Like many old-timers, I have loved Leicas for decades. Michael's review had me itching for an M8. However... also like most old-timers, image stabilization is a real necessity. And, for everyone else, image stabilization would permit slower shutter speeds, enabling lower ISO, which would reduce or eliminate much of the noise of which Michael spoke.

IS is such a great tool for showing subject movement.

Oh, I agree with the desire for IS, but I think the Leica shooting style allows for very good shots without worrying about that.

I'm not squarely in the "I wish I owned one and all the lenses" camp, but I'd sure like to have a couple of weeks' worth of hands-on experience.

I'm also an olt timer, and in my film days used Leica M2, M3, and M4s. Most have found that they can hand hold an RF camera at much slower shutter speeds than an SLR because of no slapping mirror or auto diaphragm.
Michael may want to comment on this, as I noted on several of his M8 pics the shutter speed was aroun 1/45th sec.
Dave

I'm surprised that MR did not comment on the color finging easily seen on specular reflections and high-contrast dark-to-bright edges - I've seen it on every M8 shot posted on the Web. The problem is quite common in cheap digicams but unacceptable on a Leica. The problem becomes a non-issue if the shots are converted to B&W, but it would be shameful if the camera would only be capable of good B&W shots.
I see this as an M8 fatal flaw, most probably caused by the lack of an AA filter.

I've shot about 700 frames with the camera and haven't seen anything such as you describe.

Michael

Michael, I've seen the problem in some photos on dpreview.com. It specifically showed up on the edges of some tree trunks. However, the person posting clarified that they were working with a pre-production unit. That said, I wonder if you might have time to post a few additional color shots that show off Leica's ability to get great edges with color? I like the ones you've put out on the luminous landscape site, but would enjoy and benefit from seeing some more color shots in the gallery.

Thanks in advance if you are able to find the time to do that. Thanks for the consideration if not.

Rick

The first downloadable RAW shot on Outback Photo showed this fringing on chrome, but Uwe was using white plaster walls as his highlight area, and the fringing took place on chrome surfaces that must have been 4-5 stops overexposed. No camera could hold that, without doing something truly weird. (You can see the fringing on both cars at the bottom of the photograph, in the wheel spokes of the front wheels on the car on the right, and around the gas cap in the car on the left.) One of the members of the Leica Users Group used ACR to clean it up, and reposted the shot, showing no fringing, and estimated it took him about a minute to do it. Again, this was an extreme situation of severely overexposed chrome in bright direct sunlight, and was easily fixed with the most common post-processing software.

Sean Reid showed some mosaicing in a wicker basket, also easily fixed.

I can't wait to get the camera.

JC

I'd say it's a real nicety, but "necessity"? Not so much.. Many many many very fine low-light, hand-held photographs have been produced on Leicas, and I can't see how the move to digital would change that possibility. As others have mentioned, these cameras are very easy to hold steady. That said, I'd take it in a heartbeat if it were being offered

After reading the review, I think I might just get in line for one of these myself..

(BTW, I absolutely love the front-page image, Michael)

Color fringing is not only seen on that shot around chrome. It can be seen elsewhere, namely, the white painted graphics on the building's door glass. It will show up on on every specular reflection. It is due to the lack of an AA filter - a very unfortunate decision. Whoever hails designs without AA filters is dellusional and aims to deny basic physics. I'm sure this problem will be ovelooked as this camera is quickly becoming a fashionable must have toy.

I just had a look at the wonderful review of the M8. However, I don't exactly understand what Michael says about the 8/16 bits issue of the Kodak sensor.
The M8 has a Kodak KAF-10500 sensor.
The spec sheet of the KAF-10500 says that the linear dynamic range is 71.5Db, with a note indicating that 71.5 = 20 log(Vsat/VN)
This means that log(Vsat/VN)=3.575
In log2 form, we hav log2(Vsat/VN)=11.88
Thus the sensor seems to be a 12 bits sensor, much like the sensors usually found on Nikon and Canon DSLR.
Am I wrong somewhere?

Sorry but you are wrong in my opinion. I have used a Kodak 14nx (no AA filter) for over 3.5 years and now also use a Mamiya ZD (no AA filter as standard). You can remove any color fringing quite easily, but you can't recover detail lost through the use of an AA filter. I'm glad Leica realise this and have decided, rightly in my view, not to include an AA filter.

Quentin

James,

Per my limited understanding of these complex matters, DR has basically nothing to do with the bit depth.

