M8 vs Nikon D700

[mjh]

Originally Posted by pgk

How do Hasselblad deal with variation in individual lenses due to manufacturing tolerances - or is that not a good question to ask?

They don’t – and they don’t really have to. A good deal of robustness against manufacturing tolerances – that are rather small anyway – is a general requirement when designing lenses: you don’t want a lens that exhibits a very high degree of correction, but is a bear to calibrate as image quality deteriorates rapidly with only tiny changes due to manufacturing tolerances, changes of temperature etc..

[AlanG]

Originally Posted by diogenis

... abd I doubt a mathematician can describe easily a lens, much less a company who doesnt even design lens...

I believe that all lenses are based on mathematical models. The math can be so complex that some optical designers spent much of their careers doing the complex calculations required for just a few designs. Use of computers for lens calculations was a big breakthrough as it was a great time saver.

Additionally, new construction techniques made it possible to design lenses where individual elements or groups of elements can be relocated depending on the distance setting or focal length. The new Hassleblads use the AF motors to automatically compensate for known focus shifts caused by changing the aperture.

All of this technology is being applied to get the highest quality image possible from a given design, whether that is in a $35,000 camera or a cell phone.
You just have to look at the pictures to see it. (I posted an example.) What is wrong with that?

[mjh]

Originally Posted by SJP

What it will NOT work for is bokeh & depth of field.

Yes, since both depend on how rays of light propagate in 3D space. You would need a light-field camera to capture the 3D data necessary for calculating images with any desired focus setting and any desired depth of field. But existing light-field cameras are still severely limited with regard to resolution.

[mjh]

Originally Posted by diogenis

Yes, but don't diminish the fact they might have found a decent test procedure and create their parameters from this procedure. And by assigining a specific lens as a black box, trying to find what works best, will also take into consideration manufacturing tollerances. Anyway Im not an optician, abd I doubt a mathematician can describe easily a lens, much less a company who doesnt even design lens.

Determining the correction coefficients empirically is certainly much more work than deriving those values aided by the lens design software, but in principle, the results aren’t any less valid. And of course, the results arrived at mathematically need to be double-checked empirically with actual copies of the lens.

[ammitsboel]

mjh, just a personal question, do you program?

[mjh]

Originally Posted by ammitsboel

mjh, just a personal question, do you program?

I do, but how is this relevant?

[SJP]

Just to set things straight, you CAN calculate pretty much everything you need to know about a lens concerning all of the abberations that can be corrected in software like vignetting, cyan drift, CA, barrel & pincushion (contrast and sharpness to some extent). Calculating and optimising this is exactly what is done when the lens is designed. Once designed and built according to spec it will perform as expected. There is very little room for surprises but it is complex to get the calculations right and it has taken physics/engineering the best part of 300 years to get this far.

There is little or no benifit from measuring or calibrating a lens empirically unless the variations between lenses is unusually large.

[sdai]

Originally Posted by SJP

Just to set things straight, you CAN calculate pretty much everything you need to know about a lens concerning all of the abberations that can be corrected in software like vignetting, cyan drift, CA, barrel & pincushion (contrast and sharpness to some extent). Calculating and optimising this is exactly what is done when the lens is designed. Once designed and built according to spec it will perform as expected. There is very little room for surprises but it is complex to get the calculations right and it has taken physics/engineering the best part of 300 years to get this far.

There is little or no benifit from measuring or calibrating a lens empirically unless the variations between lenses is unusually large.

Not every ideal calculation can be transformed into product due to limitations in process, material, and cost of fabrication.

You can draw many nice things on paper, but not all of them can be turned into reality.

[marknorton]

Originally Posted by AlanG

Use of computers for lens calculations was a big breakthrough as it was a great time saver.

When I designed digital audio equipment, I used a 16 bit multiplier chip from TRW which would do more calculations in 1 mS than a human could do in their allotted 3 score and ten years - and that was 17 years ago.

