Digital Photo Professional vs. Adobe image quality comparison

[Damon Lynch]

Attached are two 100% crops from near the bottom left corner, with no input or output sharpening. The lens is an EF 16-35mm f/2.8, body is 5D Mark II. The first crop is from Lightroom 5.5 with the lens profile and color correction activated, the second from DPP 3.14.5 with its Digital Lens Optimizer activated (output via LR export, with no image adjustments except for the crop). See for yourself which is clearly superior!

The third image is the final image itself, processed via DPP -> LR -> PS -> LR. As you can see the dynamic range is not too big. If it were larger then DPP would struggle because its highlight recovery is poor. Lightroom eats it for lunch in this respect.

[Eyeball]

Comparing raw developers seems like a really difficult task to me since there is no objective baseline with which to compare.
Using default or "zeroed" settings doesn't really give you an objective baseline - it just gives you a comparison of the developers at their default or "zeroed" settings.  I guess that might be of interest to people who don't want to spend time making adjustments but adjustments are easy enough to do and can be stored as presets once tuned in.

In this case, it looks to me like the DPP version has more sharpening applied, regardless of what DPP says in its sharpening settings.  It also appears to be a little noisier.
LR seems to have a little more CA and a warmer color balance but it seems to me that those things could be tweaked pretty easily, as could the sharpening.

It would be interesting to do the exercise, making whatever adjustments were necessary in LR, to see how close you could get the LR version to the DPP one.

[francois]

FWIW, the LR image could be adjusted in LR to be similar or very close to the DPP one (in sharpness).

[TonyW]

FWIW I prefer the tonality of the LR rendered version sharpening effects may be a red herring in this case

Not sure how anyone could make a judgement on superiority from one editor to another at least until you get the best you can from each and perhaps try your best to match each one with the other application.

I have very little experience with Canon products including DPP but have always been under the impression that DPP even with zero sharpening showing in the application has actually applied default sharpening under the hood so to speak.  Therefore you cannot turn it off. 

If that is the case then perhaps a better comparison would be seen by applying the best capture sharpening you can in both products then comparing output specifically for screen or print?

[Damon Lynch]

I agree that comparing RAW convertors is fraught with peril. Sticking to default processing values is typically not particularly enlightening.

However with this image, I contend that even the finest post-processing guru could not come close to the the DPP version if they used only Adobe's imaging pipeline. I am certain of this because the results of using DPP's Digital Lens Optimizer are dramatic with this image.  No capture, creative or output sharpening will compensate for the fact that the Adobe rendered version is mushy compared to DPP's DLO version. Prior to applying the DLO, the DPP version is similarly bad.

[Damon Lynch]

Damon what is your normal workflow? Do you run your photos through DPP first?

I run an image through DPP first only if DPP's DLO will result in an improved image and the dynamic range is not too great. I realize that's a particularly boring answer, but it's the reality. DPP has one massive strength -- getting the best optical quality out of selected Canon lenses -- that I suspect no other convertor will ever come close to unless it incorporates Canon's proprietary algorithms. In pretty much all other respects I find Adobe's products superior.

[TonyW]

I am sure you are correct about applying DLO.  But suspect that working working in Adobe imaging pipeline only will/can yield results just as dramatic.

I am certainly no post processing guru but just taking the presented LR image into PS and applying High Pass filter at 1 pixel radius overlay blend results in slightly increased sharpness to the DLO version reducing the opacity or slightly lower radius should get closer still if that is whats required.  

Therefore I honestly cannot see any reason that working with the raw data in LR or ACR the processing could not be matched.

[Damon Lynch]

Tony if you have access to DPP then you're welcome to do your best with the CR2 of this image and make the comparison yourself, as long as you promise not to share it   PM me if you're interested.

[Tim Lookingbill]

So what is it do you like about the DPP version which I find looks ashen though much sharper?

I'm surmising this toned down DPP version evenly distributes saturation across all colored elements that would saturate evenly adjusting the sat slider with no hot spots kicking up (color noise) while not increasing saturation in objects that appear gray such as the tree trunks and branches.

I actually have to fight that ashen look when applying CA, color noise and sharpening in both LR/ACR. I take it you like this look.

Another strange effect that happens in LR/ACR and appears in your DPP crop is when I zoom close to high frequency detail like the branches located around the bottom/side of the frame of my own Pentax PEF Raws the CA adjust actually makes that zoomed in area of the image appear more sharper.

[TonyW]

Tony if you have access to DPP then you're welcome to do your best with the CR2 of this image and make the comparison yourself, as long as you promise not to share it   PM me if you're interested.

