Pages: 1 ... 5 6 [7] 8   Go Down

Author Topic: Medium format redefined  (Read 70987 times)

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8913
Re: Medium format redefined
« Reply #120 on: September 15, 2015, 02:33:16 pm »

If you have a very good lens and sensor, and a quantitative testing method, you may be able to observe the focus peak. I saw this with my Rolleiflex Schneider lenses using imatest and an 8ft chart shooting tethered.  I was shooting at 3-4 meters and could notice a jump in sharpness at the peak of focus. I don't think I would have seen this with normal focusing, but I had my camera mounted on a geared rail and was using the rail to tune the focus following the numbers on Imatest to see the max.  Taking steps smaller than 1mm you can see this jump in sharpness right at the peak - move 2 mm steps and you might miss it. This was observed with several lenses and with different aperture settings.

It would be interesting to graph the sharpness over distance to see how the sharpness rolls off from the focus peak in either direction.  I don't have time for this now, but I think it would be interesting to compare the roll off curve between formats.

Hi Eric,

I also did similar tests (see attachment), using a Stackshot rail at some 3.34 metres distance on a slanted edge target, to find the best/highest focus/resolution possible at various apertures, as a prelude for deconvolution sharpening. In my findings I did not observe a (sharp) peak, but a gradual optimum (which is what I expected because defocus blur does not change focus abruptly, but rather more gradual).

Quote
My hunch is that the rate is faster with larger formats.

I don't see why it would. After all, all that happens when we focus, we make the focus plane intersect the exit pupil's cone of rays at slightly different positions. That cone of rays has a constant acceptance angle for each focus point/sensel, so equal steps will show a gradual increase/decrease. When the same focal length and aperture are used, then the size of the sensors makes no difference, the cones are basically the same.

Cheers,
Bart
Logged
== If you do what you did, you'll get what you got. ==

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: Medium format redefined
« Reply #121 on: September 15, 2015, 04:30:41 pm »

Hi Bart,

My experience is more a bit like your diagram. I see some maximum in sharpness, but it is more at the 2cm level at 4.0 meters than on the mm level.

Regarding "standard sharpening" in Imatest, it may be a red herring. I wrote Norman Koren about it, asking if it was an indication for good or adequate shrapening, an he cautioned about it.

My experience is that I can apply a large amount of sharpening at small radius and still have undersharpening according to Imatest.

To understand Imatest "standard sharpeinung" it may be helpful the read the relevant information at Imatst: http://www.imatest.com/docs/sharpening/

My take on the issue is that Imatest assumes a sharpening that will result in MTF 100% at 0.3 x Nyquist frequency. When we sharpen at actual pixels we tend to sharpen with a large amount at small radius. This will enhance "micro contrast". But, that "micro contrast" will normally contribute very little to perceived sharpness in a print, as medium frequency detail will dominate.

If I push pixel level sharpening in Lightroom and analyse in Imatest I will always end up with under sharpening. On the other hand I can try some more balanced sharpening like the "high frequency/landscape" presets in Lightroom and apply some sharpening say 15% at radius 2 in Photshop unsharp mask, and that will be pretty good sharpening according to Imatest.

Using "landscape" sharpening and say 15% unsharp mask at radius = 2 will give excellent SQF values in Imatest.

Focus magic gives pretty optimal results in Imatest. No MTF above 1.0 in 0-0.3 Nyquist, MTF near 1.0 at 0.3 Nyquist and a decent level of MTF at Nyquist. Also, FocusMagic doesn't enhance noise as much as many other methods of sharpening.

As a side note, the "Orange Peel" artefacts "Diglloyd" has found may be a sharpening artefact, possibly combined with aliasing. What I may have seen is that sharpening using FocusMagick will not cause some artefacts which may be the foretold "orange peel" artefacts.

So, I would say that I really appreciate FocusMagic, it is one of the best tools I have found for sharpening. But, it doesn't fit my parametric workflow. Would I do a large print, I would probably deploy FocusMagic and some other tools, but for 90% of my work I try to get around with the tools provided with LR 6.

