Digital Photo Professional - Export/Resize Sharpness

[penguinsushi]

Hi, everybody -

I've been troubleshooting this for the last few days, and it's driving me crazy.  Any help would be much appreciated.

I have noticed that when I export and resize images in DPP, they do not look as good as when I view the original raw in DPP.  I have heard it said that "well, you're not comparing the images at 100% - if you look at the raw image at 100% and an un-resized exported jpg at 100%, they'll look the same."  

Well, I have two problems with this.  

1) I find it hard-to-impossible to work with images at 100%.  I'm not that cool.

and

2) Regardless, I would expect the exported image to look like it does in DDP when I'm editing it.  To me, that's kind of the point of the editor - to see what it's going to look like when you export it.  I find it a bit ridiculous that I can take a screenshot of the raw image and the resulting jpeg looks better than anything the export produces (if anything, I would have expected it the other way around).

After doing a little experimenting, it appears that the sharpness and/or USM settings are not being applied on export.  I've exported multiple images with various settings and they all look about the same.  (I can't say they're exact, but I can't tell a difference)

I'm using DPP 3.13.51 (mac) - which is the latest version my machine will run.

Below is a photoshop-composited image comparison.  I would have expected these two images to look basically identical, but the one of the left is clearly more crisp than the one on the right - seen best in her hair, eyes, mouth and the detail on the ball. (yes, i know the images aren't exactly the same size - but I don't think that should be a significant factor here - and it always looks the same no matter how I scale the window in DPP).  This bothers me a great deal.

Has anyone else observed this problem?  Am I looking at it wrong?  Is this some issue with this version of the program?  My machine?  Something else?

Thanks,

~PS

[penguinsushi]

Re: 

"Looking back at your sample images, I see that the RAW image is "fit to window", while the JPEG is "100% view". I don't think those are comparable. The RAW image has far more than enough information to display the image at the implied resolution, while the JPEG has no extra information. The RAW display is apt to be better because the program has more to work with."

No, not really.  I see what you're saying, however, the program has exactly the same amount to work at the time of conversion.  This is what puzzles me.  The program can dynamically convert the full resolution file for display on my screen at that resolution while I'm working with it - and yet, when it does that same(?) conversion for export, it does not look as good.  Again, it doesn't seem right that I can take a screenshot of myself working with the RAW and it turns out better than a process-exported image at the same resolution.

Re:

"All that said, I'm having difficulty seeing the problem at all. The images look very similar to me. Their most prominent feature is the blown highlights, and those look about as bad in the RAW image as in the JPEG. And I don't really see that one image is sharper than the other. Maybe others don't have my problem, but I think I'd at least find the comparison more meaningful with properly exposed images."

Haha.  My wife said basically the same thing.  Yes, I'll grant you the differences are subtle, but they're definitely there.  I can be a bit overly obsessive about this, however.  It's possible (likely, in this specific case) that I'm the only one who cares.

As for the overblown highlights, sure.  And yeah, the EV is the same in both from what I can tell - no argument there.  If I try to recover the highlights, I either end up with her face too much in shadow or with unacceptably-low image contrast.  Again, I'm no pro.  I generally try to set the exposure to get the "important" part of the image, and if other parts are too dark or overblown, well, that happens.  It'd certainly be more ideal if all elements were properly exposed, but that's not always possible with bright sunlight and a running child - and that sort of thing doesn't bother me.  

I could find another, more evenly-exposed image later to re-illustrate the issue if that would be helpful - but the differences are likely to be on the same scale.  I actually found this image to be fairly "easy" to see by comparison.  Other images I convert just "seem" off, but I have a harder time identifying the problem.

I'm generally not the sort of person that finds phantoms where there are none.  I've actually written this problem off for several months - noticing differences, but just chalking it up to the difference in size, or Facebook's image compression, my imagination, or some other circumstance.  It was actually this image that caused me to experiment and notice what I identify as a problem.

~PS

[penguinsushi]

Alright, I have done some more work in an effort to make this inescapable.  I have a new image (without overblown highlights),  It's a tighter shot, making the facial sharpness more apparent (to me, anyway).  The difference here seems still more obvious.

Above is a screenshot (cropped and composited) of three images open in DPP.  On the left, is the RAW image I'm working with.  In the middle is the exported JPEG (full quality, displayed scaled identically to the raw).  On the right is another exported JPEG, this one exported and resized and being viewed at 100% but exactly the same size as the others that are "fit-to-window".

i see no noticeable difference between the JPEGs, but the RAW image is clearly sharper.  If DPP can display the RAW image for my screen looking that sharp, why is the processed/exported image noticeably mushier?

Also, I have run now this same experiment on a PC using versions 3.14.15 and 3.14.48.  Same results, exactly.

