Can you see any difference? DPP VS LR Raw conversion

[gail bisson]

I processed this image absolutely identically in DPP and LR4.
Image is full frame, nothing altered.
I increased the exposure in both programs by 1/3 of a stop. Followed Arash's guide for NR and sharpness as per his guide for DPP
and transferred images into Photoshop where I did
1. NR on BG only at 10
2. levels adjustment to +2 on left of graph and 241 on right of graph
3. increased saturation to +4
4. Detail extractor at 4%
5. JPEG then sharpened ( bird only) at 35% at 0.3 pixel.

[gail bisson]

Picture #2
Interesting that image #1 is 323.1 in size and image #2 is 344 KB in size. I wonder why?
So do any of you see difference?
I am very interested in this because I will eventually be converting all my images in DPP if Adobe tries forces me to join Creative Cloud for Photoshop and perhaps someday LR.
Gail


1DX
F 600 II
ISO 800
SS 1/1250
F 8.0

[Don Lacy]

Gail, Edit two is brighter if you take a color reading from the URC of your first edit you get the following readings for the color channels red 196 green 209 and blue 226 while with your second edit they are 206, 218, 234. You did not mention if you changed the color profile in ACR the default profile is Adobe standard you might want to try it again and change the profile in ACR to match the one you chose in DPP if you used standard in DPP then change ACR to camera standard to match. I would say the difference in the images are negligible with both converters doing and excellent job.

[arash_hazeghi]

Gail,

you will not see a difference at this size. Even output from different cameras will not look very different when you make small JPEGs.

If you want to compare you need to post a 100% crop.

[Diane Miller]

Try layering the image from one method over the other in PS (the full sized images), and turn the top layer on and off for a good comparison. (Zoom in and look around.) It will show any differences much more than a side-by-side comparison.

And do it on a number of images, especially difficult ones.

I've had some dark areas go flat on me with DPP, trying to lighten them, which I didn't get in LR. DPP has less leeway than LR/ACR to lighten dark areas, although with LR/ACR you do pay a price in noise when you bring up darks too much.

[John Guastella]

I would say the difference in the images are negligible with both converters doing an excellent job.

+1 on Don's comments.

Over the last few years I have gone back and forth many times between DPP and ACR, and have never seen any significant difference between the conversions. And I am an obsessive pixel-peeper.

Morever, ACR's features and its UI are so overwhelmingly better than those in DPP (which looks and functions like it was programmed by a bunch of second-rate coders) that, even if there was a difference in the conversions, I would probably still choose to work in ACR.

John

P.S. Lovely image of a godwit!

[Diane Miller]

One thing I forgot to mention that I have noticed is that sharpening with DPP can very quickly create tiny white squares. (Yes, I'm an obsessive pixel-peeper, too.) They're going to rear an ugly head when you enlarge from a crop, or to print.

Yes, the noise with DPP is slightly better at high ISOs -- at or above 1600 for me. But after I go to a good NR program like Nik Dfine or the like, the difference is maybe 1%. For that, although I appreciate that there may be a theoretical advantage with DPP, I too like the ACR / LR conversion engine. And it's integrated with my LR workflow.

[Jerry van Dijk]

Hi Gail, I've ignored the other posts to give you an honest view, so maybe I'm doubling things: Image 2 is a bit brighter and seems to have less contrast than the first image. Also the yellow/orange colors seem a bit more dominant (but that's maybe just due to brightness). I'd be extremely happy with any of these images (well done!) and I would judge from these images that you'd be able to get the image exactly how you want it to be in very good IQ in both workflows.

[arash_hazeghi]

One thing I forgot to mention that I have noticed is that sharpening with DPP can very quickly create tiny white squares. (Yes, I'm an obsessive pixel-peeper, too.) They're going to rear an ugly head when you enlarge from a crop, or to print.

Yes, the noise with DPP is slightly better at high ISOs -- at or above 1600 for me. But after I go to a good NR program like Nik Dfine or the like, the difference is maybe 1%. For that, although I appreciate that there may be a theoretical advantage with DPP, I too like the ACR / LR conversion engine. And it's integrated with my LR workflow.

I don't know what you mean by "white squares", maybe you are talking about halo artifacts? Can you post an example ?

You have to use the sharpness parameters as opposed to unsharp mask in DPP.

[Diane Miller]

Ive been processing as per your manual but I'll double check later today and make sure I didn't let that setting slip by. Don't have any to post, deleted them, but will try to re-create it. They are very tiny glitter-like things scattered about on edges.

[arash_hazeghi]

yes post an example, I maybe able to help.