[ammitsboel]

Originally Posted by mjh

I do, but how is this relevant?

I was just curious.
I think it's good to have a programmer on the board.

[SJP]

Originally Posted by sdai

Not every ideal calculation can be transformed into product due to limitations in process, material, and cost of fabrication.

You can draw many nice things on paper, but not all of them can be turned into reality.

They don't calculate the ideal lens, they calculate the best lens thay can build with given contraints (like no more than x ASPH surfaces, no more that y floating elements, these are the available ranges of glass types with known index and Abbe nr. etc. etc.). Once the computer design is done they know that they can build it and roughly how much it will cost.

Then they test the design in practice & start the fine tuning.

It is not black magic but it is a science, it still is an art and it still requires precision engineering.

[pierovitch]

Originally Posted by chris_livsey

Curious. What did they recommend ?
(I see CSI use Nikon I do not have sufficent product knowledge to identify and it's only TV )

Using the D3 for photography

Interesting that Nikons email disclaimer prohibits posting of its content.

[diogenis]

Originally Posted by AlanG

I believe that all lenses are based on mathematical models. The math can be so complex that some optical designers spent much of their careers doing the complex calculations required for just a few designs. Use of computers for lens calculations was a big breakthrough as it was a great time saver.

Additionally, new construction techniques made it possible to design lenses where individual elements or groups of elements can be relocated depending on the distance setting or focal length. The new Hassleblads use the AF motors to automatically compensate for known focus shifts caused by changing the aperture.

All of this technology is being applied to get the highest quality image possible from a given design, whether that is in a $35,000 camera or a cell phone.
You just have to look at the pictures to see it. (I posted an example.) What is wrong with that?

photos were locked and small, and I couldnt or didn't tried enough to spot differencies.
Anyway we are drifting away.
There are 2 philosophies here: one that accepts it doesnt matter to go for highest quality against affordability, since they can allways apply digital post processing work to correct for that, and the other is about assembling the best possible gear (expensive) that will lead in the best quality possible and all your options open, but less affordable. Whatever you choose, you get higher options to "fix" a photo when your gear is the best.

I dont mind if a photo has a couple of flaws, I am watching the big picture. And I am not convinced that it is so important to correct for mistakes when you can get your hands on a s/w tool that will do this for you automatically. The big picture is still there, so your photos won't affected that much.
On the other hand, I can understand that you, might work professionaly and that your clients might go crazy if they see an electricity cable runing in the skies, or a stain made from specs on the sensor.
What's more, I prefer to leave my photos untouched to preserve the original moment In contrast to what some digital tools can do these days. For that I am stuck with programs like C1 (which is also given free with the m8).
Here is an example:
Digital Film Tools Power Stroke
take a look at the Barbie, they tried to fix her lips/eyes. Well I don't like this type of manipulation in my photos. And I don't like the result at all! I prefer barbie with green eyes This is a plugin for Aperture 2 incidentally.

Here is another example:
We all have some way of listening to music. Well, I prefer to buy CDs from good sources, like say Telarc (try the 1812 overture, the original recording..a classic) and hear them flat to preserve fidelity. Some prefer to apply filtering to them, like bass and treble, or even more, like loudness. I don't like that. I just want to preserve fidelity as much as I can.


Back to the programs now:
We have C1 that does global raw processing and it goes free with your m8: I like this sort of manipulation, yet I am using it with caution. If I don't, I risk changing my original a lot.
Then you can get Aperture or Adobe's LR. I dunno about aperture, but in LR2 adobe is going for localized control -- even more control that is...
Bottom line is, you can do anything with those tools and the way things are going, one should also be a photoshop pro as well as a photographer if he wants pictures. Do we like that? Personally I don't

[AlanG]

Originally Posted by diogenis

photos were locked and small, and I couldnt or didn't tried enough to spot differencies...