Hi Damon,
First thanks for your kind offer and the indications that you would be prepared to trust and share your original raw file with me to see what I would do with it.  I am certainly happy to give this a try but wonder if it would be of any real benefit for several reasons.

I am not afraid to give the raw image a go in LR to see what I come up with but I recognise a few names on this forum that I would consider real pp gurus with knowledge and experience far in excess of my own.
  
What you have certainly demonstrated here is a sharper image via DPP and a particular filter of that I am in no doubt.  What I question is the methodology leading to the conclusion that one is actually much better than the other without fully applying the tools available in both applications to achieve specific results.

If the final image destination is print then until you actually make a print it is impossible to judge final quality due to your monitor not having enough resolution at actual print size.  As it stands now viewing the crops on a monitor with a resolution of 100 ppi at 100% zoom you are looking at an image enlarged 3x that of the print assuming print resolution will be 300 ppi.  To get actual print size on screen you would need to reduce the screen view to 33% and this tells you nothing about how the print will appear as far as sharpness goes – at least that is the case for me.

If the final image destination is monitor only and you size the full original shot appropriately and sharpened for screen output then I wonder if the difference in sharpness could be percieved by most or at least those using monitors with resolution not exceeding 100 ppi?

As I mentioned earlier there is a shift in colour also mentioned in more detail by Tim that I am not too keen on, but then this is not my image and may be the look you are after or maybe something that you would deal with seperately.

Anyway I did shoot my mouth off and stated I felt that sharpening in LR correctly using the tools available should level the playing field when working with the raw data.  As the only data available at this time are the crops I just had a little play with your images and brought the JPEG into ACR and used just the sharpening and noise reduction tools to see what they may bring.  Obviously not ideal as it would be better to work on the original raw data, but maybe it shows something.  And to me at least confirmed my view that much better to be had working with the tools to hand with raw data.

[Damon Lynch]

I'm surmising this toned down DPP version evenly distributes saturation across all colored elements that would saturate evenly adjusting the sat slider with no hot spots kicking up (color noise) while not increasing saturation in objects that appear gray such as the tree trunks and branches.

I actually have to fight that ashen look when applying CA, color noise and sharpening in both LR/ACR. I take it you like this look.

Hi Tim, my objective with DPP's output is to get something good to work with later, so I have maximum headroom when using plugins like Topaz Clarity and Nik Pro Contrast. For instance in my experience some of Topaz Clarity's default presets can quickly produce over-the-top results that I need to tone down. In such situations I find giving these plugins a flat-looking file is helpful. So when I use an LR-PS-LR only workflow (which is most of the time), I also almost always start out with a relatively flat image.

If in answering your question I missed your point (which is entirely possible), please feel free to clarify.

[Damon Lynch]

I am not afraid to give the raw image a go in LR to see what I come up with but I recognise a few names on this forum that I would consider real pp gurus with knowledge and experience far in excess of my own.

Hi Tony, thanks for taking up the offer to try to push the RAW to see what you can come up with in LR / ACR. I look forward to seeing your results. I would be happy if you can produce results as good as what we see from DPP. In my opinion it's good you're not one of the gurus because obviously very few of us are. What practical use would a potential solution be if one needed to be a guru to be able to use it in Photoshop? DPP provides excellent output with literally just a few clicks. Of course if a guru could provide an easy-to-use solution, then that would be great.

[Eyeball]

By just playing around a little bit with the provided Jpegs, it looks to me like you could get close (maybe better) to the DPP version in terms of sharpness by using LR sharpening at 100 Detail, Amount in the 10-20 range, and Radius of about .7 or .8.

With that high amount of Detail, the deconvolution engine of LR sharpening should be fully engaged.  It will also accentuate fine noise a bit, too, but probably not any worse than is shown in the DPP jpeg sample.

Obviously, things could be tweaked a lot more from there to further reduce noise, adjust the white balance and saturation, and maybe try some different camera profiles to get color closer to DPP.

Now in theory, DLO could be applying sharpening in a variable manner across the frame based on Canon's mapping of the performance of the particular lens but I would like to see evidence of that before actually believing it.  Up until DLO was included, Canon always took a pretty crude approach to sharpening.  If Canon started to use something like deconvolution AND applied it variably across the frame based on lens data with the inclusion of DLO, it was certainly a giant leap for DPP. 

[Glenn NK]

By just playing around a little bit with the provided Jpegs, it looks to me like you could get close (maybe better) to the DPP version in terms of sharpness by using LR sharpening at 100 Detail, Amount in the 10-20 range, and Radius of about .7 or .8.

