Why Leica did what they did....

[adan]

Advertisement (gone after registration)

I just thought I'd post some reminders of what we DO get from Leica's engineering choices vis-as-vis other digital cameras - specifically the decisions to cut way back on the layers of filtration over the sensor...IR, AA, etc. - and to develop the offset microlenses to combat vignetting.

 

1) Leica M8 + Voigtlander 15mm compared to Canon superwides on full-frame bodies - corner image quality.

 

2) Vignetting with the 15mm on the RD-1 vs. M8 (even though the M8's sensor is larger and crops less)

 

3) Epson R-D1 reflections in the chip layers. I've been informed there is a small error in the Epson/Sony sensor schematic at the top - the IR filter is the top layer, not in the middle. But it still reflects light in any position.

 

The banding/blob issues are a bug, not an engineering choice. But I'll take the magenta tints (when they occur) and fix them with external filters or elsewhere, as needed, rather than reduce the M8's optical performance to "good enough".

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[aj37]

Excellent points, especially the info about the R-D 1's sensor reflections. I had seen similar effects, but had written them off as lens flares.

 

I'm repeating something I've just posted elsewhere, but suppose that from the very first announcement of the M8, Leica had introduced it as follows:

 

"The Leica M8 is optimized for black-and-white photography, with a sensor design emphasizing sharpness and texture rendering. Although the M8 is fully suitable for color photography, photographers may need to employ an infrared cutoff filter over the lens for critical color results."

 

I suspect that if they had introduced the M8 that way, Leica buffs would have been even more enthusiastic over the camera! They would have cheered Leica's commitment to b&w tradition, and reveled in the esoteric complexity of IR-cut filters. Pictures exhibiting magenta blacks would not have undermined confidence in the camera's quality -- instead, the photographer would have been blamed for failing to follow Leica's recommendation.

 

Although I'm sure the M8's teething problems have put some chinks in Leica's reputation, I think they can still recover if they emphasize that the IR situation was created because of their commitment to maximum image quality, not because the initial design was faulty.

 

(Of course that doesn't apply to the banding/ghosting issue, but defective components can happen to anyone -- just ask all the laptop computer makers whose Sony batteries have been going up in flames!)

[Asher Kelman]

I just thought I'd post some reminders of what we DO get from Leica's engineering choices vis-as-vis other digital cameras - specifically the decisions to cut way back on the layers of filtration over the sensor...IR, AA, etc. - and to develop the offset microlenses to combat vignetting.

 

1) Leica M8 + Voigtlander 15mm compared to Canon superwides on full-frame bodies - corner image quality.

 

Hi Adan,

 

While true, it's not a good comparison. While the defects of the Canon WA options are well known, what needs to be recognized is that some Canonites use the Zeiss or Leica WA lenses instead.

 

So compare the Leica M8 +VL 15mm with the 5D/1DII and the 21mm Distagon or 19mm Leica DSLR lenses.I wonder then how the pictures would then differ?

 

The Leica M8 appeal is more than WA lenses! That, small size and unobtrusiveness are winning combinations for landscape, street and travel. All this without having my B&W film inspected at the airports!

 

Maybe Leica did what they did because camera companies are run, not by enthusiastic Cartier-Bresson devotees but people with track records in finance, management, marketing and the like. They are used to packaging brands. They move boxes to store shelves!

 

In all organizations, the marriage between these necessities and the creative but obsessional designers is often one of suffering and joy in fluctuating proportions.

 

Asher Kelman

 

http://www.openphotographyforums.com

[adan]

Note: I'm not trying to bash Canon or Epson here. Anyone can go to dpreview.com and see how well the 5D and 1Ds(Mk.x) sensors do with longer glass. I'm just trying to show what a hard nut Leica had to crack with wide-angles on digital.

 

Here's a Canon prime wide on the 5D, in case using a zoom above seems unfair (I've heard regular Canon users debate the 17-35 vs. the 20 prime)

 

Leica had specific problems to solve with short-focus rangefinder lenses - especially wide-angles. I'd say they solved those problems superbly - at the cost of one new one.

 

Asher: Well, I hear ya', but the foofarah over the M8 seems to indicate that in fact the image engineers WERE running things, not the marketers and money people. I doubt any savings in the sensor filtration department were worth the recall and production freeze and free filters and lens discounts. Leica (intentionally or unintentionally) said "we will make the image sharp everywhere on the frame - regardless of the cost."

 

I can't speak to Zeiss/Leica lenses on a Canon body - the one time I tried I had to hold the lens in place without an adapter - and I won't blame EITHER Leica or Canon for the resulting mess.

