Thursday, July 06, 2006

Who is Micky,anyway? The "small-pixels" guy,ya' know

On dPreview's FujiFilm SLR Talk there was a dust-up recently between Micky Rouse and Artichoke. Basically, Micky's contention is one that's supported by the Phil Askey review of the FujiFilm FinePix S3 Pro. Basically, Micky put together an all-new pBase user name and an all-new pBase account with one or two or three image files from an S3 Pro which, when converted through Adobe Camera Raw (hereafter called simply ACR),showed the "spots" in large, smooth-toned areas. Micky contended that the S3 Pro uses "small pixel's" which are "the size of those in my wife's digicam's" sensor. He pointed out,and rightly so according to Phil Askey's testing results, that converting S3 Pro .RAF files using ACR can cause "spots" to appear in some pictures. Well, wouldn't you know, I tried to make that same,exact point on Fuji SLR Talk some many months ago and it was not appreciated,shall we say.

The dual-pixel nature of the S3's sensor output CAN CAUSE SPOTS or digital noise to appear,especially when ACR is the raw converter used, and on certain types of scenes or subject matter, these spots MAY BECOME very,very readily apparent.The Askey review mentions this,and shows examples. Micky is correct.Micky put up some samples on his brand-new pBase account,showing how bad the S3 Pro's raw files can look when converted using ACR as the raw conversion engine. Point taken. I have to agree with him,and I had SEEN this negative consequence with my own eyes,before Micky pointed it out. Not all raw converters are created equal,and in some cases a particular raw converter may produce particularly BAD output, or unusually "different" output from other converters. Hey...it's a software thing. And it is also scene-dependent. But the thing is, shadow noise problems from the D2x, banding of several different types from the D200's sensor, track noise inunderexposed Fuji S2 files, and banding on EOS 5D files shot using some common lenses like the 50mm 1.4 USM Canon EF lens which apparently can cause some radio frequency interference problems for the EOS 5D's imager. These types of weird image artifacts are part of what Rob Galbraith refers to as "digital weirdness". We've had these types of issues with d-slr cameras for several generations, among Kodak,Nikon,Canon,Fuji,and Sigma d-slr models.

Moire, Italian flag effect, birefringence (or what Canon calls that nasty green and reddish and purpleish fringing when it's not lens-induced lateral chromatic aberration), moire problems on khaki pants and on Oxford-cloth men's dress shirts with the D70, "the jaggies" out of control on cat whiskers,diagonal architectural lines,etc on the Fuji models S1-S2-S3....good Christ--"digital weirdness" and various types of artifacting are part and parcel of d-slr use! Get used to it! Recognize it! Deal with it. Admit it when it happens. NO d-slr camera that I've ever heard of has been immune from _all_ digital weirdness. If one only prints large enough or looks hard enough, or shoots "the wrong subject matter" in a certain way, he will see that,yes,there are some digital weirdness artifacts to see,sometimes.


Anyway....I found out about the post some days after it had gotten started. You can start Micky's original post entitled "S3 Dots Explained" here,and watch it degenerate quickly http://forums.dpreview.com/forums/readflat.asp?forum=1020&message=18972715

As the post got going, I think Artichoke jumped to the defense of his beloved S3 Pro, and somehow missed the simplicity of one key part of Micky's original contention which was that that S3 Pro .RAF files when converted using ACR are MUCH more-prone to these terrible "spots", and that by default makes ACR a lousy conversion engine for S3 Pro raw files. Which, in my opinion, is simply a terrible thing, because ACR is so efficient and so amazingly capable a raw converter. ACR offers a good,fast,easy workflow,and the more I use it and learn its power, the better ACR gets. ACR also offers a good browser feature, and offers just wonderful Macintosh performance. FujiFilm's RAW-conversion and tethered shooting software releases have suffered from a long,well-documented history of terrible Macintsh support. Failures to install, problems with disk permissions under OS X, failure to install, and did I mention failure to install? FujiFilm had a littany of problems with software updating and upgrading gotcha's on Macintosh computers deployed all across the US and the UK, even among really knowledgeable power-users. The problems with getting Fuji's raw conversion software to install on Macs was almost unprecedented,in my experience.