My sample photos aren't locked and the full res versions may be downloaded. Others have downloaded them. They are available in hi-res so one can see that no detail is lost by applying the optical corrections. Even in the small samples, you should easily be able to spot the change in barrel distortion and vignetting when you toggle between the two images. I guess if you can't see it, then it won't matter to you. Here's the link in case you wish to try again.

PhotoShelter - Gallery Password Required
Password - Leica

I am not sure what you are driving at in your post. I don't think we were generalizing about manipulating photos in post. Specifically I thought we were discusing eliminating lens defects via software whether in post or in camera firmware. All lenses have some defects whether they are cheap or expensive and there is the reality or potential to improve their imaging characteristics via software.

How much adjustment, basic retouching, or other manipulation a person wants to do to a picture is a personal decision.

[pgk]

Originally Posted by mjh

They don’t – and they don’t really have to. A good deal of robustness against manufacturing tolerances – that are rather small anyway – is a general requirement when designing lenses: you don’t want a lens that exhibits a very high degree of correction, but is a bear to calibrate as image quality deteriorates rapidly with only tiny changes due to manufacturing tolerances, changes of temperature etc..

This is probably true for higher end lens manufacturers (such as Leica) but you only need to look at various photo fora to discover that there can be substantial sample variation in (not cheap) lenses!

Sounds to me as if the 'best' results will still be a combination of optical excellence (and if needed) software adjustments based on known parameters of optically excellent lenses whose characteristics can be plotted with a reasonable degree of certainty of them reflecting actual user's lenses. So what's new if you still need excellent lenses which have high QC to get the 'best' results.

I have a friend who is a lens designer and several years ago I discussed the DX format and its potential maximum MPixel limit (this was before I switched to FF). He suggested that the limit would probably be just a little more (~15) than that available from current DX cameras (12~13). His suggestion which he said was 'rough' was based on the potential cost and QC problems that he envisaged in building lenses capable of matching such sensors commercially. I'd say that the D700 (and its 'new' 'FX' format) may just be starting to show that lens costs (QC CAN be sorted) may be a limiting factor in format capacity.

[diogenis]

Yep, I got the pics.
And I see what you mean. DxO did a good job.
What lens did you used?
Vignetting is just barely noticed.

[AlanG]

Originally Posted by pgk

...but you only need to look at various photo fora to discover that there can be substantial sample variation in (not cheap) lenses!...

I am not a lens designer but I wouldn't be surprised if some aspects that can be corrected in software do not change very much due to sample variation. This may include vignetting, c/a, and distortion. DXO lets one fine tune these parameters and doesn't just leave you stuck with their default settings. I don't see why camera optical correction firmware couldn't have some kind of similar fine tuning on high end cameras. This would be similar to how some cameras allow AF to be fine tuned. Plus the lenses that people complain about may be outside the quality control limits and should be returned for adjustment or replacement. (A friend of mine had Canon adjust his 24 TSE and they made it sharper!)

[manolo]

* BACKWARD COMPATIBILITY WITH ALL M & R LENSES
(there is a beautiful history to be preserved)

* LARGER THAN FULL-FRAME SENSOR
(when possible use the same electronics for both M & R so that the upgrades are less difficult when one person has both systems) (there could even be an option to use a film back) (or a live view pack)

* MODULAR SYSTEM WITH DIFFERENT VIEW-FINDER OPTIONS
(for both M & R)

* THE MOST UNIVERSAL RAW FILE
(16 bit, red, blue, green, ir, uv) (the most information about the lens and settings so that it can be worked in the computer with the latest and any software available in the present and future)

this is the most upgradable sistem

* KEEP MAKING BEAUTIFUL LENSES

[stunsworth]

You've lost me. What's supposed to have backwards compatibility with all R and M lenses? A single camera or an R camera together with an M camera?

Don't you see an inconstency in asking for backward compatability with all lenses and have a larger than full frame sensor?

[manolo]

Originally Posted by stunsworth

Don't you see an inconstency in asking for backward compatability with all lenses and have a larger than full frame sensor?

if the sensor is larger than the lens you can crop the file any way you want.