martindpr: The images from the 5Ds have a huge redundancy in detail. Decrease a sample to 24MP, then increase it back to 50.6MP and you'll get the exact same detail as the original image. Try that for yourself with the pic no1: http://www.usa.canon.com/cusa/consumer/products/cameras/slr_cameras/eos_5ds#Sample_Images_Videos
None the less, the down-res / up-res is distinctly worse than the full 50 mp image. The 50MP image could be processed as you suggest for even finer results. Some folks will appreciate the high res image ... I suspect that scenic and some professional work will benefit.
I agree that the 20-30 MP range is the sweetspot for the 35mm full frame cameras. The 50mp range only adds somewhat to the sweetness. If I acquire the new camera, I would probably shoot it in APSH 30 mp mode for the majority of my work. My 1D-IV and 5D-III are definitely adequate for my current work.
The down/up sampled version has less detail / image crispness than the full 50.6 mp image.
DuckShots: I have L lenses. Will they be sharper? I love all of them. Will they be less sharp? Cannot spend more on glass. Don't have the money for a medium format or a Leica. Is this the camera of my present dreams?
Indeed .... all of your images with any of your lenses will be sharper
The camera sounds interesting. I suspect that the common shooting will be done in APSH 1.3 crop mode giving a 30MP image. Use of 51mp will be used "where appropriate".
I will be curious to see of the R version will deliver better results with current Canon lenses. Of course, all current Canon lenses will produce "sharper images" with the 51MP sensor on the R or S versions.
At least Microsoft does see that raw files are "important" to the users. They decided to provide this service after Apple started to support raw :) Too bad that Adobe will not provide a codec for PSD files - I have to use Fastone or Bridge to see a full directory thumbnail set.
johnmcpherson: Rangefinder cameras are best for low light photography. Period.
Those of you who are critical of this camera; you need to use it by candle light or, in a dimly lit room; then you will see the advantage.
The modern DSLR cameras will generally beat the human eye ability to focus when you get into the range of 4EV to 0EV. The rangefinder is good - but not that good. The co-incidence rangefinder of the Leica does not perform very well at low EV with lower contrast or certain fine patterns that are no problem with a modern DSLR.
On the other hand, if you do have superb eyesight (and I do mean superb), the rangefinder Leica will work well in low light.
I really appreciate the Leica concept - I have 4 M film bodies and 9 lenses. The M2/3/4/5 were the best low light cameras available in the 1960-1980 time period simply because of the rangefinder focus system. Of course, it was also the best camera for the reportage/H.C.Bresson style of work.
The new digital M cameras are so lack-luster in low light sensor performance that I have avoided purchasing a digital M. The new Leica M-E (and probably the M) in no way encourage me to purchase one of these supposedly high end cameras. I was hoping the current new series would have offered a sensor that would be in keeping with the Leica craftsmanship of the body and lenses - IMHO, they certainly have not done that.
Seems to me that my only practical option is to try to find a second-hand M9 at some "reasonable" price - simply so I can get some use of my many excellent Leica lenses. The M-E and M simply are out of bounds for what I would term "marginally realistic price/performance".
tony field: Watch out!!! If you are a serious user of Bridge6 and rely on reasonable performance with a LARGE number of images in the cache, ACR 7.1 will cause all of the images to be re-cached when you access an image directory.
For example, I have 700,000 images and am reconstructing the Bridge Cache to comply with the new SQLite file implementation. It has taken me 10 continuous days of 24 hr/day to process only 380,000 images. The new release of ACR 7.1 will negate this completely and I have to spend another 10 days simply to reprocess the what I have already done.
In fact, I think the reprocessing will be significantly longer than the 10 days since all of the raw files must be reprocessed - effectively negating the use of the BridgeCache files recorded in each directory.
It certainly is slow - no doubt. However, that is the technology with my Dlink network drives - it was the fastest available at reasonable cost when purchased a few years ago. I would hate to purchase two more four-bay systems just to gain a bit of speed on the cache rebuilds.
These ACR inspired rebuilds present no tangible benefit when browsing images - in particular, since I use the default 1024 pixel preview size.
I have no issues with the speed otherwise - the cached previews are "instant" since they are on SATA. Actual image load is quite acceptable when fetching the raw data.
The ACR rebuilds and other cataloguing issues certainly perplex me since Adobe claims that Bridge is a proper DAM and it certainly is not. Lightroom is closer to be a photographers DAM but still misses the mark for things I consider important. I have even written a pre-processor to provide additional DAM features for Bridge (which might even be portable to Lightroom).
I need all images cached - I have to go back to history often for various publishing reasons.
One of the attributes of a decent DAM is that you can quickly browse all of the image assets. As it turns out, my image archive is on networked drives (1 gbps network, the drives work at an effective 275 mbps) with the cache on a sata drive for browsing speed. If ACR chooses to rebuild, everything slows down since the cached images become almost useless while ACR rebuilds
My computer is more than competent enough.
Watch out!!! If you are a serious user of Bridge6 and rely on reasonable performance with a LARGE number of images in the cache, ACR 7.1 will cause all of the images to be re-cached when you access an image directory.
Humm, I like the interface. The only thing I immediately noted was that ACR no longer has a "fill light" - new controls to use :)
Humm, what happens if we magically have sensors that capture a 15 (or higher) stop dynamic range and suitably process the tones to fit our (screen/print) viewing devices? That is "high dynamic range" by all counts and the only difference is that it can be accomplished with a single exposure rather than multiple exposures.
You could even argue that shooting with a higher dynamic range medium format studio camera of today would be "inappropriate" since it captures more than the run-of-the mill Dx3 or 1D-IV.
Just a point in passing. Working with a couple of images, one with flat lighting on the face and the other with higher contrast with strong facial shadow, the "red channel smoothing" technique does not work too well on the darker lit images - other smoothing operations (if smoothing is actually desired) should be used - possibly by editing red channel layer for luminosity. For the darker images, the luminosity bump becomes difficult to control unless you use a very low opacity on the smoothing layer and thereby largely defeating the smoothing operation.
The basic retouch operations are, of course, "spot on" :)
Interesting article. My recap of what Jean said is:
1. Retouch the skin "as necessary"2. Perform any other adjustments "as necessary".3. Smooth the skin "as necessary"
The tutorial certainly presents a straight forward Photoshop technique to accomplish the above three steps. The presentation is well done, and the technique works well (I just edited an image using this style).
All in all, a well done article - certainly taught me some new things about Photoshop that I never knew about before.
The other thing it taught me is that the "retouching war" about appropriate retouching is as filled with opinionated and meaningless commentary just as the "camera war" is about the right tool for shooting.
techmine: personally, I hate beautifying faces with Photoshop :-)
Quite a fine article. My recap is:
1. Retouch skin as "necessary".2. Smooth skin as "necessary".