1. DR can be expressed as a ratio in db, but it could also be expressed in decimal format as the Vsat/Vn ratio.

It just expresses the ratio between the brighest recordable illumination not resulting in blown highlights (saturated output current of the sensor), and the lowest illumination not producing a signal to noise ratio higher than a given value (impossible to distinguish between current resulting from actual image information and background current of the sensor).

It is usually expressed in log 2 form mostly because photographers are used to working with stops, and because a one stop increase corresponds to a doubling of the incoming light.

2. Completely independantly from this, you can decide to code the electric current coming out of the sensor photosites as a result of the incoming light any way you like.

Things in the digital world being discrete and based on binary arithmetics, the values of current are usually expressed as a binary number. The longer the number (bit depth), the more values you can use to describe the range of electric current correponding to the values between blown highlights and deep black, and the finer the division of the available range becomes.

You could very well decide to code using 16 bits a DR range corresponding to one stop only. Conversely, you could decide to only use 4 bits to express the same range.

Think of it as a ladder. The DR is how high the ladder is, and the bit depth is how many steps there are in the ladder.

Practically speaking, there seems to be little value in using very high bit depth to code a small DR, and using too few bits to express a large range would not do justice to the sensor's ability to output subtely different currents when small variation of illumination occur (especially when those changes of illumination occur in highlights or shadows).

Anyone else, please feel free to correct me.

Cheers,
Bernard

Yes, thank you, I understand that. I was just saying that with something like a 72.5Db (1:3758) as a ratio between Vsat and Vn, it was reasonable to encode it with 12 bits, like it is usually done. A good detailed explanation of what I wrote too scarcely is available there (look for "dynamic range"):
Fundamental etc...
Of course, Leica might have decided to encode the output of a 72Db CCD sensor with 8 bits, but then it's beyond my comprehension...

Cheers..

James,

My bad. I had misunderstood the nature of your comment. 12 bits would indeed be a reasonnable value.

Regards.
Bernard

Surely this isn't the main issue with the M8, the really big M8 problem is IR spillage into the shot when used in artificial light. See here for an example

http://homepage.mac.com/billh96007/PhotoAlbum196.html

and here,

http://www.leica-camera-user.com/digital-f...o-box-goes.html

Apparently the ultra thin IR protective glass in front of the sensor means, under some conditions, black subjects (see the tuxedos and dresses in the above shots) become a ghastly magenta. Leica have said there's a solution, but it involves 6 bit coding on the lens, a special IR/UV filter on each lens, and a firmware update due late November. I'm relieved there's a solution in the pipeline, but I'm not sure it'll work for me. For example I'd planned on getting the forthcoming Zeiss M mount 21mm f4.5 for a light-weight Leica M8 travel outfit. Obviously a Zeiss lens can't have Leica coding. I'm fairly optimistic this will all be resolved before my M8 arrives in January 2007, but if it isn't I guess I'll cancel my order.

The key word in your statement above is "artificial light" -- and more specifically, artificial light high in IR or "hot" light.

I spent the past two days pushing this camera very hard to try and make it puke.

The simple fact is I can only generate the purple-toned blacks if shooting under "hot" tungsten, halogen or strobe (xeon) lighting. It was *not* an issue under my CCT (cold cathode tube) continuous lighting nor under several types of indoor fluorescents I shot it under.

Moreover, I added an absorption-style IR cut filter (daylight version) and it fixed 99% of the IR bleed probelm -- to the point where I had to make some extreme image adjustments to find any IR hot spots in the shot. I did not test it, but I am convinced had I used the stronger tungsten IR cut filter it would have totally eliminated the problem.

Since *my* use for the camera is almost exclusively outdoor, I tried everything to make it fail under sunlight -- and couldn't. I shot directly into the sun at ISO 1250, sun through trees, sun refelcting of of multiple colors of automobile paint, chrome, aluminum, house paints, fabrics, you name it and I shot it. And I had absolutely zero issues with IR bleed.

My take on it is that if somebody is going to be using the M8 in a studio situation under direct hot lighting, then they probably have cause for concern. If they are primarily outdoor shooters, it is essentially a non-issue.

If they happen to shoot indoors under avaialable lightling it may be an issue especially if they are concerned about accurate color rendering -- however, even with film this was an issue and one needed to filter for it if they expected accurate colors. And street and PJ photographers rarely bothered filtering their film for color-correct mixed lighting.

Personally, I think for most applications the M is going to be used for, this will be a small issue -- and if it a big one on occasion, you have the filter fix available.