Now, getting back to MFD, what I have seen is that there are no magick differences between my MFD gear and my Sony Alpha 99 gear. The great advantage of MFD I see is resolution, but that advantage takes like A1-size prints to show up, unless I view my images with a loupe.

I am pretty sure that those 80MP backs, paired with excellent lenses, have a resolution advantage over present day high MP 135, even if the best lenses are used. With lesser backs, say 40-50 MP I am a bit skeptical.

One interesting observation is that a frquent poster here in LuLa, Chris Barret, shifted over his work from IQ-260 (I think) to Sony A7r. He uses it with Canon T&S lenses and Hasselblad CFi glass, pretty much what I have. He feels that the old Hasselblad/Zeiss glass works very well on the A7r.

Rainer Viertlboeck, a well known German architecture photographer is also quite happy with the A7r: http://forum.luminous-landscape.com/index.php?topic=90358.0

So, it seems that real commercial photographers see real benefits with high MP 135.

Personally, I have been shooting my A7rII for just three weeks so I don't feel I can make a lot of comments.

Best regards
Erik


Hi Eric,

I also did similar tests (see attachment), using a Stackshot rail at some 3.34 metres distance on a slanted edge target, to find the best/highest focus/resolution possible at various apertures, as a prelude for deconvolution sharpening. In my findings I did not observe a (sharp) peak, but a gradual optimum (which is what I expected because defocus blur does not change focus abruptly, but rather more gradual).

I don't see why it would. After all, all that happens when we focus, we make the focus plane intersect the exit pupil's cone of rays at slightly different positions. That cone of rays has a constant acceptance angle for each focus point/sensel, so equal steps will show a gradual increase/decrease. When the same focal length and aperture are used, then the size of the sensors makes no difference, the cones are basically the same.

Cheers,
Bart
Logged
Erik Kaffehr
 

landscapephoto

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 623
Re: Medium format redefined
« Reply #122 on: September 15, 2015, 04:36:46 pm »

Quote
Just a warning: the equations you are playing with ignore all optical aberrations of the lenses (unless you are using ray tracing software). These aberrations are essential to characterise how "unsharp" the out of focus areas of a given lens are.
you can use the same MF back with the same lens and just consider a crop from it matching the area of FF sensor... that shall be better.

By changing the sensor size, you will change the apparent depth of field on prints or on screen, so: no, you can't do that.
Logged

Theodoros

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2454
Re: Medium format redefined
« Reply #123 on: September 15, 2015, 05:33:21 pm »



By changing the sensor size, you will change the apparent depth of field on prints or on screen, so: no, you can't do that.

Same reason why shooting on the same image area (like Fransisco did) and then crop the image is wrong... It's like having used an APS-c sensor on one of the images (the 50mm lens)... Like if he has used a D800 for the MF lens and a D7000 for the 35mm lens...
Logged

fdisilvestro

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1852
    • Frank Disilvestro
Re: Medium format redefined
« Reply #124 on: September 15, 2015, 06:21:08 pm »

Try this:

Take any image from your MF DB an print it A1or A2. Now cut a square at the center of the pint 10x10 cm.
Hold the A1 (or A2) print next to the small square you cut before (in other words, look at them from the same distance). Does the complete print has the same DOF than the small square or not?

Theodoros

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2454
Re: Medium format redefined
« Reply #125 on: September 15, 2015, 07:08:18 pm »

Try this:

Take any image from your MF DB an print it A1or A2. Now cut a square at the center of the pint 10x10 cm.
Hold the A1 (or A2) print next to the small square you cut before (in other words, look at them from the same distance). Does the complete print has the same DOF than the small square or not?
Off course it does Francisco... but that's not my point! My point is that the result you got is the same as shooting a 75mm (FF lens) on an APS-c camera...

EDIT: ...Since later you magnified it (and thus reduced the DOF to the magnification of the respective focal length - i.e. from 50mm to 75mm of a FF lens...)
« Last Edit: September 15, 2015, 07:21:14 pm by Theodoros »
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Re: Medium format redefined
« Reply #126 on: September 15, 2015, 07:21:58 pm »

Try this:

Take any image from your MF DB an print it A1or A2. Now cut a square at the center of the pint 10x10 cm.
Hold the A1 (or A2) print next to the small square you cut before (in other words, look at them from the same distance). Does the complete print has the same DOF than the small square or not?

