Canon's Digital Photo Professional and My Image Quality Issue

Thursday October 15, 2015 at 1:14pm photography, canon, problems Comments (0) »
 CLICK TO ENLARGE: A screenshot composite of three images in Digital Photo Professional, submitted to illustrate a problem I perceieve in the program's JPEG export process.  The differences are subtle (depending on the dot pitch of your monitor), but are visible at the top of her hair, her eyes, mouth, and a few other details.
CLICK TO ENLARGE: A screenshot composite of three images in Digital Photo Professional, submitted to illustrate a problem I perceieve in the program's JPEG export process. The differences are subtle (depending on the dot pitch of your monitor), but are visible at the top of her hair, her eyes, mouth, and a few other details.

Ready The Canons

So, let me first say that I like Canon.  Quite a bit.  Especially their cameras and lenses.  I've had my EOS 40D for almost 8 years now, and I still love it.  Sure, it's a bit dated at this point, but it produces great images (at least when you don't need high ISO).  I've also greatly enjoyed using the office's 5DmkIII (it is so beautiful it makes me cry...because it's so expensive) and the 7DmkII I rented last summer was a lot of fun as well (even if it didn't quite live up to my hopes).  My 50mm 1.4 continues to impress me, and my 300mm f/4L is a thing of long-reach beauty.

Unfortunately, I have a bit of a time using the images I so enjoy making.

I first noticed this problem...maybe a year ago?  And honestly, with everything else going on, I haven't devoted large amounts of time to the post-processing aspect of my photography hobby over the last couple of years.  Not that I've ever done that much, mind you - I prefer a simple set of tweaks and enhancements, and then, done.  I'm usually loathe even to crop images - I try to get them composed right on camera.

Digital Photo Professional is a nice program.  It's simple, it's clean, and it has all of the features I would use to enhance my photos to the point that I think they're beautiful.

At least, that's what it shows me.

Comparing What-You-See To What-You-Get

The image above shows a composite screenshot, unresized but with distracting elements such as window borders stripped out.  The image on the far left shows what the RAW file looks like in the editing window of DPP with the settings I wish applied.  It looks perfect.  The center image shows a full-resolution JPEG (dynamically scaled in DPP to exactly the same viewing size as the RAW) that was exported by DPP with those settings applied.  As you can see, it is noticeably mushier.  The third frame, for good measure, shows a JPEG that has been actually resized to exactly the resolution of the scaled viewing size of the previous two frames with the same settings applied.  It is virtually indistinguishable from the second frame.

Here's the thing: I am a stickler for sharpness and clarity, and, especially for screen-view 72-dpi images, I like them over-sharp.  With lines that really pop.  When I open a RAW photo like the one above in DPP and I adjust exposure, contrast, saturation, and (most notably) sharpness, it looks fantastic in the preview window.  And I think "Wow, this image is great.  I could totally do this professionally*."  

However, when I use DPP to resize this image to about what it is on the preview screen, and I export it as a JPEG so that I can actually use it on this site, on facebook, or wherever else, the image it produces looks significantly less crisp - even with export quality set to max.  Details become mushy, and it loses the crystal clarity I was seeing just moments before in DPP.  And I think "Wow, this image really isn't much better than the smartphone pics my friends with a shred of composition sense take.  I kinda suck*."  Now, I can take a JPEG screenshot of the crisp image, I just can't get DPP to make one.  If my monitor was larger, I just might start doing that to get the images I want.  Heck, I might end up doing that anyway, and the fact that I'm considering that as a workflow strikes me as, well, completely ridiculous.

* The accuracy of these thoughts is not the issue here

Anyway, all of this makes no sense to me, and I would really like to know why it is the case.  

I am amazed at how difficult it is to communicate this.

The Ongoing Saga

At first, when I began positing this question I was hoping that I was simply doing something wrong.  Maybe I forgot to turn off the "suck" setting, or there was some step in the process I was performing incorrectly.  Now, however, I'm beginning to think the program is simply flawed and cannot do what I require of a photo editor.

I first posted this issue to the Canon Community Forum thinking "surely I can't be the only one who's noticed this" (I still think that, but evidence is lacking).  Maybe someone will tell me how to fix it, or maybe they'll just tell me to buy Lightroom.  Either way, maybe I'll learn something about what's going on.  This seems so basic...

