Given the slowness of DxO 3.5 I find I need to batch process the images. To find the "best" settings I experimented on about 100 photos form a recent trip to Israel that were shot on a D200 with the 18-200 Nikon VR zoom. I tried 4 settings (except as noted factory defaults), outputted to DNG in Adobe RGB and viewed in Abobe Bridge on a calibrated Apple 23 inch LCD:
1. Factory default
2. Bob Johnson's default- slight exposure compensation, 75% vignetting vs 100% default and -0.25 sharpening vs default 0.0
3. Slight- Bob Johnson's settings but with slight rather than the factory default medium DxO lighting
4. Toned Down Settings- Bob Johnson's settings but with 50% medium lighting vs auto (auto initially opens at 70%), luminance noise 60% vs auto (initially opens at 70%), fine detail preservation 50% vs auto (initially opens at 0%).
I was really not able to evaluate sharpening and noise but otherwise my wife and I reviewed the 4 sets of pictures and rated the situations when there seemed to be obviously better and/or obviously worse results. In many cases the differences were small. Interestingly the toned down setting on occasion produced color/temperature changes (? because of having a fixed rather than an auto adjustment?) which did not occur with any of the other settings. Sometimes this was an improvement of an inappropriate white balance. At other time the picture became too blue. Never did the picture become redder/warmer with the Toned Down settings.
Overall, 2/3 of the time when there was a clear preference we preferred Bob Johnson's settings. One quarter of the time the toned down setting was preferred and 10% of the time the factory default was preferred. In most cases the factory default and Bob's default were quite close in appearance.
On the other hand, when one setting was clearly worse than the others, 60% of the time it was the Slight setting and 1/3 of the time it was the Toned Down setting.
For me then for now the factory default with slight exposure compensation will be how I batch process my raw files. Then I will open in Adobe Camera Raw and hopefully minimally process the majority (based on the above experience about 80%--Bob Johnson settings and the very close factory default setings results--should be in good shape) to be printed with automatic actions in most cases.
I'd be interested in other's experiences, observations.
I, too, appreciate all the experimenting. I have a new D200 on the way, and I am curious about one thing in your test: what settings were you using in ACR to view the DNG's from DxO? Were they the ACR defaults for the D200? Were any of them set to "Auto"? I have come up with some settings I find OK for the D70, but I suspect I will need completely different ones for the D200.
As Bob stated I've viewed everything with ACR off (command-Mac or control-windows U if you have not changed your ACR default). That way you are looking at what DxO did, not ACR. After opening in ACR I do like to see what the program thinks it should do automatically and after DxO the ACR automatic results seem better than those before DxO. Obviously the next step might be to come up with post-DxO ACR defaults for the D200. Let me know what you come up with.
For me, I want to next set up simple Photoshop actions--adjustment layer of auto levels and some kind of standardized sharpening in a new layer. I would then save as psd and should be able to use hopefully a lot of photos as is at that point and for 4 x 6 and just flatten the layers and do more, change sharpening etc. on those that need it or special pictures. Image Processor should then be able to convert the psd to JPEG.
Someone who knows more, Bob? should correct me if I'm wrong about what I just suggested. Also while I know every picture is different I'd love to be able to automate basic sharpening that would work reasonably for small prints most of the time. Is there a suggested USM, smart sharpening setting? Would Photokit Sharpener (which I think you use Bob) or some other plug in do the trick?
I have my ACR defaults with auto completely turned off. I have sharpening set to "preview only" since DxO has already done its thing and any additional sharpening that may be needed gets done later. Everything else is using the camera defaults I believe.
If I end up not liking the way a DNG looks in ACR I might click on some of the auto boxes to see what happens but rarely leave them that way.
I also have all "auto" turned off for my ACR default, but for my D70 I am not using the ACR default camera values, either. It sounds like you and Eliot *are* using the ACR camera defaults. When the D200 gets here I will try that first and see what happens. If you guys had some dramatic changes from the default to suggest, that would have saved me some startup time with the new camera, but it sounds like the default should get me started.
FWIW I also have ACR sharpening set to "preview only" and use Photokit Sharpener for all sharpening.
Hello, This is just a quick note to let you know that I am enjoying this thread about DxO and the right settings for incorporating into an Adobe-centric workflow.
Between this and Bob's June 18th review I became motivated to download a trial of 3.55 and try DxO again.
I think I am getting great results for a start on auto, but look ofrward to applying your research from this thread to make it even better. I have a mix of supported and not-as-well supported lenses for my D50.
Late last night I took my experimentation all the way to a final 4x6 set of prints through my Epson PictureMate Deluxe. The on-screen images look very, very good. But the prints using Photoshop CS2 look atrocious. Green vegitation and blue sky look all clipped and pixelated; no detail in the print. I'm not sure where this all comes apart so I'll research this for another day or so before I call for Nikonians help.