[Diane Miller]

I found an image at ISO 1600 that was underexposed. Here are screenshots from two conversions, LR 4 and DPP, that were displayed at 100%. All parameters were at default settings except Exposure to + 1.4.

The DPP has less noise -- in fact, its default NR was too strong. There is some oddity somewhere in it's NR/sharpening, too. I double-checked -- it was Sharpen, not Unsharp Mask. Setting was 3. If I went to the NR/ALO tab and reduced luminance NR there, the appearance was a little better, but it only matches the LR appearance at 0, where it is equally noisy. DPP just looks funky to me in some details. (Probably only noticeable at 100%.)

Look just ahead of the wing leading edge, at the small feathers. They are just strange in DPP and appear much more natural in LR. In some instances I've seen this sort of thing look even worse. Still looking for one to show the little sparkles. Bottom line, every time I've tried DPP I haven't liked it, and the workflow hassles for a LR user are an added negative.

(Canon 5D3, 600mm II)

[Diane Miller]

Converted a different image and found the flatness I referred to above. It happened when I lowered the Highlights more than 1 notch. Increasing Shadows is also not very pleasing, at least in this image. I can often take both those sliders to their full range in LR 4 and get very pleasing contrast reduction. Sometimes a little Clarity is needed to restore midtone contrast.

Side by side with this image (looking at wing feather detail in a female House Finch), LR gives me a more pleasing result when I zoom in.

[arash_hazeghi]

Hi Diane,

I am not sure about your observation and conclusion , the LR output is way too noisy to my eye, it is going to be difficult to remove the coarse grains. The grain sometimes gives the false impression of detail. it is noise.

Looking at DPP, I think you are using luminance NR, which you must turn off. use chroma NR 8-10 luminance 0. You will get poor results if you use luminance NR in DPP- bad parameters.

I also don't see the "white boxes" you were talking about. where are they?

IMO the LR output is very poor based on your own example. if you apply NR the dark area will look murky and detail-less. it will show in a large print or on a HD display.

[arash_hazeghi]

Here is an example (100% crop) of a sharp RAW file from DPP. 1DX ISO 1600.

the grain is very tight and details are tack sharp.


http://ari1982.smugmug.com/photos/i-.../i-5x5sdT6.jpg

[Diane Miller]

I haven't found an example of the glitter flecks yet. I'll post it when I do.

I see I did have Luminance NR at 3, although I see from my notes it should be at 0. (Is 3 the default? If so, it's a little disheartening that they would have chosen a less-than-ideal value.) I reduced it to 0 and now the DPP image looks virtually the same as the LR one, as regards noise, at 100%. But I realize that's not the end of the story. You spent some time in the monograph on going to a NR program after DPP, and I understand your point that the NR will be better than LR / ACR, with DPP having dealt with the noise in a camera / ISO specific manner. I'll do some testing there, too, for my images. I certainly haven't been delighted with any NR I've tried.

But the issue of being able to get better shadow and highlight detail in the new Process 2012 in LR 4 / ACR is a major item in its favor for me, except possibly for the noisiest images.

I'll keep messing with this. I do understand the theoretical advantage to DPP.

I think noise is like sharpening -- the only really good way to deal with it is to avoid the problem in the first place.

[arash_hazeghi]

Hey Diane,

I don't think it will look as bad as LR if you use the right settings. If you can upload your RAW file somewhere I can show this point.

The advantage of DPP is not theoretical. I am personally not in the business of theoretical photography. For me the proof is in the photographs and prints that I have made over the past 6-7 years

If I understand correctly Artie was using ACR for many years until last year, when he realized the advantages of DPP he switched over quickly.

Some do not like the interface, that's a personal choice... Keep in mind LR cataloging is a disaster, it consumes your disk space by creating thousands of nested folders and files, all in propitiatory Adobe format...at some point those libraries will become as useless as junk when Adobe drops support to force you into the cloud. I like to keep things simple and in MY control.

[Diane Miller]

We're getting ready to head out on a trip, so give me a while -- I'll keep looking at DPP (double-checking settings) and when I find a better example image, I'll put it in my FTP folder and send you a link.

Meantime, two questions:
You recommend Neat Image for noise, but I've found the demo too frustrating to even try. How does it compare to Nik Dfine?
And what about the additional leeway (over DPP) in shadow and highlight detail with ACR / LR Process 2012? I find them wonderful.

I'll be about the last person to join the "cloud." I'm a certified control freak. But I do like LR. I just hope that when its day passes I'll have enough time to migrate to something else.