No dice.  I got a handful of responses, none of which quite answered my question.  My last post made for "ridiculous clarity" went unanswered.

So, a few months go by and I volunteered to take some photos at our church on wednesdays for an ongoing program.  Even though the images are primarily used on a projector screen and the coordinator assured me that "whatever" would be completely fine, I like to present my best work, because it says something about me and my skills.  So naturally, this issue comes up again.  This time I contacted Canon's support via email.  Sure, I could have called them, but my foray into the support forum taught me that, simple as it seems, this issue requires some explaining.  It also requires that a person see the image quality disparity - something a telephone is ill-suited to accomplish.

So I go back and forth with Canon's email support for several messages, talking with at least 4 different people.  I explain and re-explain things - no problem, I understand how support systems work, and I don't mind bringing someone up to speed.  I am given a few suggestions that don't really solve the problem, and a couple more that deal only with tangent issues that really aren't relevant to what I'm asking. 

In the last message I sent, I outlined everything - again, for "ridiculous clarity" - just to make sure I'm being understood and the reader doesn't get side-tracked.  The response I got today told me only that, since I'm having "extended issues", "we" recommend calling the customer support line.  Yeah, reading a URL over the phone is going to be...fun.  I wonder if the poor guy just saw my wall of text and decided he'd rather pass it off than answer, but that assessment is probably overly cynical and unfair.

When I call, I might just point them to this post - that might actually be easier.  Though, I'd have to wait for them to read it.  Which I guess means the more I type here, the longer I have to wait.  If you're from Canon Support and you're reading this, I'm sorry.  I know it's not your fault.  I'd just really, really like my images to look as good as the program tells me they should, and it doesn't seem anyone else can point me in the right direction.  Help me, Obiwan Kenobe, you're my only hope, etc.

Non-Explanations

Here are the things I have been told in various correspondences.  None of them really answer the question of why a JPEG screenshot of the editor window looks better than an exported screen-resolution JPEG.

Now, my knowledge is not at all perfect on anything, and digital photography is certainly no exception.  I may be misunderstanding some aspects of how things work.  However, I am a programmer, I have studied digital photography for 8 years, and I haven't yet been told that my understanding on any of this is incorrect (and, in fact, I have tried to gently correct a couple of respondents on some things), so I feel like I'm on pretty solid ground.  

I also accept that these explanations were offered in good faith and intended to be helpful - and, though they didn't directly answer my concerns, I was most appreciative of them.  The first time I got them, anyway.  After the 3rd or 4th incarnation of the same or similar suggestion, however, I started to tire of going in circles, so I passed all of these on in the last message I sent to Canon Support in an attempt to explain what has been considered so far and how these didn't really address my problem.

So, here we go:

Non-Explanation #1: "JPEG is a lossy format, so the exported JPEG is going to look worse than the RAW preview"

Why I don't buy it: A JPEG screenshot is capable of showing the image as I want it.  JPEG as a format is not fundamentally incapable of the IQ I'm looking for.  I'm using max quality for the export.  The point of photo editor is to show you what the final version will look like.  A JPEG can look this good, and a good photo editor with the RAW file to work from should be able to produce it.  As another tangent, I have also been told I should export as a lossless TIFF and that that would look better, and it does not: TIFFs exported look virtually identical to the JPEGs to my eyes.  Even if it WAS significantly better, I don't have much use for TIFF files.  And again, a JPEG can look this good, and it should.  I've also been told, when showing RAW preview and exported images side-by-side, that a full-res JPEG being dynamically scaled down to fit the screen won't look as good as the RAW being dynamically scaled because of the format.  This I could totally believe, except that, as seen above, I see basically no difference between an exported JPEG being dynamically scaled, and a JPEG exported at the window-size resolution being viewed at 100%;

Non-Explanation #2: "The low-res JPEG is lower resolution, so it has less information, so of course the RAW file which has more information from which to sample is going to look better"

Why I don't buy it: The images in both cases are being viewed at screen-resolution - 72dpi, and exactly the same number of pixels.  Resolution alone does not account for the discrepancy.  DPP has *exactly* the same information available to it when it creates the JPEG as it does when it creates the on-screen preview, ergo, both should look the same at a given resolution.