I do not think this is a very relevant comparison.  "Depth of field" is perceptual measure that depends on a variety of factors including viewing distance (not to mention the visual acuity of the viewer etc.), and in general people do not view prints of very different sizes from the same distance.


About the best one can do is:

(a) First, give a measure of DOF under somewhat arbitrary "reference" conditions, including
-- the focal length f
-- the aperture ratio N (or one could instead use the effective aperture diameter, A = f/N)
-- the "effective format", meaning the dimensions of the part of image circle at the focal plane that is used to produce the displayed image (so that crops are format changes)
-- the viewing distance
--  some measure of the viewer's visual acuity; maybe a proxy like "circle of confusion diameter no more than 1/3000th of the viewing distance, as I believe is used in many traditional DOF scales).

(b) Then to deal with other situations, describe the changes in DOF caused by changing any of those conditions, in relative terms.
For example:
-- Doubling the focal length at equal aperture ratio or halving the aperture ratio at equal focal length (either of which doubles the effective aperture diameter) approximately halves the DOF.
-- Doubling the viewing distance or halving the displayed image size doubles the DOF.
-- Halving the (linear) size of the image (as with cropping) and still displaying the new image at the same size as the original (so doubling the degree of enlargement) halves the DOF.
-- Aging enough to half one's visual acuity doubles the DOF of all your photos!
Logged

fdisilvestro

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1852
    • Frank Disilvestro
Re: Medium format redefined
« Reply #127 on: September 15, 2015, 07:32:13 pm »

Off course it does Francisco... but that's not my point! My point is that the result you got is the same as shooting a 75mm (FF lens) on an APS-c camera...

EDIT: ...Since later you magnified it (and thus reduced the DOF to the magnification of the respective focal length - i.e. from 50mm to 75mm of a FF lens...)

I'm using an arbitrary area of the sensor  for the 80mm which has a diagonal 1.6x compared to the sensor area used for the 50mm (same aspect ratio), then for the output I enlarged the smaller area (from the 50mm) 1.6x compared to the area used for the 80mm and compared them side to side.
Any object from the scene will look the same in the output.

Yes, you can use a 75mm on FF and a 50 mm on APS-C and will get a valid comparison

Theodoros

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2454
Re: Medium format redefined
« Reply #128 on: September 15, 2015, 07:35:22 pm »

I'm using an arbitrary area of the sensor  for the 80mm which has a diagonal 1.6x compared to the sensor area used for the 50mm (same aspect ratio), then for the output I enlarged the smaller area (from the 50mm) 1.6x compared to the area used for the 80mm and compared them side to side.
Any object from the scene will look the same in the output.

Yes, you can use a 75mm on FF and a 50 mm on APS-C and will get a valid comparison

It's not the same magnification factor Fransisco... you treat the MF lens like being an FF lens...
Logged

Theodoros

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2454
Re: Medium format redefined
« Reply #129 on: September 15, 2015, 07:38:08 pm »

I do not think this is a very relevant comparison.  "Depth of field" is perceptual measure that depends on a variety of factors including viewing distance (not to mention the visual acuity of the viewer etc.), and in general people do not view prints of very different sizes from the same distance.


About the best one can do is:
.......-- the "effective format", meaning the dimensions of the part of image circle at the focal plane that is used to produce the displayed image (so that crops are format changes)........


That's right!
Logged

fdisilvestro

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1852
    • Frank Disilvestro
Re: Medium format redefined
« Reply #130 on: September 15, 2015, 07:51:08 pm »

Let my try again:

Assumption: Scene is at a distance > 1m.

Take an image with a FF DSLR (35mm) with a 50mm lens and another with a MF DB with an 80mm lens.

Then:
- take a crop of 10 x 15 mm out of the FF DSLR image (diagonal 18.02776mm)
- take a crop of 16 x 24 mm out of the MF DB image (diagonal 28.8441 mm)

Make a print of 20 x 30 cm out of both crops which means:

- the 10 x 15 mm from the FF DSLR will be enlarged 20x and the image from the MF DB will be enlarged 12.5X

Look at them side to side. Do they show different field of view? Do objects in the original scene look the same size in the prints or are they different?

