As I pointed out I'm on a very limited internet service. I can't resend the files again at the moment. When the opportunity comes I'll have another attempt.

Bruce

Just another note that may help.
My files are 6000x4000 tiffs. 300ppi. 160meg. (Originating from a D7100). Files from my D7000 which are 4928x3264. 300 ppi. 92 meg, have no problems. If I resize the large files down by changing the 6000x4000 to 4928x3264 and stay at 300ppi I still have an issue. However if I simply resize the files by changing the ppi to 150 then every thing is fine. If I take the 6000x4000, 300ppi tiff file and simply save it as a jpeg, no other changes, still at 300ppi, I get the same problem!!
It seems to be that the higher 300 ppi files at 6000x4000 are having a resolution problem. (I get the same problem on various computers). Change the pixel size ie height and width, and the same issue. But change the resolution and problem solved. No it's not because I need/want the 300ppi.

As I said above. Other programs handle the files just fine. It's only QImage and the softproof screen that the issue becomes an issue!!! And only on the 6000x4000, 300ppi files. (Plus the jpegs of 300 ppi).

regards

You should have received TWO confirmation files from WETRANSFER. One is confirming that they received the upload, and the second confirms that the recipient downloaded the file.



One more thought to try until you can get the file to Mike.

Since it's ONLY the softproof image that is a problem, and softproof is an image derived from a comparison between the Monitor profile image and the Printer profile image,
can you do some testing by varying the printer profile and/ or the monitor profile and see if you see any changes?
Try a common garden variety profile for both...... like the standard sRGB that comes with Qimage and most programs.

Fred


Fred

Thanks Fred A.

1/ only received the one email from WeTransfer. Confirmation that it had been uploaded. No indication that it had been downloaded.
2/ i 'played' with various monitor profiles. sorry but it made no difference.

If it was a monitor profile issue wouldn't it be a problem with other sized files? My brain tells me that there is an issue with large files using 300ppi. Note that all my files are 300ppi. It's only once I downsize by down sizing the resolution that things improve. 150ppi looks great. Until you zoom in!!!! Changing pixel width/height doesn't fix the problem. Just 300ppi files out of the D7100 that create the problem.

Prints are fine. Nothing else is an issue. I just like to use Qimage soft proof and worry that my files may have an issue. I sent a large number of these files off to a client today. I hope they don't come back to me with the same issue.

regards

I'm already 90% certain what the problem is: malformed TIFF where QU can only read the embedded thumbnail inside the TIFF.  I won't be able to do anything until I see the file, so when you are able to send it again on WeTransfer, please send me (or Fred) the link manually.  That is: send a message directly to myself or Fred and don't rely on WeTransfer to send the notification.  If we know when to look for the WeTransfer file, we can sort it out before it expires next time.  Having not received any emails from you since last December, we had no idea to even look for a WeTransfer.  And WeTransfer must have botched the notification.

Regards,
Mike

Hi Guys,

I've sent the file via WeTransfer again. However in the rush to do so I screwed up a wee bit. I inserted Mikes email instead of mine. So you'll get the notice!!! I won't get a notice.

Bruce

Got it.  Thanks!  I can see the problem.  The TIFF itself has what appears to be demosaicing artifacts.  Zoom in (a lot) on the rear wheel of the bike and look at the bright area through the spokes near the top of the rear wheel.  You can use any program you like to zoom in there as they all show it but I used QU's examiner.  You can see demosaicing artifacts (looks like a checkerboard pattern).  Those are what is causing the image to look noisy/jaggy.  When the image is scaled down to fit on the screen, those checkerboard patterns overlap in such a way that they create even larger artifacts.  Clear up those artifacts and you not only clear up the problem but have a much nicer and higher resolution image!

Now, as to what caused it... you'll have to find what is causing it in your workflow.  My suggestion is to just work directly with the raw images from your D7100 in QU because QU won't create artifacts like that and it'll create a higher quality final result anyway.  If you're "stuck on" your current workflow, then my suggestion is to do your workflow steps one at a time to see where the artifacts are creeping in.  That type of artifact usually creeps in when the original raw is decoded so I would suspect Capture NX (are you using an old version?).  That type of artifact is typically a result of the raw photo not being decoded properly.

In any case, this is an indication that you need to find what is causing the artifacts in your image.  Clearing those up has many benefits.

Regards,
Mike

It hit me that I should post a picture of what I'm talking about as many may not know what demosaicing artifacts look like.  You can see them in the attached zoomed crop.  Those checkerboard pixel patterns are simply creating aliasing as the picture is downsampled to fit into different sizes and when they align at certain zooms, you get the grainy look.

