Ad
Ad
Ad
Pages: [1]   Bottom of Page
Print
Author Topic: High magnification of ground glass  (Read 2383 times)
torger
Sr. Member
****
Online Online

Posts: 1400


« on: May 28, 2012, 01:56:35 PM »
ReplyReply

The attached image shows 100% crops of an experiment when a Canon 5Dmk2 with a 1:1 macro lens was focused on the Linhof Techno standard ground glass, and then live view was used as a very high magnification lupe.

I then focused on the 130 cm mark on the 45 degree slanted folding rule 3.5 meters away, and then the 33 megapixel digital back was slided into position (Linhof standard sliding back) and a picture was taken on the wide open aperture f/4.5 of the 90mm lens.

After studying the resulting image the actual focus seems to be closer to the 138cm mark (note that there's some dirt on the glass around the 140cm mark blurring it a bit). I don't think I can see any focus shift though, the focusing error is due to that the depth of field is fairly wide and ground glass resolution is limited, so it is hard to see on the ground glass where the exact focus is. This error (about 5.6 cm depth distance) corresponds to about 40 um on the focus rail. The gearing on the Techno is good so I had no troubles to set the focus where the eyes told me to despite the very small movements.

Note the quite high resolution of the ground glass. With some effort each individual dash on the mm scale on the ruler can be seen, but due to variying grain size the exact resolution varies a little. The ground glass resolution is roughly 50 lpmm, or corresponding to about 18 megapixels in the 48x36mm area.

After this experiment it is clear to me that with a 10x lupe one is noway near maximizing the potential of the ground glass. Using the 10x lupe I can see that there are numbers on the ruler but not actually read them. I'm playing with the idea to get a high quality 25x pen microscope to use in situations when really exact focusing is needed, not sure how much light loss there is in those though.

Has anyone played around with >10x magnification or has any other comments / tips / tricks?
Logged
amolitor
Sr. Member
****
Offline Offline

Posts: 801


WWW
« Reply #1 on: May 30, 2012, 03:31:14 PM »
ReplyReply

It looks like you're making movements of 40 micrometers to refine the focus here, if I am understanding your post correctly?

Do you think that when the sensor is placed into the plane formerly occupied by the ground glass, that it's within 40 micrometers of the plane the ground glass was in? I have no sense of how accurately built that component is, and how reproducible all this business is. Obviously one COULD build equipment that placed the sensor plane at the same plane as the glass to whatever accuracy you like, the question is, does your equipment accomplish that? Obviously if things are wiggling by 40um or more then adjustments on the scale are really just random perturbations.
Logged

- Andrew

My awesome blog about photography: http://photothunk.blogspot.com
Gigi
Sr. Member
****
Offline Offline

Posts: 416


WWW
« Reply #2 on: May 30, 2012, 08:48:10 PM »
ReplyReply

Interesting experiment - and forgive the silly question, but why are there many strips? One or two seem very sharp and in focus, the others not. What's theintention?
Logged

Geoff
torger
Sr. Member
****
Online Online

Posts: 1400


« Reply #3 on: May 30, 2012, 11:50:35 PM »
ReplyReply

Interesting experiment - and forgive the silly question, but why are there many strips? One or two seem very sharp and in focus, the others not. What's theintention?

I was probably not clear in the description, the folding rule is slanted 45 degrees up against a wall. I split the images in strips to make it fit better on screen, but it is originally from only one ground glass image and one image from the digital back, you can see the marks on the ruler fit together with some overlap. Attached a photo.

This means that only one part can be in focus, and the test was to see how precise focusing the ground glass *really* allows when there is adequate magnification. A common criticism of ground glass focusing for digital tech cameras is that it is not precise enough, but I think my experiment shows it is much about magnification.

Would there be focus shift in the sliding back / adapter plate / digital back the photo of the ground glass would show another section of the folding rule in focus than the resulting digital back image. I think the experiment shows that focus shift is negligible.

The 40 micrometer number is my estimation of the focus error - I tried to focus at 130 cm mark but got it at ~138 cm, and I think that is due to resolution limitation and it is not that easy to see where the *exact* middle of the f/4.5 DOF is.

