With optical bench, do MTF values correspond to resolution lp/mm or lp/ph?

l_d_allan said:

I'm a fan of blog articles from Roger Cicala (LensRentals) such as "Just the Lenses" from their OLAF optical bench without sensors ... such as:

http://www.lensrentals.com/blog/201...4l-ii-vs-sigma-35mm-art-vs-canon-35mm-f1-4l-i

I do wonder if the MTF values on the vertical axis could be read as a lp/mm or lp/ph values. For example, does 0.8 correspond to some resolution figure, like 1000? If so, would that be consistent across lenses, since no sensor is involved with the OLAF optical bench?

Or not?

Click to expand...

Not.

They are MTF (modulation transfer function) plots, not resolution plots. The vertical axis is contrast at a specified spatial frequency.

0.8 corresponds to 80% contrast. The spatial frequency for each line is specified in the caption; from top to bottom 10, 20, 30, 40, and 50 lp/mm in the figures you linked.

I don't know the details of the procedure, and in particular whether test targets with sinusoidal or step intensity are used.

HTH

In the context of MTF charts, I posted a little bit about the meaning of the y-axis a while ago. Definition of contrast etc.


It does seem to cause some confusion to people that the word contrast is used in a variety of ways in photography. Most of my cameras have a "contrast" adjustment for instance, as do image editing programs. But to just keep it simple, the y-axis of an MTF chart is defined by the formula in the post I put up.

OpticsEngineer said:

In the context of MTF charts, I posted a little bit about the meaning of the y-axis a while ago. Definition of contrast etc.

http://www.dpreview.com/forums/post/54652410

It does seem to cause some confusion to people that the word contrast is used in a variety of ways in photography. Most of my cameras have a "contrast" adjustment for instance, as do image editing programs. But to just keep it simple, the y-axis of an MTF chart is defined by the formula in the post I put up.

Click to expand...

Maybe we should call it Michelson Contrast to differentiate it from your cameras' ;-)

l_d_allan said:

I'm a fan of blog articles from Roger Cicala (LensRentals) such as "Just the Lenses" from their OLAF optical bench without sensors ... such as:

http://www.lensrentals.com/blog/201...4l-ii-vs-sigma-35mm-art-vs-canon-35mm-f1-4l-i

I do wonder if the MTF values on the vertical axis could be read as a lp/mm or lp/ph values. For example, does 0.8 correspond to some resolution figure, like 1000? If so, would that be consistent across lenses, since no sensor is involved with the OLAF optical bench?

Or not?

Click to expand...

Each of the plots represents contrast at a given resolution on the sensor (lp/mm -- line pairs per millimeter) where:

1. Black = 10 lp/mm
2. Red = 20 lp/mm
3. Green = 30 lp/mm
4. Cyan = 40 lp/mm
5. Purple = 50 lp/mm

The x-axis is a bit confusing, however. It reads as "image height" and goes out to 20mm, but I would think it should be distance from the center, which would be a maximum of 12mm for a FF sensor. Perhaps the image height refers to twice the distance from the image center?

OpticsEngineer said:

In the context of MTF charts, I posted a little bit about the meaning of the y-axis a while ago. Definition of contrast etc.

http://www.dpreview.com/forums/post/54652410

Click to expand...

Thanks. From that thread, I came across what I felt were several well written articles from H. H. Nasse (Zeiss) on reading MTF curves:



Quite a bit of the first was over my head, and I anticipate tackling the second article in the near future. (makes me wonder what cln00 thru cln29 are about)

Great Bustard said:

The x-axis is a bit confusing, however. It reads as "image height" and goes out to 20mm, but I would think it should be distance from the center, which would be a maximum of 12mm for a FF sensor. Perhaps the image height refers to twice the distance from the image center?

Click to expand...

The article refers to the average of measurements at "4 different rotations".

My guess is that "image height" is distance from centre. 20 mm takes you close to the corners of a full frame field (21.6 mm half-diagonal for 24 x 36 mm).

