The Man From Mandrem said:
In photozone.de They talk about center, edge and extreme edge in the MTF resolution tests. Was considering a lens where MTF resolution deteriorates rapidly (drops by 1/2) at edge, extreme edge as you go to wider aperture. Trying to figure out what locations that corresponds to in the image (Couldn't find where they explain test).

The have a "Lens Test FAQ" where they explain their test and also the locations. In the top menu go to "Lens Reviews" and fourth entry from the bottom is the Lens Test FAQ.

I had a quick tangential question. Searching forum the only closer thread was closed.

In photozone.de They talk about center, edge and extreme edge in the MTF resolution tests. Was considering a lens where MTF resolution deteriorates rapidly (drops by 1/2) at edge, extreme edge as you go to wider aperture. Trying to figure out what locations that corresponds to in the image (Couldn't find where they explain test).

When the discussion comes up with something useful can you interpret this in a manner an old person can understand? It seems to me, in spite of all the tests, the final analysis will be in what comes out as the final product. A result of all factors. A mix of lens and sensor resolution, shooting technique, and post processing. So, is there an answer to the question of lens vs. sensor resolution? I suppose it is about the idea of the weaker link.... the photo will be at the resolution of the image in its least sharp mode, either coming through the lens or as it is received post sensor by the buffer. Ah well..

LP/mm = lines pair per mm

Correlli said:
As interesting as I find the discussion about resolution and wether sensor out resolve lenses or vice versa, to me this is a discussion about optics, not photography.

I have to agree with your statement, my original reason for posting was questioning/pointing out that the techniques listed in Nikon's tech note were as important to D7000 camera as they are to the D800.

I took the use of live view to mean that you should use manual focus and the magnified live view to get the sharpest possible focus. Although my eyesight is no longer 20/20, I have never considered the auto focus as accurate as manual focus with live view magnified.

Most importantly, as paperman noted, live view eliminates the vibration induced by the mirror flip.

Mike,
I did go back to the tech guide to read between the lines to see if it was actually saying the D800 will focus better & sharper in LV mode in the context that traditional way of AF via viewfinder was not perfect because it was missing the focus ( and LV wasn't - due to different detecting systems )

One understands it is not the case ( thank God ).It is simply because the LV offers shooting without mirror slap ( also possible with non LV focusing ) and that there is the huge advantage of zooming into the object and then focusing.

Nothing to worry about, in other words :-))

Mike Gunter said:
At issue is the nature of losing the viewfinder to the Live View in order to facilitate a better Phase Detection focus system in the camera

It's the other way around; with Live View, phase detection is turned off and the camera switches to contrast detection.

If you 'have to use "Live View'" to focus in order to get sharp images, you need to tell your users up front that before they pay their money.

You don't "have to" use Live View to get (the sharpest) images. It depends on the situation.

Contrast detect does have advantages for "static" situations. You can place the focus point virtually anywhere, you don't have back/front focusing issues, and you can "zoom in" closely to check focus (especially with the help of a loupe).

Hi Paperman,

I'm not down on Live View, but on rather that using it something more than it is to get a sharper image.

What it is, is finding focus, not aiding focus, that's a huge sight different. What I'm driving at is that you are using a different tool with which to focus when you switch to Live View, and that matters - a lot.

If you 'have to use "Live View'" to focus in order to get sharp images, you need to tell your users up front that before they pay their money. I'm not sure Nikon is doing that right now, nor am I sure they want to sent that message out, but I'm sure that is exactly the message that is in the Tech Note I read.

My best,

Mike

Mike Gunter said:
The problem I have is the somewhat glossed over nature of switching to "Live View".

Been using the D300 for 3 years now I think. Probably clocked over 60,000 clicks.

And I must admit, I don't know how to use the Live View properly ! !. Yesterday, I decided to fool my baby daughter by shooting while not looking thru the VF ( she stops smiling when she loses eye contact ).

Couldn't manage a single decent shot. When is mirror up, when is it down.? Why does it "clonk" and not take a picture sometimes and sometimes it does ? Why does it feel like I have taken 2-3 shots when I have taken only one ! Drove me crazy; made me feel like a fool. I know the mirror has to get out of the way but it's like the whole world is shaking when it does that.