 

I want an M-sized camera. That includes M-sized lenses, that sit 15-28mm from the image plane rather than 45mm (as do all SLR lenses). That has implications for imaging on silicon.

 

As to wide-angles - yep, the first lens I've bought in every camera system since 1978 has been an 18-20mm. That field of view earns more for me than most of my other focal lengths combined. So performance in that range is paramount - everything else will more or less fall into place.

 

If Leica can increase IR filtering internally without fouling WA performance on the M8 - more power to them. Otherwise, I will take external filters (or magenta issues) over any loss of sharpness and clarity.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[Guest stnami]

... still Leica could have started afresh... a digital rangefinder... but the chickened out

[adan]

... still Leica could have started afresh... a digital rangefinder... but the chickened out

 

I think Leica will evolve a new digital M system slowly. The M8 is designed with a lot of thought to the whole existing M line (except for 135 lenses, perhaps) and thus makes at least one compromise (magentas) to accomodate those lenses in other ways.

 

I think the newest lens designs, especially those introduced since the "digital M" became a commitment (16-18-21, 75 ASMA, 50 ASPH, 28 ASPH, perhaps the 28 'cron) probably all have "digital" design improvements.

 

By the time an M9 gets here, likely more of the lens line will have migrated towards digital friendly designs, so that the M9 can have, say, a stronger internal IR filter without affecting the resolution of the new lenses. Or a larger sensor. Or both.

 

I don't know that Leica "chickened out" - they just figured they would be better off getting to the same end without ticking off an "installed customer base" completely.

[j. borger]

Advertisement (gone after registration)

Finaly a good thread about all the M8 "issues" ....... thanks for putting it all in the right perspective.

I do not think a Digital rangefinder from scratch would have been a good idea .. it's all about the glass .. old and new .. ... that's one of the charms of a rangefinder: to have a very broad choice of lenses.

New digital only lenses .. .... if they ever considered it they where right about chickening out ... they sure would not have gotten away with this .. especially not if you see all the comotion around the current M8.

[jaapv]

Thank you for reminding us that there is no such thing as a free meal. When are you starting your new job with Leica as PR manager in charge of the introduction of new products? Seriously - it should have been Leica to point this out, not a member of the customer base. As it is, the choices Leica made have put a photographic tool in my hands that I can only describe as magnificent, even at this point of time.

[Joseph S. Wisniewski]

I just thought I'd post some reminders of what we DO get from Leica's engineering choices vis-as-vis other digital cameras - specifically the decisions to cut way back on the layers of filtration over the sensor...IR, AA, etc. - and to develop the offset microlenses to combat vignetting.

I'll grant you the wide angle quality of symmetrical (or at least "less retrofocus") rangefinder lenses, and the advantages of "my" offset microlenses.

3) Epson R-D1 reflections in the chip layers. I've been informed there is a small error in the Epson/Sony sensor schematic at the top - the IR filter is the top layer, not in the middle. But it still reflects light in any position.

 

The banding/blob issues are a bug, not an engineering choice. But I'll take the magenta tints (when they occur) and fix them with external filters or elsewhere, as needed, rather than reduce the M8's optical performance to "good enough".

The Epson pictures you post do not indicate the problem that you claim, a reflection between the layers of the filter stack in front of the sensor.

 

Your flare increases in size with each repetition. A flare that increases in size without inverting shows a net convexity, either a reflection two mildly convex elements, a convex element and a flat one, or a convex element and a concave element with a less severe curve. That flare comes from either within the lens, or from between the front element and a front filter. Since Leica's decision will mandate filters in many situations where we would not normally have used them, it is a cause of this type of flare, not a cure.

 

As an aside, the correct sequence of filters is one cemented package with the first layer (closest to the lens) having a blue-green dyed glass filter that performs a combination of IR blocking and red level reduction (or the red channel would blow too easily), then two layers of LiNbO3 AA filter, then typically one very thin layer of clear glass. The glass and LiNbO3 layers have similar indexes of refraction and are cemented, so you do not get the reflections between the layers that you have drawn. The front and back glass layers have anti-reflective coatings.

 

There is an air gap between this filter pack and the sensor cover glass, and the sensor cover glass is also coated.

 

I have never seen a real shooting situation in which reflections between sensor layers created a flare that was noticeable. Such flares are masked by the much more radical flares between lens elements (such as the one you have shown) or flares between the rear element and the first glass surface (whether it's the sensor cover glass as in Leica or the filter stack cover glass as in Epson).