It is interesting that the largest magnification Linhof provide themselves in their system is only 3 times! The photo of the ground glass one could say corresponds to about what you would see in a 50x magnifier. The typical saying about ground glass is that 6x is the optimal magnification, if you have any higher you just see grain but no more detail. For 8x10" 6x is probably adequate. For a digital view camera with 48x36mm image area it is clearly not. It is also clear that 6x is noway near the optimal magnificaton for critical focusing -- starting to see grain is not the same as stopping to see detail. Visible grain actually provides the advantage that you can see you loupe is correctly focused.
« Last Edit: May 31, 2012, 01:10:09 AM by torger » Logged
torger
Sr. Member
****
Online Online

Posts: 1400


« Reply #4 on: May 31, 2012, 12:13:17 AM »
ReplyReply

It looks like you're making movements of 40 micrometers to refine the focus here, if I am understanding your post correctly?

Do you think that when the sensor is placed into the plane formerly occupied by the ground glass, that it's within 40 micrometers of the plane the ground glass was in? I have no sense of how accurately built that component is, and how reproducible all this business is. Obviously one COULD build equipment that placed the sensor plane at the same plane as the glass to whatever accuracy you like, the question is, does your equipment accomplish that? Obviously if things are wiggling by 40um or more then adjustments on the scale are really just random perturbations.


Hopefully my reply above helps to explain a bit more. The 40 um number is an mathematical calculation backwards from the 5.6 cm focusing error how much that corresponds on the focus rail.

I have used my hand to turn the focusing knob in normal fashion and used the very high magnification setup instead of a normal loupe to try to focus as exactly as possible on the 130 cm mark on the folding rule. I took a photo of the ground glass as I saw it, and then I slided in the digital back and took a picture.

The digital picture seems to have focal plane closer to the 138 cm mark, 8 cm off my aim which divided by sqrt(2) (from the 45 degree slanting) is about 5.6 cm which for the 90mm focal length is about 40um on the focus rail.

My conclusion is that this error is not due to focus shift - I can't see any in the pictures, but rather that it is simply hard to see where the center on the f/4.5 DOF is on the ground glass.

I don't consider this a large focusing error either - a tech camera is most often shot at f/8 - f/11 and then this error would be masked. Another aspect valuable to know is that if you reduce the focal length the DOF correspondingly increases so it is sort of a myth that you need higher focus precision with shorter focal lengths - to hit an exact distance mark higher precision is indeed required, but to hit within the same range within the DOF (which is what counts in practical photography) no more precision is needed.

The takeaway message is 1) ground glass focusing precision is better than most seem to think, 2) much higher magnification than normally used is required to max out the ground glass potential.

I'm now trying to figure out what type of magnification device I should get to my camera.
« Last Edit: May 31, 2012, 12:14:56 AM by torger » Logged
torger
Sr. Member
****
Online Online

Posts: 1400


« Reply #5 on: May 31, 2012, 12:22:53 AM »
ReplyReply

Another experiment I did was to focus as precisely as possible at the 130 cm mark using the normal 10x loupe. These are the results after 9 tries:

try #1 of focusing at the 130 cm mark: result: actual focus at the 90 cm mark
try #2: 90 cm mark
try #3: 95 cm mark
try #4: 112 cm mark
try #5: 118 cm mark
try #6: 75 cm mark
try #7: 115 cm mark
try #8: 100 cm mark
try #9: 102 cm mark

That is quite far away from the 130 cm mark. The reason the precision is much worse is that with 10x magnification I cannot see the numbers on the ruler.

The reason I always focus closer in this experiment is my focusing technique which is to always turn the knob outwards for the final adjustment since it is better to focus a bit too close than too far away especially if focusing at infinity.

It should be said though that in practical landscape photography (which is my business) this "poor" precision is good enough, and when using tilt the precision requirement goes down further. The worst case would be to focus on a flat wall, so I guess for architecture photography a higher magnification would be nice.
« Last Edit: May 31, 2012, 12:26:41 AM by torger » Logged
torger
Sr. Member
****
Online Online

Posts: 1400


« Reply #6 on: May 31, 2012, 12:30:42 AM »
ReplyReply

I think these should be really interesting experiments to any digital view camera user out there, but obviously I need to work a bit on the presentation Smiley
Logged
henrikfoto
Sr. Member
****
Offline Offline

Posts: 708


« Reply #7 on: May 31, 2012, 01:29:20 AM »
ReplyReply

Hi Torger!