Non-Explanation #3: "You're not viewing the files at 100%.  When you export the JPEG at 100%, it looks the same as the RAW at 100%"

Why I don't buy it: This might actually be true - I'm not terribly good at comparing 100% images.  The on-screen dot pitch is too large for me to notice the kinds of details I'm examining.  That said, this is still completely beside the point: While it would be good news for the rare occasions when I print a photo, in general, I don't want a 100% image that's thousands of px on a side, I want a screen-resolution image.  DPP can show me a screen-res image in the preview that looks perfect.  I can take a JPEG screenshot of it.  Why can't it export a JPEG that looks like this?  The case that 100% JPEGs are perfect would be completely irrelevant to my observation that scaled JPEGs are not because it is possible to have a low-res image that looks perfect.  That, and if I export the image at full size and then use Photoshop or some other program to scale it down, it doesn't look any better than when DPP scaled it, so I'm not sure the claim is true to begin with.  

Non-Explanation #4: "When you edit a file, it changes the file.  (Lots of successive) edits result in a loss of image quality"

Why I don't buy it: This seems irrelevant to my case, but ok.  My modus operandi is to open images in DPP, apply a standard set of edits, export, done.  I don't do a lot of post work, that's not what I love about photography.  I generally only edit photos once - MAYBE twice if I notice something later.  I have not edited the files enough to cause image degradation - in fact, I know for a fact the image of the boy in the chair [referenced in a previous message] was edited exactly once.  Though I know it to be true of repeated edits to JPEGs (which is something I never do anyway), I'm not actually sure the statement is accurate with respect to RAW files in the first place.  DPP seems to save the *settings* you're applying to the RAW file separate from the RAW file itself, though perhaps I am mistaken.  Also, the *kind* of quality difference I'm seeing isn't consistent with what I have seen in JPEG/image degradation, though my experience there is less than vast, and I would be hard-pressed to articulate the difference.

Non-Explanation #5: "There must be something wrong with your computer/software or your camera."

Why I don't buy it: I thought this for a while, too.  I guess it sill could be the case.  However, evidence points to the contrary.  I have observed this problem in two different versions of DPP (3.11.x, 3.13.x) on my macbook using files from my 40D as well as the office's 5DmkIII, a different version of DPP (3.14.x) on a different PC using files from both of those cameras, and DPP 4.x on still another PC using files from a 7DmkII.  I believe the problem is in the program(s) - and apparently a core aspect of it that does not change from version to version - and not in any hardware or software I'm using.

My Suspicions

I have spent some time speculating about the actual cause of this issue, since it doesn't seem there is an easy solution.  I'm wondering if that fact is why I've been given a bit of a run-around.  I passed these on to Canon Support as well.  Well, except the first one.

Possible Explanation #1: I'm insane.

Ok, this really isn't an explanation, but I'm starting to feel this way.  Except when I look at my images, then I remember that there is a problem and the puzzling fact that no one else seems to acknowledge it starts to seem less significant.

Possible Explanation #2: DPP fails to apply certain settings.

If, when viewing the RAW image, I remove the applied sharpness (and possibly contrast) tweaks, the image looks basically identical to the JPEG resulting from an export made with those settings theoretically applied.  I have also exported images with no sharpness tweaks and found that they look virtually identical to the ones that supposedly had those settings applied.  I have not done an intense side-by-side comparison on this, however, and comparing two "meh" photos is more difficult than comparing one "meh" and one "good".

Possible Explanation #3: DPP applies certain settings before scaling, and, due to how DPP scales the images, their impact is being dramatically lessened because of this - whereas, in the editing window, perhaps it is dynamically applying those settings to an image that is already scaled to the screen and so they are more visible on the displayed image.

This is just speculation, but as I have been told the 100% full-res images RAW & JPEG look the same and I have to assume this might be true, this eventuality might explain why the resized images that I want to actually use might not look as good.

Possible Explanation #4: DPP's conversion method(s) and/or resize/scaling algorithm isn't all that great.

There is an obvious lack of quality with no obvious cause.  On its face, the program simply seems to do a poor job producing images for use.  I really hope this is not the case, but no one has offered proof to the contrary.

 

So, there you have it.  Pretty much the beginning and end of my problem that no one can answer.  We'll see what (if anything) happens when I find time to give them a call.  Maybe I'll find out I'm wrong about everything.  I don't really care as long as I can get my photos to pop.

 

~PS

Submit a comment...

NO HTML ALLOWED [because: spam]