Theodoros

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2454
Re: Medium format redefined
« Reply #131 on: September 15, 2015, 08:05:46 pm »

Let my try again:

Assumption: Scene is at a distance > 1m.

Take an image with a FF DSLR (35mm) with a 50mm lens and another with a MF DB with an 80mm lens.

Then:
- take a crop of 10 x 15 mm out of the FF DSLR image (diagonal 18.02776mm)
- take a crop of 16 x 24 mm out of the MF DB image (diagonal 28.8441 mm)

Make a print of 20 x 30 cm out of both crops which means:

- the 10 x 15 mm from the FF DSLR will be enlarged 20x and the image from the MF DB will be enlarged 12.5X

Look at them side to side. Do they show different field of view? Do objects in the original scene look the same size in the prints or are they different?

You should have used the scale on the diagonal Fransisco... use lenses of the same focal length and blow (instead of cropping the FF lens) the MF lens result by 4x... (because the image area that the lens projects is 4x).... Otherwise, the magnification factor is less than it should be...

EDIT: The magnification factor you got out of your method was only of (about) 2.5x (1.6 ^2)....
« Last Edit: September 15, 2015, 08:18:07 pm by Theodoros »
Logged

fdisilvestro

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1852
    • Frank Disilvestro
Re: Medium format redefined
« Reply #132 on: September 15, 2015, 08:17:31 pm »

You should have used the scale on the diagonal Fransisco... use lenses of the same focal length and blow (instead of cropping the FF lens) the MF lens result by 4x... (because the image area that the lens projects is 4x).... Otherwise, the magnification factor is less than it should be...

The scale of the diagonal is the same 1.6X, image area has nothing to do

Theodoros

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2454
Re: Medium format redefined
« Reply #133 on: September 15, 2015, 08:26:05 pm »

The scale of the diagonal is the same 1.6X, image area has nothing to do

Obviously you posted after (before you see) my editing... That's exactly why I'm saying that you've treated the MF lens like being a FF lens...  But if you was using both lenses at their full format (to have the same AOV on the diagonal), the MF lens with your method is already cropped...

The suggestion for the diagonal is only so that accuracy is maximized (since the same AOV for comparison can only be to the diagonal as formats are different)...

EDIT: You are now (with your method) missing the difference of the magnification factor (and have some small difference because of not using the scale on the diagonal) by what the difference of (about) 4x is with (about) 2.5x....
« Last Edit: September 15, 2015, 08:34:27 pm by Theodoros »
Logged

fdisilvestro

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1852
    • Frank Disilvestro
Re: Medium format redefined
« Reply #134 on: September 15, 2015, 08:55:19 pm »

Obviously you posted after (before you see) my editing... That's exactly why I'm saying that you've treated the MF lens like being a FF lens...  But if you was using both lenses at their full format (to have the same AOV on the diagonal), the MF lens with your method is already cropped...

The suggestion for the diagonal is only so that accuracy is maximized (since the same AOV for comparison can only be to the diagonal as formats are different)...

EDIT: You are now (with your method) missing the difference of the magnification factor (and have some small difference because of not using the scale on the diagonal) by what the difference of (about) 4x is with (about) 2.5x....

A 50mm on 24x36 mm sensor will have the same angle of view than an 80 mm on a 38.4 x 57.6 mm sensor. We have to keep the same aspect ratio otherwise it will be comparing apples to oranges

80mm is 50mm X 1.6
38.4 is 24 X 1.6
57.6 is 36 X 1.6

The diagonal of 24X36 is 43.2662mm
The diagonal of 38.4X57.6 is 69.22658
69.22658 is 43.26662 X 1.6

The area of 24X36 is 864 mm2
The area of 38.4X57.6 is 2211.84 mm2
The difference between 2211.84 and 864 is 2.56 which is the same as 1.6^2

The difference in the diagonals of the experiment I proposed is 1.6

There are no small differences in my calculations. I use as many decimal digits I need to maintain accuracy