Maybe I'll learn enough to contribute to the discussions, but I fear something "easy" is going wrong and I'm missing it. But meanwhile, I look forward to seeing more discussion about good DxO and ACR setings for this type of workflow.
Hi Bob, I did DxO (any of the 3 modes) to DNG. DNG to ACR. Correct Exposure there and then open in Photoshop CS2. Adobe RGB colorspace from the start into PSCS2
From inside photoshop I print by 'Print with Preview...' and get it fit into the available paper space, etc. "Let photoshop determine colors" using a printer profile I got from another message board with Perceptual Rendering intent. It was almost 2am when I did the bad test prints. I may have been using Relative Colorimetric intent. May have been cramming some out of gamut colors on the edge of the gamut giving that flat monochromatic look in some greens and blues.
I am ashamed to admit I cannot completely reproduce the problem. I think some of it was banding caused by the print head. But the rest may have been out-of-gamut rendering problems.
I'm looking at what I have gotten in new prints today, at they look sa good as I think you can make a 4x6. Adding some 60 to 75% 1.4 radius smart sharpen makes them stand out better.
When starting the whole process in DxO, are you correcting exposure there before converstion? Or do you fix that in ACR afterward? Same for White Balance? Auto WB on a D50 can be weak sometimes and not get it right.
Thanks and sorry for the apparent false alarm on printing.
I did not change the Calibration values but rather the Exposure, Shadows, Brightness, and Contrast values on the main tab. Specifically, I use: Exposure: 0.0 Shadows: 5 Brightness: 70 Contrast: +25
These are now my defaults for the D70, but IIRC the Brightness default for the D70 was 50. I don't remember what the Contrast was.
In thinking about this, these values were from the CS days, before there was an "Auto". When I installed CS2, it changed the camera default to "Auto" for all of these settings, but I saved the numbers above as my default for my D70. What numbers do you have as the default for your D2x (which I presume are the values used for viewing the DNG's from DxO)?
Those values certainly affect how a DNG created by DxO looks when pulled into PS. And while I agree that you probably don't want to change Exposure, or possibly Shadow, doesn't Brightness control the gamma correction of the linear RGB values in the DNG into a gamma-corrected space (e.g. ProPhoto RGB or Adobe RGB) in PS?
In any case these values all have to be set to something when you import a DNG. I think Exposure at 0.0 for the DNG from DxO is fairly obvious, what about the other three values?
Thanks, Bob. That is basically what I was wondering about. I find myself tweaking the brightness up a bit with the D70 images, even from Dx0, but it sounds like I may not need to with the D200 (just couldn't justify the D2x to myself). When the D200 gets here I will do some experiments and report back.
>Yes, that is indeed what ACR calls "brightness." > >Truth is, while it is not to have the option too, I rarely >change any of the settings in ACR when opening the DxO >generated DNG.
Is there any problem with making adjustments to Exposure, Shadows, Brightness in ACR on the DNG form DxO when it still looks a bit dark or too bright? (Or do you prefer doing in Photoshop at this stage?) I found I wanted to do that up to make adjustments about 20% of the time (a combination of white balance, exposure/brightness/shadow, straightening horizon and cropping).
At this point if I make exposure adjustments is there any benefit to using the technique of holding down the alt (PC) or option (Mac) key while adjusting exposure to get rid of blown highlights? After DxO this required much less of an adjustment than without DxO but sometimes I found doing this with Exposure and then correcting the darkness with brightness left the picture look almost the same with extra work.
The DNG created by DxO is a three-channel (RGB) image but the data is still linear gamma so you would be better off correcting exposure (if need be) in ACR rather then once you have it open in Photoshop. using the Alt (Option on MacOS) key is indeed the way to go if you do need to adjust Exposure (white point) or shadows (black point).
> >FWIW I also have ACR sharpening set to "preview only" and >use Photokit Sharpener for all sharpening.
I also asked this of Bob in this forum but I'd be interested in what you do. Can photokit output sharpening be done in a batch mode? (I presume since you use DxO you don't use capture sharpening and only use output sharpening) Is it only based on output medium? Or it it also includes types of photos--portrait, fine detail, etc. can this input coupled with output medium allow batch settings?
I had of course read your tutorial on Smart Sharpening previously as well as a number of your other tutorials.
I want to understand a bit more about Photokit Sharpener because maybe it is what I'm looking for. It sounds like you could batch process a set of images if they came from the same source and were being output to the same source. For example, if I had pics from my D200 that were going to be made into 4 x 6 prints. As for capture sharpening in Photokit, right now I am doing first pass batch processing of images with DxO which includes lens and camera specific sharpening which I would presume is equivalent to capture sharpening. If I am interpreting what you said correctly I could then follow with Photokit Sharpener for the output sharpening. Am I correct here? Do you find the settings different for Photokit if you have already used DxO lens specific sharpening?