I have attempted to try it many times during long exposures just for the sake of having tried it ( they say you can focus better when you zoom in LV ). Too bad it doesn't work simultaneously with the self timer on the D300 - the only times I can find good use of.

I do know how to operate the Live View in video mode though - on my wife's D5000 . That's easy :-)

It looks like I am not the only one who will never get decent use of the LV feature.

Hi all,

"leave the pixel peeping to others - until someone gives me a D800 as a present :))"

I like presents, too. ;-)

As for pixel peeping, I hear that a lot, but really don't know what that means.

The PDF that dehidding referred to has some good sense shooting tips that - and I think that this is extremely important - largely don't limit themselves to the D800/D800E.

There is nothing there that doesn't translate into good shoot technique for DSLRs.

The problem I have is the somewhat glossed over nature of switching to "Live View".

At issue is the nature of losing the viewfinder to the Live View in order to facilitate a better Phase Detection focus system in the camera, or better said, a better "interface" for the user to place a target location of the focus points on the sensor to get that focus.

IMHO, that stinks. I think it can be engineered better.

My best,

Mike

@dehidding: thanks for the info. I did some more research at photozone. Their values are lw/ph. So getting 4000 lw/ph on a camera with a vertical resolution of 4000 pixel reads to me like a very, very good value.

It is also interesting what they say about the lw/ph in their FAQ: "You shouldn't take the LW/PH value itself too seriously because it is dependent on a number of factors. The analyzing tool (Imatest) is quite vulnerable to the quality of the source material."
And: "All PZ sample images are taken as RAW files and converted via Photoshop ACR (default settings without automatic image correction and contrast set to 0). If you convert RAWs via other imaging applications the LW/PH figures will be lower or higher due to the different sharpening & contrast algorithms."

As interesting as I find the discussion about resolution and wether sensor out resolve lenses or vice versa, to me this is a discussion about optics, not photography. I will stay with my old AI-S lenses and leave the pixel peeping to others - until someone gives me a D800 as a present :))

TaoTeJared said:
Ade I'm just shaking my head. Once you start with saying "lpmm" is not the same...
Add to that both teams were using ISO certified charts.

What is it that's hard to understand?

Some charts have markings to 2000 lpmm. Others have much finer markings to 4000 lpmm. Obviously, 2000 is not equal to 4000. And yet other charts have no lpmm markings at all. Not to mention in the digital world, lpmm is a less meaningful metric anyway due to differing sensor sizes.

And how does one know that "both teams were using ISO certified charts"? ISO does not certify any of these charts! You can print your own ISO chart using an inkjet printer. Is it certified? No. Can you compare its numerical lppm results to an ISO chart printed on photographic paper? No. Is one more more valid than the other? No, they are just different. Neither are "certified", and "lppm" results will NOT come out "the same" between them.

Not to mention that ISO target designs are relatively old, and we have better designs now for computer analysis. Imatest recommends SFRplus charts for use with their software.

I've seen the Photozone test chart and it doesn't look like an "ISO chart" to me (and no lppm markings, anyway):

I've explained how factors such as shooting distance and lighting will affect the results of these tests (see Imatest's test procedure for example.) I've also mentioned that both sites are likely using two different metrics (lines vs. line pairs).

Ignore any info as you wish.

I suspect that someone, somewhere, has mixed abbreviations. The standard resolution lens tests output a number that is specified as line widths per page or line pairs per page. The two are not the same, line widths are individual lines and line pairs are a count of each pair of black/white lines. The "per page" refers to the page height of the test chart - 200mm.

The standard resolution test requires one to focus on the ISO target so tha the target (200mm high) fills the camera view. Obviously different lenses will force the camera to be different distances away from the chart. With the chart filling the view, the tester has "equalized" different focal length lenses. A series of photos are taken adjusting the focus. Examination of the photos determines resolvable lines per page.

If we take the 200mm page height and divide it by 4000 lines per page, we get a actual resolvable line width of .05mm (when the entire photo target/subject has a height of 200mm).

No you read the number correctly. That is my point of the tests not being the same reference point but the end conclusions of "this lens is sharpest at F/x.x" or one lens is sharper than another.