Again, these artifacts are in the TIFF itself: they are not created by QU.

Click the thumbnail below to enlarge.

Mike

Thanks Mike.
I see what you are saying, But? There's always a BUT!!!!!

But why is it only an issue in QImage. Especially the soft proof view. If I view at 100% in a number of other programs I don't see the demosaicing. (100% is a realistic limit without pixel peeping). Sure if I keep zooming in as you have done it is there. So yes it's there but only obvious in QImage.
I use Nikon's own Capture NX-D, (yes it's a beta version), one would think Nikon knows how to convert an NEF. The rest of my work flow has been the same forever.
So the question comes back to why only in QImage with only tiffs out of the D7100? And why at views far less than 100%? ie a 20% view in QImage still shows the problem. Almost as though QImage is exaggerating it.

I'll continue to use QImage as I believe in the printing capabilities of it. It's just unfortunate that the issue with soft proofing has an influence on what you see.

regards
Bruce

OK I've got a big spoon for my share of humble pie!!!!!!!

I've been doing some MORE tests!

I went back to an old version of AfterShotPro. Converted said files and started over. Guess what? Go on guess!
The issue all but disappeared. (Caveat. At 100% in PSPx6 neither conversion shows any sign of the problem. At 200% the first signs begin with the NX-D conversion. At 300%, a silly pixel peeping exercise, both files show the problem).

So somehow Nikon's Capture NX-D and the D7100, without an antialiasing filter, can't handling the files as well. Well does Nikon have a problem???
AfterShotPro doesn't give as clean/sharp a result but nor does it bother the issue.

Leaves just one question though. Why does QImage 'exaggerate/show' the problem even on a relatively small soft proof screen? ie the soft proof screen on my work computer is about 25% view. Yet at 100% in PSPx6 there is no sign of the issue let alone at 25%. Only at much high zoom rates.

regards

As I said, the problem is caused by aliasing: QU doesn't do antialiasing outside of the image editor because that is not needed unless there is a problem with the image.  QU is not the problem here.  Your images are ruined and QU is just telling you something is wrong.  If you only used those other tools that "hide" the problem, you would never have known that it is time to use something other than Capture NX.  Use QU and your raws won't be ruined.

Mike

Thanks Mike. I followed your advice and tried a few images using QU for raw conversion. Sorry to say that I still get the same results. Even the same NEF. It's something more than antialiasing/demosaicing. Even uniform surfaces and their edges come up with the 'jaggies' in QU.

For some reason QU is doing/highlighting an issue that other programs can handle.

Thanks for the help. I'll stick to my known work flow and hope that the penny falls at sometime.

regards

Please send a RAW file to us..   That will solve the whole question.

Fred

Fred A I've shot off a raw to Mike. Please keep in mind though that I'm heavily editing my files. So the issue itself may not occur with just a raw conversion. (It does if you push the conversion. I've tried that> I also realise that editing will highlight any issues). It maybe/is my work flow that's exagerating the problem. My testing has always been with the full edited version/s. The file we have been using is about the worst of the examples. I have others where you really have to look hard and only experience the 'jaggies'. The example I sent you does have a 'conflict' area with the siding/boards in the background. The horizontal grain is a problem! On the sample I've sent you take note of the horizontal lines or straight lines. They get the 'jaggies'. Look at the top of the Firestone sign. Those get very jagged and don't do so with other programs.

I'm interested in what's happening. Why is the issue only apparent in QImage and not in other programs? (It is in the files. But in QImage they appear at 25% while other programs I have to zoom into 200-300% to see the problem which is not realistic). And why is it only from files/nefs out of the D7100? Everything else works just fine. Good god I've converted, edited many files from the D7000 and D7100 and run them through various converters testing. Only the D7100 files present a problem and only in QImage.

I know I'm spending too much time on this. I can ignore it and the prints are just fine out of QImage. My files are fine in other programs. I just want to know WHY? And sleep at night!!!

Regards

How and where did you send it?  I've received nothing from you from your email address or WeTransfer.

Mike

Just a thought, I tried to send a raw file to Fred/Mike last week and two transfer sites including WeSendIt would not have it, my upload just hung at 20%.

I used send space and it uploaded no problem. Note I have a slow connection.

 https://www.sendspace.com/

Jeff

Mike I've sent a PM with the WeTransfer link.

Bruce did you ever solve your problem? If not here is a test you may want to try.

