NR/detail retention using ACR and DPP

Arash H. and I have started discussion in the following thread on the NR/detail retention of RAW conversions using the latest versions of ACR and DPP.

http://www.naturescapes.net/phpBB3/view ... 1&t=180241

I reckon it's better to open a new thread so as not to hijack the linked thread.

I propose to Arash that we do the following workflow:

1. I've uploaded the 7D RAW file (ISO 1600) to yousendit, it may be downloaded now by Arash and others who may be interested.

https://www.yousendit.com/download/aHlU ... TW52Wmc9PQ

2. Arash will try his best to convert the RAW in DPP to his taste, after which I suggest that he email me the recipe file (*.vrd).

3. I'll convert the same RAW file in DPP using Arash's recipe. I'll then try to emulate the noise levels in the shadows and OOF areas using ACR 6.2 RC (same exposure and similar contrast). I suggest both conversions use the as shot WB and standard picture style. Once I'm done, I'll email Arash my ACR conversion recipe (*.xmp).

4. Once Arash and I are in possession of each other's recipe (DPP for him and ACR for me), we can produce identical crops at both ends, which can be posted for comparison. When posting, I suggest we take side-by-side screenshots of both conversions at 200% view, and saved as PS quality 10 jpeg.

This should be a fun exercise.


Romy

I am looking forward to seeing the results,
Gail

The example crop you posted in the other thread exhibits the aspect of DPP's conversion that I like least -- the digital artifacting. Little horizontal and vertical bars, the result of forcing the demosaic interpolation to be either entirely horizontal or entirely vertical. They don't sharpen well and are hard to clean up with NR. It also seems that ACR did a better job of pulling out feather detail.

Looking forward to seeing the results guys.

ejmartin wrote:The example crop you posted in the other thread exhibits the aspect of DPP's conversion that I like least -- the digital artifacting. Little horizontal and vertical bars, the result of forcing the demosaic interpolation to be either entirely horizontal or entirely vertical. They don't sharpen well and are hard to clean up with NR. It also seems that ACR did a better job of pulling out feather detail.

Emil, those are my impressions on DPP and ACR as well.

But I might not be optimizing my settings at DPP, while Arash seems to prefer DPP over ACR. This exercise will either confirm my test crops are indicative of the relative performance of the two converters, or Arash has a good workflow to allow DPP to be better than ACR in NR/detail retention.

While waiting for Arash to do his DPP conversion of the uploaded RAW, here's another quick comparison.

Uncropped full frame resized to 750x500, Collared Kingfisher (Todiramphus chloris), EOS 7D + 400 2.8 L IS + Canon 2x TC, f/5.6 (wide open), ISO 800, 1/80 sec, bean bag:
I used the same WB (daylight) and picture style (neutral) in ACR 6.2 RC and DPP 3.8.1.0. However, the color response of both converters are not the same. I then tried to equalize the contrast and attempted to achieve a fine balance between NR and detail retention during conversion. Each file is untouched after conversion, except saved as PS quality 10 jpeg (and addition of labels/watermark).

When I compare both files, I usually open both in PS, paste one as a layer on top of the other, then toggle the visibility/invisibility of the top layer and view at 200%. This makes it easier to see the differences in NR and detail retention.

1. Link to ACR conversion:
http://a.imageshack.us/img42/443/img1053acr.jpg

2. Link to DPP conversion:
http://a.imageshack.us/img23/9919/img1053dpp.jpg

Hi Romy,
Sorry I have been very busy with a conference recently, haven't had much time for other stuff but here is a 5 min quick process with DPP, this is one noisy file. BTW the color/WB difference between ACR and DPP is very striking, one of the other issues I have with Adobe is the inability to replicate the camera WB and accurate colors that I can get with DPP. BTW, the KF file is a better example IMO, the first one is just too noisy and neither sw can pull a keeper at the end of the day, IMO. for the KF sample if you look at the uniform BG area DPP seems to be less grainy detail wise critical focus seems to be on the back feathers instead of the head so looking at that area ACR looks sharper but you can increase sharpening in DPP too, For files that are too noisy I sometimes turn sharpening off in the RAW converter and apply sharpening to the output only...

http://ari1982.smugmug.com/photos/96727 ... tRC-X2.jpg

Got your recipe file, Arash! Will work on the ACR conversion shortly and email you the ACR recipe.

ahazeghi wrote:BTW the color/WB difference between ACR and DPP is very striking, one of the other issues I have with Adobe is the inability to replicate the camera WB and accurate colors that I can get with DPP.