My .02 only,

Bernard,

I know your analogy of the ladder is also made by Bruce Fraser (I think he used a stair case), but I do not think it is correct. Current digital sensors use linear analog to digital converters and the potential dynamic range is limited by bit depth. Twelve bits can encode 4096 levels and 14 bits can encode 16,384 levels, giving DRs of 4096:1 and 16384:1 respectively. To demonstrate this, look at Norman Koren's chart and the text under the dynamic range heading.

http://www.normankoren.com/digital_tonality.html

A bit depth of 12 has a dynamic range of 9 zones (allowing for 8 levels in the darkest zone) and a bit depth of 14 gives 2 additional stops of DR. However, in practice the theoretical dynamic range is not reached as explained in this reference:

http://www.photomet.com/library_enc_dynamic.shtml

If the step size is fixed (with a geometric scale in this case), you need more steps to reach a given height. If a log AD converter were used, you could encode a greater DR with fewer bits and your analogy would be true.

Bill

Jack, that's very interesting. Would the list of offending illumination include a regular flash like Leica's own unit?

I didn't try a camera-mounted strobe, but I suspect it probably would also offend as most small strobes are xenon tubes too...

(And I admit to a bit of personal bias here, given my old-school M roots -- I'll even go on record saying that a real M shooter does NOT use flash! Handhold the thing at 1/8th if you have to and to heck with the little bit of blur! )

Seriously, I think bottom line is if you want to use strobe or shoot a lot under hot lights AND you don't plan on converting the images to B&W, you will probably want to use the stronger lens-mounted IR cut filter (or your DSLR) in those situations.

Cheers,

Jack,

I have been concerned enough about the magenta issue that I'm verging on returning my M8. Try as I might, however, like you, I cannot reproduce the magenta cast under my shooting conditions (outdoor natural light and indoor mixed tungsten/flourescent/daylight). I've shot all kinds of fabrics and I cannot reproduce the problem. I was still considering returning it, however, because I was assuming that I just hadn't hit the "magic combo" of causing the magenta cast and would run across it eventually. But, like you, I don't plan to use the M8 in the studio with strobes at all (and therefore didn't test under those conditions); it'll be a natural light/outdoor camera exclusively for me.

Has anyone else confirmed your experience that you only get the magenta cast when shooting with ""hot" tungsten, halogen or strobe (xeon) lighting?" If others are having the same experience as you and I seem to be having, then I'll probably keep the M8, as I don't plan to use it under those conditions.

On further thought: I'm not sure your conclusion is correct. I've seen pics shot outdoors in daylight that exhibit this problem. One example is attached. (I hope the author of that thread wouldn't mind me posting this, but otherwise, you'd have to be a member of the Leica Forum to see it).

William, there's some examples here that illustrates the problem under conditions that don't include flash, incidentally the photographer assures us that this is not a function where a high percentage of people happened to turn up in magenta suits and dresses!

http://homepage.mac.com/billh96007/PhotoAlbum196.html

Gary: Those were the original set of images that started all the fuss -- they were shot indoors under halogen (hot IR) lights. (If you read my comment above, I mentioned continuous hot light sources -- tungsten and halogen -- AND strobe.)

William: Is there some way to confirm it wasn't actually that color to begin with? Note that everything inside the window under store lighting looks pretty normal to me -- even the blue jeans on the woman, blue shirt on the man and the lights themselves. To buy the raincoat, we have to assume the photographer is remembering the coat accurately and by his own admission he doesn't remember what color it was -- only states it wasn't purple... While it is entirely possible, I personally don't place much credibility in his memory; it is the only "outdoor" shot Ive seen the problem crop up in if that coat was black or blue...

More to the point, the real color of the raincoat in that image (and most street type shots like it) is irrelevant anyway! What do we care if it really was black, blue or purple? It doesn't change the meaning of the image at all.

Most importantly, I would say that if *you* have not seen it with *your* camera in *your* shooting situations, then why worry? The camera produses phenomenal files...

My .02,

Jack,

That may be true, but that doesn't speak to the daylight pic I attached. Moreover, my concern, as an "available light" M shooter, is that I not be limited to a certain type of natural light (i.e., daylight). If I'm shooting in a cafe for personal work, or doing an editorial portrait in a client's office, I have no control over whether there will be halogen light. I'm not aiming this at you, just trying to talk it thru for my own purposes of deciding whether to return it.