Micky also contended that the S3 Pro uses a good deal of noise reduction as ISO values go upward. He pointed out the Phil Askey comments about how the S3 Pro uses detail-robbing noise reduction routines as ISO values go up, whereas the Canon EOS 20D does NOT cost one nearly so much detail as ISO values go up. Micky referenced URL's to some dPreview pages where Askey compared the S3 Pro and 20D, and so I went there and looked at the comparisons Micky pointed the forum towards. Hell, I had looked at those same comparisons months and months ago, and agreed with Micky--according to the Askey samples, the EOS 20D does NOT penalize the user with a lot of detail-robbing as ISO values go up. I am the guy,remember, who BOUGHT HIMSELF A 20D and a three-lens kit with the money he would have spent on just an S3 Pro body, back when the S3 Pro was still $2,499. I did a blog entry on that very thing, how the S3 Pro's price and feature set with warts and all, drove me a Nikon user since 1982,to investigate the EOS 20D's capabilities for myself.

The funny thing is, like Artichoke alleged, I too think Micky used to post in Fuji SLR Talk under a different user name. I say that because I recall _somebody_ using that phrasing "small pixels" and "my wife's digicam" many,many months ago,as a way of putting down the S3 Pro. I cannot for the life of me recall who used to post using those phrases, and Micky Rouse is a very new dPreview member and a very new pBase user also. Anyway, I still think Micky's right--the Askey review of the S3 Pro shows the problems ACR can have in decoding some types of scenes shot in raw with the S3, and that, yes, Fuji DOES APPLY some pretty high noise reduction at higher ISO's on its in-camera JPEG files. AND, and this is the big "and", this ties in with what I wrote on June 22-----that SOFTWARE IS WHERE IT'S AT. Interestingly, Micky Rouse opened his dPreview account on June 24. The fact that the S3 Pro uses an offbeat sensor design means that Adobe's raw conversion performance in blending of the S and R pixels is,at times, pretty out of whack,while the Fuji-designed software does NOT suffer from the same malady. So, ACR is not the best raw file converter for the S3 Pro, and high-ISO JPEG capture shots with an S3 Pro means a fair amount of noise reduction is needed to make the beautiful picture files Fuji S3 Pro users want and expect. So? Is this news to anybody? Isn't this information at least 12 months old, at the very least!!?? Hasn't this all been covered, months and months ago?

I dunno....Fuji, and Nikon digital SLR files, to ME, seem to SUFFER DETAIL-LOSS when noise-reduction is prformed on them,either in-camera or in software. I do not own the S3 Pro, but I have good eyes, and I have seen the noise reduction effects HU brings in its two quality modes,and I have SEEN the small-pixel noise problem in broad-toned areas from S3 RAFs converted in ACR. What was kind of odd to me,as I read through the thread Micky had started, was that Micky made MANY of the SAME criticisms of the S3 Pro as I did, back in the day. It was eerie to me,actually,and I wondered if he had read through some of my posts; and I'll admit it, when Artichoke implied that Micky used to post in FSLR Talk under another user name, I thought he might have suspected that _I_ was Micky. But, no, I'm not Micky. But I do agree with Micky's original post, in almost all respects. S3 files are not optimally converted using ACR---I think Artichoke made that point unequivocally clear,and rightly too, that S3 output can be handled best by Fuji's own HU software.

A point I'd like to make though, is that while HU seems to decode "normal" S3 raw files better than ACR, there is some very strong evidence in the Askey review that ACR has BETTER highlight recovery capabilities on grossly over-exposed RAF files than HU offers on the same badly over-exposed files. And, in my opinion, the SAME, exact thing is true on Nikon NEF files which have been severely over-exposed: ACR kicks Nikon's capture's ASS up one side and down the other when you've over-exposed very,very badly and need heroic rescue measures on burnt-out highlights.. ACR seems to me to offfer MUCH better recovery on those really,really F-'d up over-exposure cases. C'mon, you know what I'm talking about....camera's in M, you think it's still in A, you wind up 3 stops over....ACR will save your ass, but Nikon Capture will not.... Small point, but Askey makes it in the review of the S3 Pro,and he faults the Fuji software for this. Must say, there's a point there....