 

(my spell checker says "indexes" is OK, and "indices" isn't!!!!!)

[Joseph S. Wisniewski]

Excellent points, especially the info about the R-D 1's sensor reflections. I had seen similar effects, but had written them off as lens flares.

You were correct in your initial assessment. Lens flare has repeating images that increase in size as they become more distant from the source of the flare, or decrease in size for some repetitions, then invert, and increase in size for successive repetitions. A flare from layers of parallel glass would maintain a constant size in precisely the way that Adan's examples don't.

I'm repeating something I've just posted elsewhere, but suppose that from the very first announcement of the M8, Leica had introduced it as follows:

 

"The Leica M8 is optimized for black-and-white photography, with a sensor design emphasizing sharpness and texture rendering. Although the M8 is fully suitable for color photography, photographers may need to employ an infrared cutoff filter over the lens for critical color results."

But this would not be true. Increased IR sensitivity "sabotages" black and white photographs in much the same way it does to color photographs.

 

See this post on the five effects of infrared contamination.

 

http://forums.dpreview.com/forums/read.asp?forum=1038&message=20838611

 

Here's how those five effects will affect a B&W image.

 

1) Infrared sensitivity won't make the blacks turn "magenta" in a B&W photo, but it will lighten them, causing shadow detail to have less clean blacks.

 

2) It will cause surface veins to be more apparent in skin, enhance stubble that would normally be invisible below the skin surface in male subjects, and create a general "blotchy" appearance in skin.

 

3) Whether B&W or color, there will still be "an increased tendency to see underwear through clothing, grafitte that has been painted over, blemishes that you thought were covered by makeup."

 

4) There will be an increase in flare, ghosting, and a general decrease in contrast.

 

5) I'll quote this one in its entirety: "Sharpness issues: Lenses are only color corrected across the visible spectrum. This is why you need to compensate your focus for infrared photography. Since infrared contamination is causing the image to be a composite of both infrared and visible light, both the visible and IR can't be in sharp focus at the same time. You lose resolution, or 'microcontrast'."

I suspect that if they had introduced the M8 that way, Leica buffs would have been even more enthusiastic over the camera! They would have cheered Leica's commitment to b&w tradition, and reveled in the esoteric complexity of IR-cut filters. Pictures exhibiting magenta blacks would not have undermined confidence in the camera's quality -- instead, the photographer would have been blamed for failing to follow Leica's recommendation.

If Leica had wished to demonstrate a commitment to B&W, they could have released the camera as a B&W camera, no color filters on the sensor. This would have caused a two stop increase in the visible light sensitivity without increasing the IR sensitivity, for a net result of a two stop decrease in IR contamination, with all it's effects (loss of contrast and sharpness, blotchy skin, visible veins, etc) on B&W. And they'd have a camera that went to ISO 10,000 instead of 2,500.

Although I'm sure the M8's teething problems have put some chinks in Leica's reputation, I think they can still recover if they emphasize that the IR situation was created because of their commitment to maximum image quality, not because the initial design was faulty.

They can't "emphasize" that, because it simply isn't true.

(Of course that doesn't apply to the banding/ghosting issue, but defective components can happen to anyone -- just ask all the laptop computer makers whose Sony batteries have been going up in flames!)

Or all the Canon, Nikon, Casio, etc. P&S cameras that have a Sony sensor with large numbers of field failures due to a chemical contamination problem during manufacturer.

 

But my personal opinion is that the banding is a circuit design issue, not a defective component issue.

[marknorton]

But my personal opinion is that the banding is a circuit design issue, not a defective component issue.

 

I agree, there's a consistency across the cameras out there which suggests a design problem. That said, 2 - 3 weeks to get a fix into production is not long, so the fix is probably quite simple but unfortunately requires the camera to be opened up. Sadly, not everything can be fixed in firmware though you can be sure they looked to see if this problem could be.

[albertwang]

Finaly a good thread about all the M8 "issues" ....... thanks for putting it all in the right perspective.

I do not think a Digital rangefinder from scratch would have been a good idea .. it's all about the glass .. old and new .. ... that's one of the charms of a rangefinder: to have a very broad choice of lenses.

New digital only lenses .. .... if they ever considered it they where right about chickening out ... they sure would not have gotten away with this .. especially not if you see all the comotion around the current M8.

 

Wrong. I firmly believe that a new digital rangefinder would have been ideal. In fact, having to be pulled merely by tradition isn't the best approach. Maybe Zeiss will come up with something better?