Cheers,

alanr0 said:

Great Bustard said:

The x-axis is a bit confusing, however. It reads as "image height" and goes out to 20mm, but I would think it should be distance from the center, which would be a maximum of 12mm for a FF sensor. Perhaps the image height refers to twice the distance from the image center?

Click to expand...

The article refers to the average of measurements at "4 different rotations".

My guess is that "image height" is distance from centre. 20 mm takes you close to the corners of a full frame field (21.6 mm half-diagonal for 24 x 36 mm).

Click to expand...

Ah -- makes sense. Thanks for clearing that up!

Those are very good articles. A while ago I printed those out and went through them in detail on a nice long airline trip.

The only thing I felt was missing from them was the practical ascpect that the MTF of a lens changes with the distance between the camera as the object plane. I guess the author felt that was outside the topic he was covering. Roger Cicala mentions that a few times on his website, one time even saying he had considered buying a bill board for test wide angle lenses. Since he found in practice wide angle lenses that tested horribly on small test charts in a lab often turned out to be fantastic out in the field with landscapes.

OpticsEngineer said:

The only thing I felt was missing from them was the practical aspect that the MTF of a lens changes with the distance between the camera and the object plane. I guess the author felt that was outside the topic he was covering. Roger Cicala mentions that a few times on his website, one time even saying he had considered buying a bill board for test wide angle lenses. Since he found in practice wide angle lenses that tested horribly on small test charts in a lab often turned out to be fantastic out in the field with landscapes.

Click to expand...

Yes, I often wonder about that. I think we make an implicit assumption that non-macro lenses are designed to focus a perfectly flat incoming light wave front perpendicular to them, and that 15-30X focal length achieves that condition for all intents and purposes. Is that not the case, or is there something else at play?

Should that rule of thumb be based on angle of view instead?

"I think we make an implicit assumption that non-macro lenses are designed to focus a perfectly flat incoming light wave front perpendicular to them, and that 15-30X focal length achieves that condition for all intents and purposes. Is that not the case, or is there something else at play?"

When it comes to lens design, it is a pretty wide open field as to what the design goals can be. In general what you stated would be a pretty normal design condition. But when it comes to a specific lens, we often don't know what the goals the lens designer had. And then with a big company, marketing gets involved. Things happen like a lens is designed, but then marketing changes it mind and so the lens gets advertised as being good at things that really were not part of the optimization.

As an example of a lens that breaks ones expectation, we have the Sigma 18-200 OS HSM. Sharp at 18 mm. Sharp at 200 mm. Soft at 70 mm. Not at all what I expected when I bought it. It is the only lens I have bought that I just really cannot find a use for. (I should have read the dpreview review on it before purchasing it.)

Great Bustard said:

l_d_allan said:

I'm a fan of blog articles from Roger Cicala (LensRentals) such as "Just the Lenses" from their OLAF optical bench without sensors ... such as:

http://www.lensrentals.com/blog/201...4l-ii-vs-sigma-35mm-art-vs-canon-35mm-f1-4l-i

I do wonder if the MTF values on the vertical axis could be read as a lp/mm or lp/ph values. For example, does 0.8 correspond to some resolution figure, like 1000? If so, would that be consistent across lenses, since no sensor is involved with the OLAF optical bench?

Or not?

Click to expand...

Each of the plots represents contrast at a given resolution on the sensor (lp/mm -- line pairs per millimeter) where:

1. Black = 10 lp/mm
2. Red = 20 lp/mm
3. Green = 30 lp/mm
4. Cyan = 40 lp/mm
5. Purple = 50 lp/mm

The x-axis is a bit confusing, however. It reads as "image height" and goes out to 20mm, but I would think it should be distance from the center, which would be a maximum of 12mm for a FF sensor. Perhaps the image height refers to twice the distance from the image center?

Click to expand...

The maximum image height for a full-format detector is 21.64mm. This corresponds to the radial extent between the center of the frame and the absolute corner.

Image Height as an optical term simply describes the distance from the center of the image.