~PS

[RobertTheFat]

---

@penguinsushi wrote:

Alright, I have done some more work in an effort to make this inescapable.  I have a new image (without overblown highlights),  It's a tighter shot, making the facial sharpness more apparent (to me, anyway).  The difference here seems still more obvious.

Above is a screenshot (cropped and composited) of three images open in DPP.  On the left, is the RAW image I'm working with.  In the middle is the exported JPEG (full quality, displayed scaled identically to the raw).  On the right is another exported JPEG, this one exported and resized and being viewed at 100% but exactly the same size as the others that are "fit-to-window".

i see no noticeable difference between the JPEGs, but the RAW image is clearly sharper.  If DPP can display the RAW image for my screen looking that sharp, why is the processed/exported image noticeably mushier?

Also, I have run now this same experiment on a PC using versions 3.14.15 and 3.14.48.  Same results, exactly.

~PS

---

All other things being equal, the RAW image should always look a tiny bit better. Information is bound to be lost in the conversion to JPEG, but DPP has all the original information available to it when displaying the RAW image. One thing you could do is try somebody else's RAW-to-JPEG conversion (Lightroom's, for example). If it works better, you have the option of using it instead. Of course it's possible to wonder how many independent RAW converters are actually out there; if the others all use Canon's RAW Codec, you may not see any difference.

Another thing you could try is converting to TIFF and then to JPEG. TIFF is supposedly a lossless encoding, so that step shuoldn't cost you much resolution. And there may be more TIFF-to-JPEG than CR2-to-JPEG converters out there. TIFF files tend to be inconveniently large; but if you don't have to send them off-site, you may be able to live with that.

[penguinsushi]

"All other things being equal, the RAW image should always look a tiny bit better. Information is bound to be lost in the conversion to JPEG, but DPP has all the original information available to it when displaying the RAW image. One thing you could do is try somebody else's RAW-to-JPEG conversion (Lightroom's, for example). If it works better, you have the option of using it instead. Of course it's possible to wonder how many independent RAW converters are actually out there; if the others all use Canon's RAW Codec, you may not see any difference.

Another thing you could try is converting to TIFF and then to JPEG. TIFF is supposedly a lossless encoding, so that step shuoldn't cost you much resolution. And there may be more TIFF-to-JPEG than CR2-to-JPEG converters out there. TIFF files tend to be inconveniently large; but if you don't have to send them off-site, you may be able to live with that."

Should it?  According to my understanding, the RAW file really isn't "what you're looking at" - you're looking at the program's interpretation of the data supplied by the raw file, with the config you've applied to it through the program.  That is, it's dynamically generating a preview of what the photo would look like.  It's basically a proof.  If anything, I would expect an actual processed export to look a tiny bit better.

Am I misunderstanding something?  This is what I keep coming back to - what confuses me.  The program can take my RAW+config and show something on the screen.  Something that I can take a JPEG screenshot of and end up with a JPEG that looks significantly better than the JPEG produced by the output of the actual processor, pixel for pixel.  If I had a monitor big enough to view the entire photo, I'd just about say this is the preferable solution to my problem.  And that fact strikes me as completely ridiculous.

I have tried to export as TIFF, and the file was not noticeably better than the JPEG at full quality (at least not at the dimensions I'm wanting).  I may play around with that a little bit more, at your suggestion.

I have considered getting an Adobe CC Photography  subscription for Lightroom and Photoshop after consulting a pro friend of mine who told me that's what he uses.  Unfortunately, I would have to buy a new computer (or start commandeering my wife's) to use it, so this isn't the preferable solution - and it also feels like overkill, considering that I don't do much to my shots in post.  Again, if I'm going to put $$ into photography, I have lenses I want and I'd kinda like an upgrade from my good ol' 40D.

Don't get me wrong, RobertTheFat, I do appreciate your help - and thanks very much!  It just doesn't quite add up to me yet...

~PS

[Bertinsky]

Hi, penguinsushi

I just happened to come across your post since I was trying to figure out why images viewed at 100% in DPP appear nowhere near as sharp as they look when viewed at 50%. Like your sharpness problem, this somewhat bugs me and I've yet to discover a logical explanation (I'm sure it's a combination of  technical and viewing distance factors).

However, I was intrigued by your problem and set about replicating it (and, yes, there is a noticeable difference in sharpness between your two images). Lo and behold, I noticed a difference in sharpness between my RAW and exported TIFF or JPEG images!

But I think I've solved your 'problem': if you set your Sharpness settings to ZERO under the RAW and RGB tabs in the Tool Palette  -- by sliding the Sharpness sliders to (0) and 0 or, under the RAW tab, selecting a Picture Style of Neutral or Faithful -- a RAW image will appear as unsharp (or fuzzy) as its exported version.

Why exported images do not retain the Sharpness (or Unsharp mask) settings, I do not know. Of course, as you probably have already been doing, exported images will still have to be sharpened ...

Note: I'm (still) using DPP version 3.13.45.0.

Let me know if that helps.