1. Display your problem image in Qimage Soft Proof, and measure the width of the image on the screen in inches.
2. Obtain the resolution of your monitor from the specifications in pixels/inch. Most monitors are around 100, but it is best to get the exact number.
3. Calculate the width of the displayed image on the screen in pixels. For example if it is 15 inches then the displayed image on the monitor would be 1500 pixels if your monitor is 100 pixels per inch.
4. Take your high resolution image that is problematic and then down sample it to 1500 pixels in width using the Bicubic method, or if you do it in Qimage use Lanczos.
5. Now take that down sampled image and display it in Qimage Soft Proof.

If that solves your quality problem, it suggests there is a down sampling method problem, potentially in Qimage itself. The testing I have done suggests that may be the case, and in certain high resolution images artifacts appear that resemble oversharpening.

I have laryngitis this morning

Set to High

Fred

"You can't be serious" as John McEnroe used to say to the umpire.  ::)
Terry

I can only say what I see when I compare the exact same high resolution image displayed in Adobe Camera RAW, Capture One Pro 7, and Qimage Ultimate. See this thread.

http://ddisoftware.com/tech/qimage-ultimate/downsampling-high-resolution-images/ (http://ddisoftware.com/tech/qimage-ultimate/downsampling-high-resolution-images/)

If you have another theory I wouldn't mind if you posted it in that thread.

That's fine. You only say what you see, but your conclusion as to where the artifacts originate is drawn from conjecture.
Set the setting for Antialias to HIGH and Qimage is now locked to disallow any artifacts. You might see a slightly softer looking image due to that settings, but you can rest assured, the artifacts you see were not created in Qimage.

See attached screen snap.

You asked for another theory?
How about a noisy sensor and other software, is masking, and Qimage is showing accurately.

No debating from me. I have done my best. Any further need for explanation will have to come from someone smarter than I.

Fred

Fred, that dialog box you have snipped, I gather is the one under Edit, Preferences, Printing Options...? If so, I would expect it would be in play for down sampling of images sent to the Printer or File? If so, that is not a solution for the issue I am seeing. It is the screen display in Qimage Ultimate of a high resolution image that is showing artifacts looking like over sharpening. The printed output is fine. The same issue that Bruce has identified in the original post. And I recall I down sampled in Qimage using Lanczos and the resulting image also displayed fine in Qimage. And further if I down sample the image in Photoshop with Bicubic it also solves the problem. The issue only occurs when Qimage Ultimate displays the image using the high resolution original. That does not leave many places where the problem could be coming from. To my thinking it can only be the down sampling method used to prepare the image for screen display in Qimage.

Again, erroneous conjecture.
That routine is used for all downsampling; to make web copies, email copies... anything!

I'm done! You can believe the sky is falling tomorrow at noon and no one can change your mind.

Stay well.
Fred

I tried the "Antialias When Downsizing" slider in the dialog box. I could see no effect on the screen display image in the Editor between the far left Off position and the far right high position. Any other suggestions?

Yep! What you see is already in the image.

Then I guess my Adobe Camera RAW and Capture One Pro 7 software is defective then. I can't see the artifacts when I view the same image with that software. Further if I down sample the image with Lanczos, Bicubic, Bicubic Smoother, and then display it in Qimage, the artifacts go away. If I down sample the image with Nearest Neighbor and display it in Qimage, the artifacts are back, and are a good match to the artifacts I see displaying the high resolution image in Qimage. I still can't find any other explanation for the artifacts other than that they are being created in Qimage by down sampling for monitor display with a method very similar to Nearest Neighbor.

We need to see examples of these "artifacts".  Screen shots or crops of screen shots would be fine.  When Bruce sent samples, I discovered that his images were defective and had a 2x2 checkerboard pattern (see previous page).  You do NOT always want to antialias when downsampling on screen because it can affect perceived sharpness and it slows down display speed.  And you generally only need to antialias when either: (a) your image is damaged and has a repeating "noise" pattern, or (b) you shoot something with a repeating pattern like bricks, etc. that repeat every other pixel or every 3rd/4th pixel, etc.

If you start out with defective images that have (for example) a 2x2 repeating defect, you don't expect software (like QU) to "fix" that problem when it downsamples because the defect becomes more noticeable at different zooms.  That's like Town A having a dam that just sprung a leak and is now flooding basements in Town B.  So Town B residents with the flooded basements send a complaint letter to their own mayor complaining that the town didn't budget for bailing buckets.  Let's see if we can fix the problem at it's source since that's what we did in Bruce's case, so let's see an example.

Regards,
Mike

I will try to screen capture some comparable examples and post them in the thread I initiated on the issue. If Bruce's problem is solved, then I do not want to hijack this thread further... Although I have never seen his images, the issue sounded similar.