With that high amount of Detail, the deconvolution engine of LR sharpening should be fully engaged.  It will also accentuate fine noise a bit, too, but probably not any worse than is shown in the DPP jpeg sample.

Obviously, things could be tweaked a lot more from there to further reduce noise, adjust the white balance and saturation, and maybe try some different camera profiles to get color closer to DPP.

Now in theory, DLO could be applying sharpening in a variable manner across the frame based on Canon's mapping of the performance of the particular lens but I would like to see evidence of that before actually believing it.  Up until DLO was included, Canon always took a pretty crude approach to sharpening.  If Canon started to use something like deconvolution AND applied it variably across the frame based on lens data with the inclusion of DLO, it was certainly a giant leap for DPP. 

 Isn't DPP designed/aimed primarily for CR2 files?

[Tim Lookingbill]

What practical use would a potential solution be if one needed to be a guru to be able to use it in Photoshop? DPP provides excellent output with literally just a few clicks. Of course if a guru could provide an easy-to-use solution, then that would be great.

I'm confused by your goals here. First you want an easy-to-use solution (everyone is looking for that even gurus) and you find it doing minimal work using a few clicks in DPP and do most of the tone mapping/Clarity/Contrast in a pixel editor. But then you say this...

For instance in my experience some of Topaz Clarity's default presets can quickly produce over-the-top results that I need to tone down. In such situations I find giving these plugins a flat-looking file is helpful. So when I use an LR-PS-LR only workflow (which is most of the time), I also almost always start out with a relatively flat image.

So how do you decide on which image is made flat looking in DPP and which for LR? And what happens to this flat file made in LR (or DPP) when applying clarity/contrast in the Raw converter?

Can you show us comparison images doing it all in DPP or LR vs using the Topaz pixel editors?

[Eyeball]

Isn't DPP designed/aimed primarily for CR2 files?

Yes, but I'm not sure why you are asking.  I mentioned playing around with Jpegs because I didn't want to bother the OP for the raw file if that is what caused some confusion.  The adjustments I suggested in LR would ideally be made to the raw file.

[digitaldog]

Comparing raw developers seems like a really difficult task to me since there is no objective baseline with which to compare.

Exactly! And the skill of the user moving all those sliders.

[digitaldog]

Hi Tim, my objective with DPP's output is to get something good to work with later, so I have maximum headroom when using plugins like Topaz Clarity and Nik Pro Contrast. For instance in my experience some of Topaz Clarity's default presets can quickly produce over-the-top results that I need to tone down.

The problems are being caused by Topaz then. Setting everything to zero is a bit like trying to take a good photograph with one hand tied behind your back and one eye closed. Why? You've got all the data and rendering potential with that raw data, why cripple the first and most important part of the processing by zero'ing the sliders? Would you do that in the darkroom with a neg? My suggestion is KISS, get the best quality data and image rendering (color and tone) using the raw converter as you can, then worry about the need (or probably lack) of some 3rd party plug-in that's only going to take you longer to process and throw data away.

[Damon Lynch]

Hi Tim and Andrew:

In practice I find the results that I get from using some combination of Topaz tools and Nik Pro Contrast gives me better results than I can get sticking to LR / ACR and its tools. I agree it's a shame to be working on TIFFs instead of the RAW data but in the end if the results are better I'll take the pragmatic approach. One thing I didn't mention is that of course I always adjust shadows, highlights and overall exposure before leaving LR or DPP.

Tim please give me a little time to produce comparison images of the kind that will add to our discussion, which I am enjoying.

I have examined some a couple of versions of LR/ACR edits kindly produced by Tony on the CR2 file. I have also shared with him the TIFF produced by DPP. I'll wait for him to comment before sharing my own thoughts.

Hi Eyeball:

I don't know what Canon's algorithm is. Whatever it does my guess is that it is not uniformly applied across the image, because after applying the DLO the size of the CR2 doubles in file size. If there was a uniform transformation, why would that be needed?

[TonyW]

...I have examined some a couple of versions of LR/ACR edits kindly produced by Tony on the CR2 file. I have also shared with him the TIFF produced by DPP. I'll wait for him to comment before sharing my own thoughts.
...

Damon
Thanks for the TIFF I have replied and put a comparison sections of TIFF on top of yours.  You are correct in seeing sharpening artefacts on mine - I went a little OTT forgetting that really we were looking at capture sharpening.  I can be a little heavy handed with sharpening and often seem to work on the principle that a little is good therefore more must be better  
Still think the proof of the pudding is how the versions either print or respond to downsizing for display