One last thing confuses me about Smart Sharpen or USM settings. Playing around I find I like the look of higher radius, lower % , sometimes at an extreme, which was a technique I saw one person (ASwash) talk about on Nikonians that is quite different from your approach and most others. For example settings like 10-20% with radius of 64. This adds a remarkable crispness without noticeable artifact that is particularly impressive in removing haze. While it does not seem to let you make out finer details the picture looks sharper with this or settings like 55% and radius of 5 with less artifact than when I've tried more conventional settings like 150% and radius of 1. At least this is true after batch processing in DxO. The comments by those who discuss the process are that it is increasing local contrast rather than true sharpening but I don't understand the difference myself. (See Using Unsharp mask to boost contrast http://www.luminous-landscape.com/tutorials/contrast-enhancement.shtml and the other links from that site)
On files already processed by DxO, I don't use any input sharping with PhotoKit. Output sharpening remains the same with or without DxO though, based on the output medium as normal. So I guess you could say that dxO Lens softness takes the place of PhotoKit input sharpening.
With any of the radius based sharping techniques, images with fine details tend to look better with small radius values and higher amounts while those with less fine details tend to benefit more from larger radiuses and smaller amounts.
As a sneaky way to adjust contrast, really large radius values can be quite effective in some images.
>On files already processed by DxO, I don't use any input >sharping with PhotoKit. Output sharpening remains the same >with or without DxO though, based on the output medium as >normal. So I guess you could say that dxO Lens softness >takes the place of PhotoKit input sharpening. >
Can photokit output sharpening then be done in a batch mode? Is it only based on output medium? Or it it also includes types of photos--portrait, fine detail, etc. can this input coupled with output medium allow batch settings?
I'll answer you here instead of up above to keep all the sharpening stuff in one place.
I always use Photokit Sharpener's Capture Sharpening when I do *not* use DxO (e.g. quick web images, emailed pictures, etc.) Sometimes I use Capture Sharpening with DNG's from DxO. I have not had the problem with over-sharp pictures that you have had from DxO - possibly due to the camera differences.
If I am making JPEG's for a website, distribution to parents at school, etc., I use an action I created that does the capture sharpening, resizing, output sharpening, etc. all in one step. I then use this action on a batch from Bridge. I am not sure if this is the way you are talking about doing a batch or not, but you could create an action with just the sharpening action you want then use that action for a batch of files. However, then you have to pick one setting for the sharpening and leave it at that.
As for output sharpening, I use Photokit Sharpener for non-printing output, but when printing I use QImage. I did some testing with 16"x24" prints and found that I preferred QImage's "click and print" output to anything I was able to achieve with Photoshop.
> >As for output sharpening, I use Photokit Sharpener for >non-printing output, but when printing I use QImage. I did >some testing with 16"x24" prints and found that I preferred >QImage's "click and print" output to anything I was able to >achieve with Photoshop.
That's a new one to me. I haven't heard of QImage. Is it available for a Mac?
I don't remember doing anything special when using Photokit Capture Sharpening with DNG's from DxO. In other words, I just used the appropriate edge size and resolution for the camera. I might have ended up making the sharpening layer group less opaque than normal for the DxO images, but I don't remember.
Theoretically lens sharpening and capture sharpening are two different things, but in actual practice it might be overkill. I will have to pay more attention to exactly what I am doing in this situation.
>On files already processed by DxO, I don't use any input >sharping with PhotoKit. Output sharpening remains the same >with or without DxO though, based on the output medium as >normal. So I guess you could say that dxO Lens softness >takes the place of PhotoKit input sharpening. > >With any of the radius based sharping techniques, images >with fine details tend to look better with small radius >values and higher amounts while those with less fine details >tend to benefit more from larger radiuses and smaller >amounts. >
Having not tried Photokit I don't have a good sense of it even though I looked at their pdf manual. I'm trying to clarify things about it enough to see if it would work for me. For output sharpening does it primarily use the medium and size which would be easy to batch process. Or does it combine this with type of image--more or less detailed. I guess the later might work too if I separated images into a couple of groupings.
What I am trying to do is batch process a large number of images from a trip to 4 x 6 prints. The DxO step with some fine tuning in ACR for about 20% of images has done well. In fact in only 1 case, a sunset that DxO dulled the bright sky too much, of a few hundred did I feel the need to go back to the original nef! (DxO also tended to overbrighten the night sky in some pics but this was readily corrected in ACR.)
>Photokit goes basically just by output medium and >size/resolution.