Note my previous post as well (below) - There is no such thing as a 1 to 1 where light on a sensor pixel = pixel in picture. All the pixels in a picture are calculations based on multiple sensor pixels with (and probably a very large and complex) algorithm that estimates the color that should be represented in the photo.

Point being - it doesn't matter what light hits what sensor pixel - it is all software estimations. Software has just as much to do with resolution as the lens and sensor.

TaoTeJared said:
What people keep forgetting is that each "pixel" on a sensor is only records Red, Green, or Blue. Software then takes those readings compares them to adjacent and regional cells to create a color out of the full gamut of colors for each pixel that is then exported. All a sensor really does is record a value of one color, not the actual image as film did.

TaoTeJared said:
They also have numbers for the D3x that where the Zeiss 25mm goes almost to 4,000 lpmm!

Hi Tao, I am absolutely sure that the Zeiss lens is fantastic, but either I read the unit wrong or something else is not correct. I read lpmm as lines per millimeter (please correct if wrong). So 4000 lines per millimeter would mean a line width of 0,25 µm or 250 nm (that is the wave length of UV-C light!). With a pixel size of 5,94 µm for the D3x that would mean that they get roughly 24 lines on one pixel. They would not be able to measure this resolution with the D3x... Or did I misunderstand something here?

I just came across a website that actually posts crops of lens tests and allows one to compare two lenses just by "mousing over". You can actually compare the 85mm/1.4 against the 85mm/1.8
Check out: http://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx

The also have a help page describing how they run the ISO-12233 test, which is informative: http://www.the-digital-picture.com/Help/ISO-12233.aspx

The site is oriented more towards Canon than Nikon, but they do have good number of the Nikkor lenses tested.

Ade I'm just shaking my head. Once you start with saying "lpmm" is not the same...
Add to that both teams were using ISO certified charts. The topic is the difference between the two disused sites, not to argue just to argue.

spraynpray said:
Well, my take on this is that Nikons worst lens designer forgot more about lens designs TODAY than I'll ever know in my lifetime. I believe that Nikon would not make such a fundamental mistake as to put in sensors that out-perform their current range of lenses. We are not talking about a 'bloke in a shed' making this stuff, as the sensors come from Sony and are sold to several manufacturers, we are saying they are all stoopid enough to have not thought of that before spending millions on R&D etc? Unlikely.

Sorry guys - just my common sense speaking, no offense meant.

I agree. Nikon's current MF 50mm 1.2 design is how old? 25years? And it still outperforms most lenses let alone what else has been designed since then.

What people keep forgetting is that each "pixel" on a sensor is only records Red, Green, or Blue. Software then takes those readings compares them to adjacent and regional cells to create a color out of the full gamut of colors for each pixel that is then exported. All a sensor really does is record a value of one color, not the actual image as film did.

Well, my take on this is that Nikons worst lens designer forgot more about lens designs TODAY than I'll ever know in my lifetime. I believe that Nikon would not make such a fundamental mistake as to put in sensors that out-perform their current range of lenses. We are not talking about a 'bloke in a shed' making this stuff, as the sensors come from Sony and are sold to several manufacturers, we are saying they are all stoopid enough to have not thought of that before spending millions on R&D etc? Unlikely.

Old lenses that may have been knocked about etc. are excluded from the above.

I think we are all getting a little to cleverer than we have the knowledge to be.

Sorry guys - just my common sense speaking, no offense meant.

TaoTeJared said:
Test chart type, shooting distance & lighting setup (for this instance) doesn't matter much if any for resolution tests. Lpmm are the same on every chart and how the light falls on it doesn't change it.

Sorry but the above is incorrect.

First, "lpmm" is not the same on every chart. Some have different scales, while newer charts tend not have any line count markings at all, only geometric shapes (such as slanted edges useful for algorithmic analysis).

Second, even assuming the same type of chart (e.g., an ISO 12233 test chart), if printed on different type of papers they may produce different numerical results.

Third, most tools (including Imatest) work by measuring sharpness, expressed as MTF. Sharpness directly relates to contrast, so anything which affects contrast -- including lighting setup and shooting distance -- will affect the results. Heck even the choice of RAW processing software used during the test may affect the results.