Anyway, Micky directed some dPreview FujiFilm users to my blog spot, so now you can read a little bit of my take on it. And yes, I did go to Artichoke's pBase site and waded around and left some comments in a place that Micky had, I thought unfairly, targeted for derision by using the headline "pussy shots and voyeur pics"....Micky pointed out that Artichoke had a lot of cat photos, mostly taken with electronic flash. Yeah, he does have a lot of cat snaps, shot with flash. I left a few commentaries demonstrating some understanding of photographic principles and practices underlying some "simple" cat photos. Here are the photos, and my comments.

http://www.pbase.com/artichoke/image/58434542
http://www.pbase.com/artichoke/image/27853499
http://www.pbase.com/artichoke/image/61592503

So yeah, Micky, I have to agree with you...the S3 Pro is not a perfect camera with regard to software compatibility. I have been complaining about the FujiFilm software disadvantage for quite some time,as have other observers and users of D-SLR's. I feel that FujiFilm has offered VERY,and I mean v-e-r-y poor software for image editing with the S2 and S3 cameras. Nikon software is much more mature than Fuji software,and Fuji's inability to save just a small list of changes and adjustments to the RAW camera files is a HUGE drawback compared to the way Nikon's .NEF files can be modified and adjusted and then just a short few lines of code can be added to the raw data so that the raw data is kept original and undisturbed.A 12.2- megapixel Nikon D2x raw file can be opened,adjusted,and dare I say perfected, and then saved as a .NEF file still remaining at its original size, of around 10.9 megabytes compressed. A 6-megapixel S3 Pro raw file can be opened,adjusted,and perfected, but it must be saved as an approximately 70-megabyte 16-bit TIFF file to preserve the changes made in post production because Fuji's software in UN-able to save back to the .RAF format!!!. This is a software issue that Fuji has not seen fit to correct. Nikon has,and has done it this way since 1999. D2x RAW files are 10.9 megabytes each in compressed NEF format, while S3 RAW files are 12.5 or 25 megabytes,depending on quality, but must be saved as HUGE, 70-megabyte 16-bit TIFF files if changes are made and the files optimized,which is a 70-megabyte to 10.9 megabyte disadvantage for Fuji.

Bottom line? Software is where it's at. Read last week's blog entry. Software is how we interface with our photographic images these days. I have long been critical of FujiFilm's poor sofware for its S-line of D-SLR bodies.So has Thom Hogan.Fuji's software leaves a lot to be desired, but it can make good TIFF files! Still, the FujiFilm D-SLR bodies are capable of making some very,very beautiful images. The lenses are still very important,as are the choices of shutter speed,aperture,ISO,and lighting. Hey....I use flash on a lot of my images indoors....often it's not in the EXIF because I'm triggering the flash from some other way than the hot shoe. Firing a low-powered stobe pop at 3 stops under ambient is a technique I'm well-acquainted with. Canon has better noise-avoidance and better noise-reduction capabilities than FujiFilm or Nikon. Okay....I'd agree....I'm impressed with the 20D's higher ISO performance. So the S3 pro's RAW output does not always process the best using ACR--now that is a SAD,sad state of affairs. But then, the same holds true for my D2x's files--at times, Nikon Capture works a bit better than ACR does, and at times, I will make the effort to process very,very difficult exposures thru Nikon Capture into 16-bit TIFF files, and then I will finish those off in PS CS-2. Such is life.

One of Micky's underlying contentions which I agree with wholeheartedly is that a camera which does not work really,really well with ACR's raw conversion routines is a hindrance and a disadvantage. FujiFilm ought to offer vastly better software,with workflow issues and storage problems SOLVED, once and for all. Nikon has more software features, but still needs better Macintosh support,and better speed on the Mac; Nikon's Capture NX promises much improved handling of files over the earlier versions of NC. I have personally felt the limitations of Fuji software before. And despite the fact that FujiFilm's software is located a little bit south of "nowhere" on the software map, the S3 Pro is capable of making some beautiful pictures,as long as it's used to its best advantage. I am not sure who Micky Rouse is; he's been on dPreview since June 24,2006,and has been on pBase about the same amount of time. He made some valid points about the S3 Pro and software and the S3's sensor performance, and yet he seems to have a score to settle as well. Kind of a weird situation really.

No comments: