Adobe Camera Raw 3.3 quality?

[peterkrogh]

Rick,
Your conclusions?
Peter

[Rick McCleary]

Peter,
I was actually waiting to hear back from Mike.  His initial post expressed frustration with the ACR conversion.  I'm looking forward to hearing his reaction to my attempts.

However, based on my test, I come to the same conclusion as I have in the past:  C1 still gives me a more detailed, more open file.  Of the files Mike sent, the pub window shot doesn't show a lot of difference between ACR and C1. The Clinton library shot demonstrates a greater difference between the two processors.

As an aside, both of these files lent themselves nicely to being corrected in Lab.  I was able to get more dimension and and color fidlelity in Lab than I was in RGB.

This is fun!  Now, back to the game.  (Currently, it's tied at halftime between LSU and Texas.)

Rick

[Rick McCleary]

Peter,
BTW, I'm only able to compare the ACR and C1 conversions.  I haven't been able to look at the DxO conversion that Mike sent to you.

Rick

[Michael_S]

Rick,
Sorry I was not able to respond as quickly as you could.  I was swamped with a Request for Proposal response to a major defense contractor.  But I did get a chance to look at your results.

I'll email you the link to a high quality JPEG made from the DxO TIFF of both photos.  I'd like your opinion on those.

My opinions of your results:

Wine Cellar -- I like the Capture One conversion a bit more than the ACR.  There seems to be a broader dynamic range of colors.  Also the colors seem generally closer to what I recall.

Clinton Library -- I'd call that a tie.  But they aren't the same, if that makes any sense.  Capture One appears to have a fill light feature that brings out the shadows better than I can in ACR.  Although I've been playing with Curves and improving that.  But C1 clearly has the edge on that.  I think the ACR conversion is a bit closer to the actual lighting and white balance of the actual scene.  Both conversions are much brighter than the "actual", even counting on the fact that I mistakenly under exposed it "as shot".  But I have done the same thing to some extent.  Otherwise, it would be a bit dreary. 

For comparison, I do my conversions in ProPhoto RGB most of the time.

I can't tell you how happy I am to have more experienced photographers look at my stuff and try their hand at converting the raws.  Thank you again.

--Mike

[peterkrogh]

Mike,
I played with it a bit tonight, and here's what I thought.
The DxO conversion is quite different from ACR, and in some ways more pleasing.  I has a smoother texture, which first came across as nearly grainless, but later started to look to me a little plastic and "digital"

The part I liked best was a certain localized contrast.  I would be interested to see what could be done in DxO trying to keep the shadows from clipping for further adjustment in Photoshop.

I'll send a copy of the file once I have had a chance to look at it with fresh eyes tomorrow.

Can DxO open a DNG file created by Adobe?  If so, I would consider it a good candidate to add to my RAW processing stable.
Peter

[Rick McCleary]

Michael -

I agree with Peter about the "plastic" quality of the DxO conversion.  Perhaps this is inherent to the DxO engine, or perhaps the noise reduction feature is just being overdriven a bit.  In any case, I actually like a little more noise than what I see in the examples you sent.  To me, no noise looks fake.

The main thing I look for in evaluating a RAW processor is how it maps the tonality of the image - things like separation in the highlights (or 1/4 tones), separation in the shadows (or 3/4 tones.)  To me, this is the area that distinguishes one processor from another.  If I read your notes right, you used the default settings when processing these files in DxO.  I'd be interested in seeing you deviate from the default settings to bring out more shadow detail - particularly in the wine cellar shot.  The shadows look clipped.

Rick

[Michael_S]

Peter,
Seems as if you are seeing it similarly to me.  I think the default settings on noise suppression in DxO are too aggressive.  Either reducing the amount or turning it off altogether might be a good solution.  Otherwise it can look like an advertisement from a Plastic Industry trade group!  Better living through plastics.....

I just played with the wine cellar file in Photoshop and discovered that the default setting histogram came out underexposed.  A quick Levels adjustment of only the white point yielded a much more pleasing result generally.  However, the detail in the window reflection was lost and it looked much more like Bibble, ACR, or C1 in that small region of the image.

I thiink the DxO noise suppression removed some of the wood grain details in the wine racks as well as the texture of leather seats.  Lost some 'reality' in the process.

As far as I know, right now DxO cannot accept DNG as input.  Also, their DNG output is linearized, but ACR can read it.
DxO is neither very fast nor easy to use in its interface.  Try the trial download and see how you like it.

I have read that some people save it for "tough" images.  But a few others are making it their main converter (e.g., Bob Johnson of earthboundlight.com and regular at nikonians.org).

As a side note, I'm still grumbling that PC users don't have a crystal clear path to an all-raw workflow.  My hopes are on Lightroom at the moment.  I'll give Bibble Pro 4.6 its final investigation this coming weekend before my trial license self-destructs!  And then there's the whole Nikon Capture NX later this spring.