 

I theoretically think the M8 is a fabulous design but the technology hasn't gotten to the point where we can be satisfied with it for color photography.

[georg]

"16-18-21, 75 ASMA, 50 ASPH, 28 ASPH, perhaps the 28 'cron) probably all have "digital" design improvements"

 

What are "digital design improvements"? It's a marketing trick - my 35Asph is over 10 years old and is better than any other 35mm-lens of any system - with film and digital, just because it is a good lens.

The short distance between film/sensor and lens is not a design flaw or a disadvantage of the M-system, it's even one of it's biggest advantages! What should they change in a "new digital rangefinder"? Increasing the distance like in the 4/3-system were the lenses are of less quality but four times as big?

 

You have to make compromises, the ideal digital systems doesn't exist, digital technology has a long way to go. But right now, the M8 seems to be the best compromise for many purposes (high-quality, small body/lenses) while other systems like Canon are better for sports-photography... and paying a high-price for these abilities.

 

There isn't the "perfect camera" - you have to find the best solution for your purposes!

 

The bad edge-quality and CA maybe hasn't too much to do with the sensor-filters, sometimes these lenses are really that bad (my 26 years old 19mm Leitz wasn't great either).

 

Here is a little fun-comparison I've made between a 7000€ and a 2000€-system stopped down two full stops, no WA, just a "simple" normal focal lengh! Of course Leica is better (otherwise it would be horrible) but many people bashing Leica or the M8 (while never used them) sometimes compare the M8 with consumer-cameras where you have pay just for the red dot... The second one compares the far edges with the 35Asph at f1,4 and f2,8.

A fully usable digital camera with real megapixels at f1,4 over the whole frame - hard to find, the M8+Summilux Asph is one of those.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[adan]

Joe: 'The Epson pictures you post do not indicate the problem that you claim, a reflection between the layers of the filter stack in front of the sensor"

 

I beg to differ. Light (except for lasers) spreads with distance, even in the absence of a convex surface. Every bounce the light takes extends its travel distance and it spreads a bit more, thus each successive reflection is larger (and weaker, as the light is attenuated by distance and spread)

 

This effect a) does not appear on film, and does not appear with the Leica/Kodak sensor. It appears more or less identically with a Leica 21mm lens and the Cosina 15mm lens - two very different optical designs.

 

If you can explain how two very different lenses can create an identical flare pattern - and only on the Sony sensor, but not on the Leica/Kodak sensor or on the 10 or so varieties of film I've used with these lenses, and without this effect - please go right ahead.

 

I'l let you and Vivek Iyer fight over the layer stacking in the R-D1's sensor - he claims he removed his IR filter and it was the top layer - but that's between you two.

 

(BTW - You might want to tone down the preachy, "smarter-than-thou" attitude - a 50% track record (while not bad) doesn't support it).

 

Georg: yes, on a 1.33-crop sensor with less-than-ideal infrared filtration, the Summilux, and my Voigtlander, and a host of other M-mount lenses perform exquisitely.

 

But the tradeoff is that one must use external IR filters. And live with a crop that turns the 'lux into a $3500 "normal" lens.

[jaapv]

Wrong. I firmly believe that a new digital rangefinder would have been ideal. In fact, having to be pulled merely by tradition isn't the best approach. Maybe Zeiss will come up with something better?

 

I theoretically think the M8 is a fabulous design but the technology hasn't gotten to the point where we can be satisfied with it for color photography.

 

I seriously doubt that you have even tried it.

[georg]

@adan

At least you can use the image area that is given. What is 24x36mm worth when you cannot use them?

I'm working with my 35Asph since two years, using it with Technical Pan and Imagelink HQ - it has excellent perfomance for 24x36mm!

That's one of the few things I don't like about the M8: knowing that you don't use the full poencial of your expensive lenses, they will really shine with a 18MPixel-24x36mm-M9!

But even today the quality is comepletely different than with mass-production-lenses.

Ok, your image angle shifts a little bit, you need to keep that in mind when buying lenses (I'm thinking about a 24mm) but on the other side my 90AA becomes a fabulous 2/120Apo!

I've made some DNG-shots at a presentation with the new Tri. 21-28mm high-quality digital-images with a <1kg-combination! To beat that you need an Alpa with a MF-back!

 

With the correct profiles in 90% of the situations you won't need IR-filters. But so what? They're highest quality B&W (Schneider-Kreuznach with glas from Schott) for free (at least two of them) while the problems with mediocre lenses and extremly expensive 24x36mm-DSLRs cannot be solved that easy.