Where do you get 4X from?

synn

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1235
    • My fine art portfolio
Re: Medium format redefined
« Reply #135 on: September 16, 2015, 04:12:05 am »

My reaction every time I see a new post in this thread:

Logged
my portfolio: www.sandeepmurali.com

Theodoros

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2454
Re: Medium format redefined
« Reply #136 on: September 16, 2015, 04:26:12 am »

A 50mm on 24x36 mm sensor will have the same angle of view than an 80 mm on a 38.4 x 57.6 mm sensor. We have to keep the same aspect ratio otherwise it will be comparing apples to oranges

80mm is 50mm X 1.6
38.4 is 24 X 1.6
57.6 is 36 X 1.6

The diagonal of 24X36 is 43.2662mm
The diagonal of 38.4X57.6 is 69.22658
69.22658 is 43.26662 X 1.6

The area of 24X36 is 864 mm2
The area of 38.4X57.6 is 2211.84 mm2
The difference between 2211.84 and 864 is 2.56 which is the same as 1.6^2

The difference in the diagonals of the experiment I proposed is 1.6

There are no small differences in my calculations. I use as many decimal digits I need to maintain accuracy

Where do you get 4X from?

56x56 is close to four times the area of 24x36.... If you shoot on the diagonal, you don't have to keep the aspect ratio because the diagonal is (supposed to be) a diameter of the image circle no matter what the aspect ratio is.

EDIT: The projected area of the 6x6 is exactly 3.63 larger (not 2.56x) and the diagonal is 1.9x (not 1.6x)... If you insist to stick with the 3:2 ratio (which you don't have to because the angle of the diameter of the image circle doesn't matter) you'll have to choose a 3:2 rectangle with a 1.9x larger diagonal (by using a Pythagoras theorem) which will end up to the same result... What you now do, is treating the MF lens like if it was of a smaller format that projects an image circle of (about) 30% smaller area image circle (and thus different COC size).
« Last Edit: September 16, 2015, 05:19:33 am by Theodoros »
Logged

fdisilvestro

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1852
    • Frank Disilvestro
Re: Medium format redefined
« Reply #137 on: September 16, 2015, 06:00:28 am »

The projected area of the 6x6 is exactly 3.63 larger (not 2.56x) and the diagonal is 1.9x (not 1.6x)... If you insist to stick with the 3:2 ratio (which you don't have to because the angle of the diameter of the image circle doesn't matter) you'll have to choose a 3:2 rectangle with a 1.9x larger diagonal (by using a Pythagoras theorem) which will end up to the same result... What you now do, is treating the MF lens like if it was of a smaller format that projects an image circle of (about) 30% smaller area image circle (and thus different COC size).

I have no problem in using a different ratio (btw, the diagonal of 56x56 is 79.196, which is 1.83X the diagonal of 24x36 or 43.267), but the essential issue is:

- You are convinced and will not change your mind that the image circle is related to CoC
- I am convinced and will not change my mind that the image circle does not have anything to do with CoC

As a result of this, I better follow Synn advice and move on.

Have a nice day 

Jager

  • Full Member
  • ***
  • Offline Offline
  • Posts: 207
    • E vestigio
Re: Medium format redefined
« Reply #138 on: September 16, 2015, 06:20:21 am »

My reaction every time I see a new post in this thread:



+1

landscapephoto

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 623
Re: Medium format redefined
« Reply #139 on: September 16, 2015, 06:24:52 am »

Try this:

Take any image from your MF DB an print it A1or A2. Now cut a square at the center of the pint 10x10 cm.
Hold the A1 (or A2) print next to the small square you cut before (in other words, look at them from the same distance). Does the complete print has the same DOF than the small square or not?

The classical formulas for depth of field suppose that smaller prints are viewed at a shorter distance than larger prints. So, for this particular test, you will need to look at the crop at close distance and compare it to the whole print hung further away. You will then find out that they do not have the same DOF.

Probably, I should also disappear in a bush as Homer Simpson in the animation, but I've never liked that series anyway.
Logged
Pages: 1 ... 5 6 [7] 8   Go Up