Meanwhile, I'll send out a link for the JPEG of the Clinton Library where we see the rookie (me) try using Curves in ACR for the first time!  Actually I think it looks good.

[jeremyrh]

If you guys could put together a web page with your results, parameter settings, key areas of difference etc that would be a really useful document.

regards,

Jeremy

[Michael_S]

I'm open to that concept of a web page if someone has the ability and time to host the images.  Especially if it is educational to anyone.  I've used a hadnful of images as a way to explore more of the converters.  I've even tried running the library shot through RawShooter Essentials 2006 just to see what it does with it.

[Michael_S]

Jeremy,
Would you like to take a crack at the files with Nikon Capture?  I would love to see the results sinec the trial time-out policewon't let me look at NC again even though I removed it only after a few days of loading it.  I'll wait for Capture NX to try it all again........

Anyway, I think it would be great to see what you can come up with in Nikon Capture.  Like I mentioned at Nikonians.org, I had a tough time with the interface.  Now that I feel more adept at raw converters in general, I wish i could take a stab at it again for a few days next month.  I suppose if I knew the secret files or registry changes it made I could be set, but I'm not that PC savvy.

Let me know if you want to take my two NEFs for a spin with Nikon Capture.

--Mike

[Michael_S]

Rick,
Yes, I didn't change any strength settings in DxO.  Whatever tool it suggested be run against the file I let it have at it.

Definitely agree on the noise reduction.  I think it could be reduced and yield a better image.  My trial of DxO is over so I cannot try any of this again.

Also, I agree about the shadows in DxO.  Simply taking the TIFF it made and resetting the white point yielded a much better image.  I expect that manually watching the exposure and white/black/gray points would give nice results.  But that's only informed speculation without being able to test it.

Thanks again for participating and discussing the results from different converters!

--Mike

[AlanDunne]

Michael et al,

some comments about DxO. When you put it in the full autopilot mode it will frequently create "plastic", over processed look, especially with tough images. It can produce some amazing results, but the user interace is quite difficult to use to get those results.

Since the objective was to compare ACR, if have only tried Michael's files with ACR. Have have been quite busy for the past few days and was unable to re-join the dialog. How is everyone sharing the results? I could give both Capture and DxO a whirl as well.

Let me know how to share the results!

Cheers ... Alan

[AlanDunne]

Here are some other thoughts on this raw conversion exercise. So far I have only converted the wine rack image ...

• I found this suprisingly challenging. This was the first time I had attempted to do a raw conversion on someone elses image. Not having taken the image, I was unsure what the intent was for the image. I was lost without a roadmap. Sometimes you want to see detail in the shadows, or "openness". Sometimes the best interpretation call for shadows to turn into an inky blackness. I did not  know which direction to take.
  
• Further complicating things, I do not own  D50. When I first pulled these images into ACR, the D50 defaults were set to full auto. So ACR went off and came up with its own interpretation. My first viewing of the images was with this ACR auto, and that coloured what I thought the image should look like. By the time a set auto off to be the default, the damage (in my mind) was done). In the wine pack picture, ACR's auto tried to drastically reduce the exposure setting, I believe triggered by the bright, bordering on blown out, highlights in the window to the outside. The the darnkness of the interior that followed was compensated by cranking up the brightness setting. I ultimately found a custom curve gave better results.
  
• In the end I settled for a "technical" iterpretation, trying to bring detail to the shadows since the purpose of the exercise was to put ACR through its paces. Although I did want to get a solid black point as a foundation.
  
• Since the exercise was primarily about raw conversion, it did not do further processing in CS2. If I wanted more shadow detail, then I would have played with shadow/highlight. In fact I would also look at mutiple conversion (2) and layer blend them in CS2 with masks, but I ahve not done that and I do not think that was the purpose of this exercise.
  
• As above, the shadlow/highlight controls in CS2 are similar to what can be accomplished by D-lighting in captue and DxO-lighting in DxO, although shadow/highlight is not done as part of the raw conversion
  

Cheers ... Alan

[jeremyrh]

Jeremy,
Would you like to take a crack at the files with Nikon Capture?  I would love to see the results sinec the trial time-out policewon't let me look at NC again even though I removed it only after a few days of loading it.  

Michael - I'd be happy to contribute, though time is limited just now, with work, and also spring has arrived here in Holland, so I am planning to do more photography and less of what my wife calls "playing on the computer" (whaddya mean, playing???   )

[AlanDunne]

Well I have run Michael's two images through ACR, DxO (manual) and Nikon Capture. It is interesting to compare the results. I made no effort to modify white balance, nor impart any artistic judgement (not that I posses any), as I think the exercise in question should be focused on lighting, tonal range, shadow/hightlight detail, from a technical perspective.

I will set up a web page and post my results in a day or so.

Cheers ... Al