So it is only if you use the creative sharpening (I believe that is what it is called) that you view the picture at 100% and adjust sharpening? Am I missing something as it seems I could just take the DxO output, resize to 4 x 6 and 300 dpi and apply Photokit for output if I did not think further processing is needed? Or does everything require/benefit from the creative sharpening process?
The need for creative sharpening depends on what you are trying to acheive. For most images that attempt to literally record what you saw, there is no need for it at all. If you want to render more the experience of seeing it though, you might choose to emphasise some aspect of the subject. While "always" is hard to use definitively, I think it's fair to say that creative sharpening would always involve some form of selection or masking and thus could not be done batched.
Output sharpening is another matter. Indeed what Photokit attempts to do is to automate the decision making process, making it so you can batch it.
>The need for creative sharpening depends on what you are >trying to acheive. For most images that attempt to literally >record what you saw, there is no need for it at all. If you >want to render more the experience of seeing it though, you >might choose to emphasise some aspect of the subject. While >"always" is hard to use definitively, I think it's fair to >say that creative sharpening would always involve some form >of selection or masking and thus could not be done batched. > >Output sharpening is another matter. Indeed what Photokit >attempts to do is to automate the decision making process, >making it so you can batch it.
This is a great suggestion! Is there another way to quickly batch by selecting folders, without adding images from the other window? For me, this is what I hate about DXO the most. It takes so long to process 800+ images.
>This is a great suggestion! Is there another way to quickly >batch by selecting folders, without adding images from the >other window? For me, this is what I hate about DXO the >most. It takes so long to process 800+ images. > >Chris Baltazar >Northern VA
I select them all and do the settings and let it run overnight.
Following this discussion about DxO settings I have run a series of experiments using what are probably the most difficult types of images for automatic processing portraits. I find that the skin tone only has to be off a little for the whole image to look bad.
I found that the factory default settings in DxO did a really good job with only three changes.
a) I selected DNG as the output format
b) I selected Adobe RGB as the color space (this is the space I use as default generally)
c) I changed the local contrast setting to 50 (the default is 35) this very much improved the skin tones in the photos, I dont know why I just hit on it by trial and error but it seems to work
Color space is irrelevent for DNG output. The data in the file is still linear and not in any color space. Since ACR freely allows you to select color space when converting the DNG (and indeed ignores any color space tag in the file), you can defer color space choice until ACR.
If you are creating output files from DxO in other formats though, the color space choice is important.
>Color space is irrelevent for DNG output. The data in the >file is still linear and not in any color space. Since ACR >freely allows you to select color space when converting the >DNG (and indeed ignores any color space tag in the file), >you can defer color space choice until ACR. > >If you are creating output files from DxO in other formats >though, the color space choice is important.
Ok, I tested the workflow earlier and it worked like a charm. Though I ran the same workflow last night with 800+ images for a wedding. I found that the DNG files are so much larger than the original NEF files. Does anyone have a fix for this or why thie DNGs are so much larger.
>Ok, I tested the workflow earlier and it worked like a >charm. Though I ran the same workflow last night with 800+ >images for a wedding. I found that the DNG files are so much >larger than the original NEF files. Does anyone have a fix >for this or why thie DNGs are so much larger. > >In comparison, the NEF is 9mb vs DNG @ 32mb.
Can anyone address the size issue? Am I the only one experiencing this?
You are not alone at all with the size difference.
For example, my D50 NEFs are around 5MB. The resulting DxO generated DNG file will be about 18MB to 20 MB.
Remember, that the DNG out of DxO is not like a camera captured raw file. The colors of each pixel have been determined and written in the output DNG (still no gamma correction, or whatever). In DNG Converter speak, this means it is a linearized file. The original NEF has been interpolated, and thus partially rendered, through a demosaic process.
Going all the way to TIFF "fully bakes" the DxO output for both color values in three channels everywhere as well as gamma.
You are seeing the behavior we should expect to see when we linearize a raw file image. But it is a bit "piggish".
Actually, just a small gramtical point: the gamma in the DNG has not been linearized since it was in fact linear to begin with. "Linearized" implies it was non-linear before hand. Applying a gamma curve to the eventual TIFF or other format makes it non-linear.
What has happened though as you say is that it has been de-mosaiced. So rather than a single channel, it now has three: red, green and blue.
>2. Bob Johnson's default- slight exposure compensation, 75% >vignetting vs 100% default and -0.25 sharpening vs default >0.0 >
Just want to check, I think I must be slow. When you say slight exposure compensation do you mean slight highlight recovery in the exposure comp window, or slight in the auto settings intensity of dxo lighting? thanks Gary
Just an addendum to my earlier posts in this thread. I forgot to mention that if you use DxO to output in DNG format and then open in ACR it is very important to to make sure that Use Auto Adjustments in ACR is turned OFF. I find the auto adjustments on top of DxO really clobbers the image.