I would have thought this issue is largely gone these days with the profiler software and associated presets that one can download from Adobe Labs.

ejmartin wrote:

ahazeghi wrote:BTW the color/WB difference between ACR and DPP is very striking, one of the other issues I have with Adobe is the inability to replicate the camera WB and accurate colors that I can get with DPP.

I would have thought this issue is largely gone these days with the profiler software and associated presets that one can download from Adobe Labs.

Did try the profiler, not extensively though. If you know any profile that I can download that would make LR output exactly the same as DPP I'd appreciate if you can provide the link, I have tried equalizing the output by adjusting a dozen slider in LR but unfortunately I did not fully succeed, when I save the profile for one photo and apply it to another photo that was shot under different light, it often doesn't work... which means I have to process photos case by case, difficult when you have 3K photos and just a Sunday evening to go through

IMO it would be the best if Adobe paid Canon/Nikon and used their algorithms with their own efficient side tools plus their excellent workflow and programing efficiency, but greed on all sides will prevent that from happening.

Exactly reproduce DPP? Probably not. But as of ACR 5.2 Adobe includes "Camera Profiles" as presets for ACR which are meant to mimic the manufacturer's color profiles. Though neither that nor the manufacturer's profile is guaranteed to be accurate color; better to shoot a color chart in the range of light you shoot in and use the profiler to develop your own. The profiles are available for ACR 5.1 and earlier from

http://labs.adobe.com/wiki/index.php/DNG_Profiles

ejmartin wrote:Exactly reproduce DPP? Probably not. But as of ACR 5.2 Adobe includes "Camera Profiles" as presets for ACR which are meant to mimic the manufacturer's color profiles. Though neither that nor the manufacturer's profile is guaranteed to be accurate color; better to shoot a color chart in the range of light you shoot in and use the profiler to develop your own. The profiles are available for ACR 5.1 and earlier from

http://labs.adobe.com/wiki/index.php/DNG_Profiles

Thanks , these are the ones I had tried before...The Canon/Nikon neutral and standard profiles usually do an acceptable job, not 100% accurate but good enough for web and prints, with Adobe need to work a lot to get it right, even with what is supposed to mimic camera profile (can see from Romy's shots how much they differ). Agree best is to shoot a color chart would do so if I were a studio photographer but I am a BIF photographer for fun, so many images, so little time prefer to spend the time in the field and making photographs...

They also had a lot of issues with Nikon profiles, here is one thread by myself, never got answered http://forums.adobe.com/thread/468122?tstart=0 hope it has been fixed in the new version...

Ok, I've just finished the ACR 6.2 RC conversion and have already emailed the recipe to Arash. I'm posting below links to PS quality 10 jpegs from conversions using the recipes of Arash and I. I'm also posting links from which the recipes may be downloaded so the conversion can be replicated by anyone interested. The DL link to the RAW file is in the OP.

1. DPP 3.8.1.0 conversion, recipe by Arash H.:

http://a.imageshack.us/img707/3935/img0 ... recipe.jpg

DL link for Arash's recipe:
http://www.yousendit.com/download/aHlRe ... R0d4dnc9PQ


2. ACR 6.2 RC conversion, recipe by Romy O.

http://a.imageshack.us/img337/544/img03 ... recipe.jpg

DL link for my recipe:
http://www.yousendit.com/download/aHlRe ... R0kwTVE9PQ

I think the results can speak for themselves.

Yes, there's a difference in color response even using the same picture style and WB, but to my color-blind eyes, the ACR conversion has less or equal noise and a bit more detail.

Thanks to Arash for taking part in this fun and informative exercise. I'd like to hear his comments too.