William: I did address your earlier post -- see right after my response to Gary's that you quoted and right above yours

The problem is supposed to be a lot wider as per this post:

http://forums.dpreview.com/forums/read.asp...essage=20806987

Excerpts from it are:

It causes most black materials, whether synthetic and organic textiles to appear purple. The purple effect is caused by the transparency of many colored organic compounds to IR. You'll see it on black human hair, a black fur coat, black cat or dog fur, and black dyed fabric. You'll see it even if the fabric is naturally black, like black casmier. The only relatively common black pigment that does not suffer from IR contamination is carbon. This is used in some (but not all) black cosmetics, but not in fabrics.

IR contamination also causes many green materials (in particular, a rare and uncommon green plant called "grass") to appear yellow. It causes red flowers to lose saturation, and often acquire violet tints.

And it causes human skin to appear "blotchy". Areas with less subsurface blood appear unusually light, areas with increased subsurface blood appear as darker blotches. (Such irregularities in blood concentration can be due to circulation problems, or to quite normal physiological reactions that constrict or dialate capillaries). There's also a tendency for surface veins to be unusually prominent, especially in fair haired people.


Further it says:

Unfortunately, IR contamination is a "many to one" problem. You can't fix it with white balance: try to set WB to get rid of yellow IR grass, and green paint, green cosmetics, or a green dress suddenly turn blue. Try to WB out the magenta tint on the black dress, and black eyeliner turns green (and skin gets much less healthy looking).

Shouldn't this also adversely affect images around sunset? Maybe even the metering?

Jack, I think you're getting to the heart of the issue with this comment, and you're one of the few (only?) M8 users that I've read with practical experience of the effect of IR/UV filtration with the M8. Sean Reid suggested filtration is the solution, the post that "aaykay" linked to also suggested it, but you've actualy done it while they haven't.

Can you tell us more about these IR/UV filters as I'm sure I'm not alone in never having encountered them before. What do you mean by different strengths? How much are they? Are they widely available in different sizes? Is there a filter factor? Do they have any other negatives? And finally, is there any way you could post shots showing the practical effects of filtration?

Thanks.

Hi Gary:

Without going into great detail and sound like I'm lecturing I will summarize and skip technical accuracy for more easily understood plain English

There are two styles of IR cut filters, "hot-mirror" or "aperture grille" and "absorption". The hot-mirror actually have a metal oxide deposited on them that has slits that allows visible light to pass through but prevents the larger IR rays from passing. (Kind of like the door to a microwave oven, only for blocking light and not high energy radio waves.) The metal oxide gives them a mirror finish on the outside, hence the name. By varying the diameter of the slits in the oxide coating, any exact cut-off point of IR band can be implemented. Usually, this will be somewhere between 700 nanometers (blocks IR in tungsten light) and 770 nanometers (blocks the IR present in visible light).

The absorption units are colored a light cyan, but attenuate IR by altering thickness and density of color, not just density. And the final thickness is critical in determining the exact cut point, so manufacturing is more difficult than the hot mirror -- thus these absorption/attenuation type filters are about 2x the cost of the mirror units.

The problem with the hot mirror is that the slits effectively get narrower as light begins to enter the filter from the side -- basic trig. This in turn can cause a visible color shift as the slits narrow to the point where they start to cut visible light rays. (Now you know why certain medium format digital backs have issues with color shifts across the frame when using wideangle or shift lenses -- they have hot-mirror cut filters over the sensor instead of absorption type )

This problem is exacerbated for Leica in the M8 since the lens sits so much closer to the sensor than a comparable DSLR. Because of this, even normal lenses have too extreme an entry angle to use a hot mirror filter, so the absorption style must be used. However, the thicker the absorption style, the more refraction you will have at the image corners, and refraction directly over the sensor will impart a form of CA that is not lens related and is not easy to correct in post... So for Leica this was a design nightmare, basically having to walk the razor's edge and balance no CA and a bit of IR leaking over blocking all the IR and having CA in the corners of images.

In the end, it appears Leica chose the 770nm cut point as it deals effectively with IR present in visible light. However, I am of the opinion Leica will have to offer buyers this choice and offer to install the thicker 700nm cut filter as a sensor option -- photgrapher's, pick your poison...

To put price in perspective, a 3" x3" square absorption filter costs around $300. Smaller circular sizes are not much cheaper as they need to be cut and there is wastage. By contrast, circular hot mirror filters will be in the $100 range depending on size and manufacturer.

I'll try to post some example images later this morning.