"What is considered a large noticeable difference in everyday shooting. 30, 50, 200, 500 lines?"

Now we can see that partly this question is impossible to answer since -- as we've been discussing -- the numerical values from different tests are not strictly comparable.

Another factor to consider is the final print size. A lens may seem sharp when its images are printed at 5x7, but might look a little soft when enlarged to 13x19 assuming normal viewing distances. So two lenses differing by a certain lpmm may appear at the same subjective sharpness at a certain print size but not at much larger sizes.

[Photozone and Lensrental] obviously use two different standards to represent resolution.

Looking at the numbers, my guess is one is using lines (LW) while the other uses line pairs (LP).

Ade Barkah said:
Both Photozone and Lensrental use Imatest.

I saw that in the comments section but couldn't find anything on their site about it. I may just be missing something.

Ade Barkah said:
Lens testing numbers are not comparable unless all (other) parts of the test system are the same (camera model, test chart type, shooting distance, lighting setup, etc.)

Isn't that what I said? Test chart type, shooting distance & lighting setup (for this instance) doesn't matter much if any for resolution tests. Lpmm are the same on every chart and how the light falls on it doesn't change it. Other tests it certainly does though.

I'm just laughing at how people think they need to sell their 85mm D for the 85mm G because of difference of "89" of whatever that number means. Stupidity resides in ignorant impulses.

TaoTeJared said:
Photozone's software tools cost around $3,000. I'm not sure what Lensrental uses but it looks like DxO which is only a few hundred bucks.

Both Photozone and Lensrental use Imatest. Lens testing numbers are not comparable unless all (other) parts of the test system are the same (camera model, test chart type, shooting distance, lighting setup, etc.)

They also have numbers for the D3x that where the Zeiss 25mm goes almost to 4,000 lpmm!

Here is an example of the difference between the two sites for the Nikon 85mm f/1.8 D.
(Not easy to find the same lens tested at both.)

Lensrental: 1053 / 942@ f/8
Photozone.de APC D200: 2188 @5.6
Photozone.de APC D7000: 2716 @5.6
Photozone.de FX D3x: 3779 @5.6

I am in no way by pointing this out that I believe that one is right and one is wrong or to bring into question either of their testing methods. They obviously use two different standards to represent resolution. Photozone's software tools cost around $3,000. I'm not sure what Lensrental uses but it looks like DxO which is only a few hundred bucks.

This where focusing on only a number is very wrong to do. That may baffle some how numbers can be different, but that is the wrong focus since at the end of the day what that exact number means is actually nothing.

The key is for a tester to pick the testing solution and to stick to one for all their lens tests. Once we know that, all we need to take notes on are the lenses that fall outside the boundaries (one being really low) and to note the F-stops where the lens is performing at it's optimum. As you can see between the two sites, the optimum F-stop was basically the same (lensrental gives only one but Photozone, F8 was almost the same as 5.6) even if the numbers are vastly different.

The LensRentals tests: You can also see (from both sites,) Zeiss out resolves Nikon glass but not by much. From the "best" to the "worst" lenses tested is damn near nothing. I think it just shows that all of Nikon's glass is good to go on the D800.

What I never like about these tests is that they never give the reader the Perspective they need to approach the tests with. As in, What is considered a large noticeable difference in everyday shooting. 30, 50, 200, 500 lines? They don't say. Test the old kit 28-85 lens or even the old 24-120vr (always known as soft) and put that up to see the difference. If they test close to the group, then smaller difference in numbers matter. If they are 100-200+ lpms lower, then you know that the difference between 100 lines is nothing.

Tao Te,

Photozone.de has numbers for both the D200 and D7000. I think D200 numbers max out around 2000 and the D7000 numbers max out several hundred above that.

There are different ways to test lenses and how they calculate the "end" number. My guess it is just a difference reference point/ technique. I have seen other testers using end results in the 1,000s and others with results in the 2,000s so there must be a couple of standard programs out there - probably many more. Note that Photozone.de uses a D200 10mp which is 3x smaller in # of pixels. Usually the more pixels, the more resolution and a slight increase in results as smaller pixels can resolve more lines.