[John Chardine]

Here is an example (100% crop) of a sharp RAW file from DPP. 1DX ISO 1600.

the grain is very tight and details are tack sharp.


http://ari1982.smugmug.com/photos/i-.../i-5x5sdT6.jpg

I would be interested to know what the DPP techs are for the owl image. Severe pixel-peeping, which I unapologetically engage in (!), shows sharpening halos around high-contrast borders like the pupil and iris. Perhaps DLO does this?

[arash_hazeghi]

Hi John, I do not see the halos you are talking about but If you notice carefully there is some moire in the image because the 1DX has very weak low-pass filter.

I don't use DLO for super-telephoto lenses.

[Don Nelson]

John
Are you referring to the halo around the black pupil or the edge of the eye? Or both?

[arash_hazeghi]

hey Don,

I am not sure is that is a sharpening halo. If you magnify any digital image to this extend, it will look like that.

This is because digital cameras have a Bayer sensor, so the edges of detail cannot be abrupt or "single pixel". There is always bleeding to and from neighboring pixels because of de-mosaic process. That is normal.

Also you don't really need to blow up an image to this level to find halos. sharpening halos are usually visible at 100% view or even much smaller size as we sometimes see in the images posted.

Note that unsharp mask has a higher tendency to produce halos, because it is applied to the final image post demosaic. versus the "sharpness" parameter in DPP is calculated during demosaic so it has a lower tendency in producing halos, but it is more prone to moire.

[John Chardine]

I've been doing some research on my new Sigma DP2 Merrill, which has the superb Foveon sensor and for that have been using ImageJ to look at sharpening. The Sigma RAW processing software actually sharpens at a setting of 0 but fortunately the scale runs down to -2 so you have the ability to turn it down (is -2 off?). My interest in this subject is that I don't want to send an image on to post processing with an unknown amount of sharpening occurring during the RAW development.

I could see an indication of sharpening in the 100% of the Great Grey Owl- just an overall look- so had a look using the simple technique I've been using with ImageJ. Attached are the results. First the image clearly shows the telltale sign of sharpening- a ring of darker pixels on the inside of the edge between the pupil and iris, and a lighter ring on the outside. The graph shows the grey value (luminance) of pixels along the thin yellow line from left to right, from ImageJ. You can clearly see evidence of sharpening in the forms of the bumps at each side of the transition from dark-pupil to yellow iris. With no sharpening, this line should ramp more smoothly with little or no evidence of bumps.

Arash- I really don't think this is caused by the de-mosaicing process, which would cause a blurriness to the transition, not these bumps.

Now I'm beginning to worry that DPP is sharpening RAW images even when the sharpness settings are set to 0 (like Sigma does!). Perhaps ACR does not do this. It would be worth a test I think.

Gail- sorry for this diversion but I think it is relevant to your OP.

[Don Nelson]

John
Arash/Artie flow sets Sharpening >0, which does raw sharpening. It would be informative for Arash to repost with Sharpen set to 0.
The effect seen isn't due to Bayer Pattern or anti-aliasing filter, but the edge contrast enhancement that passes for "sharpening"
Don

[arash_hazeghi]

John, you are working on a compressed JPEG, if you want to use numerical tools you need to be working on TIFF directly. There is also compression artifacts. Your curve does show edge high-pass. it's not bad IMO though.

Of course Sigma is not a Bayer sensor. There is no demosaic and therefore there is no need to cancel its effects by applying a high-pass filter. It should not need much sharpening. RAW files from other cameras will look soft in comparison because of Bayer demosaic and optical low pass filter so sharpening is necessary.


Don is right I do not set sharpness to 0. Because detail that is lost due to demosaic cannot be recovered later by unsharp mask. you can reduce it if you wish to. There might be other artifacts too at this mag, but it does not affect the output.

getting a bit off topic here.

[arash_hazeghi]

I've been doing some research on my new Sigma DP2 Merrill, which has the superb Foveon sensor and for that have been using ImageJ to look at sharpening. The Sigma RAW processing software actually sharpens at a setting of 0 but fortunately the scale runs down to -2 so you have the ability to turn it down (is -2 off?). My interest in this subject is that I don't want to send an image on to post processing with an unknown amount of sharpening occurring during the RAW development.

Off topic, but I bought a Fuji X-E1 that uses a propitiatory Fuji sensor with a non-Bayer color filter and no optical low pass filter. The JPEGs from this camera look like RAW files from DSLRs...you should also look into this camera if you like compact cameras...the ISO performance is better than DSLR APS-C sensors.