Romy

Thanks for posting Romy,
to my eyes (viewing at 100%) ACR looks a tad better but it is marginal IMO, the color difference is more significant than he difference in noise and details. I am not familiar with this bird so don't know which one is the real color. Also it is interesting to note that each photo has different hot pixels it means that neither can remove all the hot pixels. Looks like Adobe output also has some luminance noise reduction with post sharpening, I prefer to do luminance NR with Neat Image that allows me to do variable noise reduction based on spatial frequency. That said, I never deal with files that are so noisy to start with since I have the 5D2 and MK4, for those cameras often noise is not so much and the standard chroma NR in DPP works well without needing luminance NR applied to the bird itself... The luminance NR in DPP is very poor so I never use it. Looks like Adobe have improved a lot since the last version as you mentioned, I sold my 7D before the new version came out so I had not tried so extensively, thanks for posting and discussion. For now I will stick with the free DPP to see what comes next! Thanks for taking the time to process and post...


Here is the side by side crops from the area that appears to be in sharpest focus...first one DPP second one ACR

liquidstone wrote:
Yes, there's a difference in color response even using the same picture style and WB, but to my color-blind eyes, the ACR conversion has less or equal noise and a bit more detail.

.

Romy

Why color-blind Romy?

ahazeghi wrote:Thanks for posting Romy,
to my eyes (viewing at 100%) ACR looks a tad better but it is marginal IMO, the color difference is more significant than he difference in noise and details.

I think it will be fair to say that ACR 6.1/6.2RC is at least equal or even a teeny-weeny bit better than DPP 3.8.1.0 in NR/detail retention. This has been my experience with both converters in hundreds of my recent bird photos. The difference in color response IMHO can be dealt with easily in WB setting and/or in PP.

ahazeghi wrote: Why color-blind Romy?

It's an in-born issue with my eyes (acutely color-blind), Arash. When converting my work, my 13-year old son sits beside me to assist me in setting the WB. My perception of fine detail though is I think good.

liquidstone wrote:

ahazeghi wrote:Thanks for posting Romy,
to my eyes (viewing at 100%) ACR looks a tad better but it is marginal IMO, the color difference is more significant than he difference in noise and details.

I think it will be fair to say that ACR 6.1/6.2RC is at least equal or even a teeny-weeny bit better than DPP 3.8.1.0 in NR/detail retention. This has been my experience with both converters in hundreds of my recent bird photos. The difference in color response IMHO can be dealt with easily in WB setting and/or in PP.

Agreed especially since the last versions was plain horrible this one is really good, just need the time to sit down and optimize the color and WB for each

ahazeghi wrote: Why color-blind Romy?

It's an in-born issue with my eyes (acutely color-blind), Arash. When converting my work, my 13-year old son sits beside to assist me in setting the WB.

Oh, I didn't know, I am really sorry, I apologize though it doesn't look like you are missing much and your son sure does a great job. Thanks again appreciate if you also post in the avian forum so we can see the wonders of south east Asia.

Best.

ahazeghi wrote: Oh, I didn't know, I am really sorry, I apologize though it doesn't look like you are missing much and your son sure does a great job. Thanks again appreciate if you also post in the avian forum so we can see the wonders of south east Asia.
Best.

I'm like a camera sensor - colorblind - and my son serves as my "color filter array".

I haven't posted in the Birds section for a while, will do so soon. Meantime, here's a processed and downressed version of the bird, plus some info.

Oriental Reed-Warbler (Acrocephalus orientalis)
7D + 500 f4 IS + Canon 1.4x TC, 700 mm, f/5.6, ISO 3200 equivalent (ISO 1600 with +1 EC in RAW conversion), 1/200 sec, bean bag, manual exposure in available light, 7.63 meters distance
And another high ISO shot, converted with ACR 6.2 RC:

White-breasted Waterhen (Amaurornis phoenicurus)
7D + 500 f4 IS + Canon 1.4x TC, 700 mm, f/5.6, ISO 3200, 1/200 sec, bean bag, manual exposure in available light, uncropped full frame, AI servo, IS on, 15.3 meters distance