[Joseph S. Wisniewski]

Joe: 'The Epson pictures you post do not indicate the problem that you claim, a reflection between the layers of the filter stack in front of the sensor"

 

I beg to differ. Light (except for lasers) spreads with distance, even in the absence of a convex surface.

Only perfectly collimated light, whether or not it's from a laser, does not spread with distance. Lasers are merely more collimated than more common light sources, they're not perfectly collimated.

 

That is not the issue. The issue is the distance between the reflecting surfaces and the angle at which the non-collimated light is spreading. The distances between the layers of a multiple layer filter pack are too small for the light to spread as much as it did in your examples.

Every bounce the light takes extends its travel distance and it spreads a bit more, thus each successive reflection is larger (and weaker, as the light is attenuated by distance and spread)

 

This effect a) does not appear on film, and does not appear with the Leica/Kodak sensor.

I have seen "this effect" on film, many times. And your Leica and Epson examples were taken months apart under radically different conditions.

It appears more or less identically with a Leica 21mm lens and the Cosina 15mm lens - two very different optical designs.

 

If you can explain how two very different lenses can create an identical flare pattern - and only on the Sony sensor, but not on the Leica/Kodak sensor or on the 10 or so varieties of film I've used with these lenses, and without this effect - please go right ahead.

As I said, different conditions.

I'l let you and Vivek Iyer fight over the layer stacking in the R-D1's sensor - he claims he removed his IR filter and it was the top layer - but that's between you two.

Vivek Iyer and I don't fight. We go way back. He removed the entire cemented pack.

(BTW - You might want to tone down the preachy, "smarter-than-thou" attitude - a 50% track record (while not bad) doesn't support it).

First, as far a toning down a preachy attitude, have you heard the one about people living in glass houses?

 

Second, my track record is way past 50%.

[gumshoecamus]

as usual andy makes sensible and informed observations with empirical evidence.

a new-design digital rangefinder might be in the future, but with leica's financial situation, small (but loyal) customer base and the huge r+d costs, it would have been a collosal risk. (maybe the tentative moves into 4/3 system could be an experiment.) leica's only possible step was to develop a digital m that supports most of the m lenses out there. ignoring digital would be to choose obscurity and eventual collapse.

I appreciate some people are disappointed, but it seems to me that -- the faults in the first cameras aside -- the IR/full-frame sharpness compromise is the right move. I don't like the idea of adding an IR filter to my lenses, but I would be even more unhappy with r-d1-like problems with corners, vignetting, internal reflections, etc (although I love my r-d1).

I have read some posters suspecting a finance- or marketing-led consipracy at leica, where the design was delibertately botched to save money and clever marketing led customers astray. that is utter nonsense. leica is about the most engineering-led company I can think of (a quick glance at the woeful balance sheet confims that). in all its pro products there seem to be no compromises made for cost (or even practicality), the marketing is inept and usually poorly translated german.

I admire many poster's demands for perfection -- leica's perfectionist customers ensure its survival -- I am a perfectionist too, but now and again I just accept the best-possible solution and enjoy it.

dave

[Guest stnami]

..most engineers prefer to create something new.... not a fix-it job. A new camera still would have made more sense, as for the lenses there are plenty of film photographers out there willing to buy and salivating at the prospect of owning a Leica lens. Sure there may be a temporary drop in price, then nostalgia hits and the pockets empty willingly.( Leica sells new lenses)

Olympus went for broke, sure yet to really pay off but it is a genuine attempt to look after the OM customers. With the advent of small DSLRs they will probably get there. Their new top lenses are pricey but superb and are waiting for the cameras to catch up and take advantage of the optics.

 

p.s. with the speed of changes to the M8 (2 weeks or so), is really an indicator that Leica were never transparent in their dealings with their pre-order customers. Then again as with most things the customer expects too much from the product and the company expects too much money

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[farnz]

..most engineers prefer to create something new.... not a fix-it job.

 

I disagree. I am a Professional Design Engineer who has worked with a large number of Engineers over many years. I have no doubt that Engineers prefer to improve on a sound design rather than "create something new" for the following reasons:

1. A proven design avoids the "bathtub curve" where, statistically, a high incidence of faults occur in (give or take) the first 5% and the last 10% of product lifetime.

2. The nature of the design is well understood because it has already performed.

3. Early-life bugs have been identified and designed out.

4. It has a known history which can be extrapolated to help identify the optimum path for improvement and paths to avoid.

5. Creating from new typically requires extensive research to prevent accidental plagiarism or infringement of other's patent rights.

 

Pete.