Jack thanks for this, I've learnt something useful here. So these are the type of filters that would have to be fitted to the lens?

interesting indeed. So how hard is it to make the chip's microlenses behave as a kind of (additional) (absorption type?) IR filter?

Here is an example. Both shots using M8 AWB and shot with a single Profoto D4 head.

First, the image straight out of the camera:


Note that everything in the image is black in visible light -- THe backdrop is the back (dirty) side on my velveteen background cloth. The camera body and plastic lenshood stays black, but the black flocking inside the hood turns purple as does the black anodized aluminum RRS L bracket and parts of the all-black camera strap.

Here is the image shot with the Daylight absorption IR filter clipped on the lens:


It is not perfect -- there is some slight IR bleed visible in the hotter areas of the image (seen directly under the center of the strobes light cone and fading towards the outer edges), but IMO it is significantly improved over the one before it. I wish I had a Tungsten-cut IR absorption filter to test, as I am 99% certain it would eliminate the last remnants of IR bleed visible here, but I do not -- sorry.

Gary: Yes, this is a lens mounted filter.

Opgr: This is the issue I mentioned above -- it is not a microlens fix, but rather the choice of IR-cut filter/protective glass mounted *over* the sensor itself. This is a servicable part, though it requires dissassembly of the camera's sensor to replace it.

PS: I fogot to address filter factors...

The hot mirror style have a stated "0" factor, but I am dubious of that claim -- I feel it is only "0" for visible light energy. The absorption type documentation says it's variable depending on the light source. Since IR bleed adds significant energy to the image, it can affect overall exposure -- and usually does. However, depending on your meters spectral sensitivity, your camera may not register it.

I have seen, depending on the light-source, anywhere from 0 to 1-1/2 stop change using the absorption filter. By example, the filtered shot was 1-stop more than the non-filtered shot above it.

Cheers,

Jack, great answer.

You should post these shots more widely as I'm not aware of anyone else showing a practical, side-by-side demonstration of the efficay of lens filtration with an M8. After wading through all the normal internet bluff and bluster, posts like this remind me just how useful and informative forums can be.

Ah, yes, but I meant in an earlier design stage. As you mention, it apparently requires a careful balance between absorption and aberration. If somehow the microlenses on the chip itself could be designed to handle some of the absorption...

I agree, and very comprehensive as well...

Wow, that is quite a color cast in that image up there. Im kind of surprised that Leica would release a camera without dealing with that issue fully.

Is the need to blast these products out into the marketplace that strong? It's like a black hole, sucking products out of the hands of manufacturers before they can let the paint dry.

It seems to be a monster we've created alright. Just like the software we buy.

I used to work in software (and firmware) development and when we'd discover a new bug in a product that we swore was tested and bug-free, my favourite line was that it's a miracle our cars start every morning.

I hope they don't build airplanes and medical equipment this way.

Official "Leica Statement on M8 Issues" at DPR.

Nill
~~
www.toulme.net

And beyond IR spillage, Glass cover edge refraction effects (CA-like)... everybody is forgetting (for the moment) the aliasing artifacts (easily seen) due to the lack of the AA filter. Stay tuned.

I had great hopes for this camera.

Thanks Gary, thanks Opgr -- always good know one's inputs are appreciated

I've seen perfectly sharp M8 pictures shot handheld pictures at 1/8 second. Something about Leica!

Asher

No, that software is usually far more shoddy.

I can't mention specific software names for airplanes, but in some laser eye surgery equipment, you'll find ancient, un-patched versions of Windows 95 that crash often or give pop-ups that the operators automatically click away because they're so used to seeing the warning that they know doesn't mean anything (but what if there's another warning?). Other medical equipment may even be networked, and will likely run old, unpatched versions of their operating systems, because the medical software requires exactly that version and patchlevel of the OS to run properly and/or not to invalidate the service agreement.

And ... well, you really don't want to know.

But anyway, the problems we see with modern photographic equipment, are peanuts.

[QUOTE]I hope they don't build airplanes and medical equipment this way.


They don't [U]build airplanes this way. Trust me.

Two simple and recent examples (and if you are in the know, and we can trust you, I'm sure that you can cite dozens more):

Airbus A300 or A310, software problems in the synchronization of the five "independent" controlling computers led to a total system crash in certain conditions if only one of the computers crashed.

The December 2000 Osprey accident, a software glitch when resetting the system caused significant pitch and thrust changes in both rotors.

But this is far off-topic. My point is that you'll find shoddy programming anywhere, even in airplanes and medical equipment.