Thanks for presenting this experiment. This is more or less a problem for all users of these cameras.

Do you think this error is caused by focusing-problems like lacking fine-adjustments on the camera,
or do you think it's the sliding-back thats causing it? (that the ground-glass is not in the same position
as it sensor is placed)

I have had the same problems with the sliding-backs. These system should have some kind of shimming-
system. The backs are not prodused to 0 tollerance. In fact the producers of DB normally have a 30-40 microns
tollerance in there systems. Without shimming there is no way to get this accurate with a ground-glass.

Henrik
Logged
torger
Sr. Member
****
Online Online

Posts: 1400


« Reply #8 on: May 31, 2012, 02:33:42 AM »
ReplyReply

Thanks for presenting this experiment. This is more or less a problem for all users of these cameras.

Do you think this error is caused by focusing-problems like lacking fine-adjustments on the camera,
or do you think it's the sliding-back thats causing it? (that the ground-glass is not in the same position
as it sensor is placed)

I have had the same problems with the sliding-backs. These system should have some kind of shimming-
system. The backs are not prodused to 0 tollerance. In fact the producers of DB normally have a 30-40 microns
tollerance in there systems. Without shimming there is no way to get this accurate with a ground-glass.

My evaluation of the result of my own camera system (Linhof Techno with standard ground glass and Linhof sliding back, plus Linhof-made H-mount adapter plate and H-mount Aptus 75 digital back) is that there is negligible focus shift problem, that is the precision in the system is adequate. I base this on that I cannot see that the ground glass image and resulting digital back image is shifted in any way. There still could be a shift -- since GG resolution is lower than the digital back resolution it is not possible to see where *exactly* the focal plane.

From what I have heard, Linhof is one of the manufacturers that have good reputation concerning good precision in sliding backs, and this experiment fits into that. I have heard much more criticism of Arca Swiss view cameras that the ground glass is not matching. I have no possibility to check whether this is true or not though.

The small focus error (which I think is acceptable) -- that I aimed for 130 cm mark on the GG but the resulting image shows ~138 cm as far as I can see -- is not due to focus shift or difficulty to turn the knob precise enough, but due to that GG shows detail corresponding to about 15-18 megapixels and the sensor is 33 megapixels i e it is not feasible to with the eye see better precision than that. I had no problem to use the knob and turn the lens into the position my eye said it should be in. It would be more elegant with a dual geared focus rail, one finer gear for fine-adjustment, but it is really not necessary, with tight gearing as Linhof has micro-adjustments can be made by hand.

I may not been that clear but my thinking around this is that focus precision of the ground glass is adequate, but the manufacturers (Linhof in this case) do not take the need for magnification seriously and that is the real problem.

I'm a bit of a view camera fanboy Smiley so I'm also of the view that there's a little bit of FUD coming from the pancake camera manufacturers, exaggerating precision errors and also exaggerating the practical impact of such errors. You don't really need to nail focus within 0.0001 cm at 10 meters when you're doing f/11, and an important principle is that with reduced focal length the DOF increases correspondingly so the precision requirement on the linear focus rail does not increase in practice - it is indeed harder to hit a specific distance but DOF is wider so it does not matter.

The experienced increased difficulty to focus wide angles I think is mostly due to lack of magnification (and dimness of ground glass) rather than a precision problem in the equipment.

I'm a bit skeptical about the need for shimming, unless there really is poor manufacturing precision, which would be a scandal concerning the price of these things. It may be a greater need for shimming in pancake cameras since I guess there's more of a risk to focus past infinity there than with proper view camera focusing technique.
« Last Edit: May 31, 2012, 02:36:40 AM by torger » Logged
Gigi
Sr. Member
****
Offline Offline

Posts: 416


WWW
« Reply #9 on: June 03, 2012, 06:07:01 PM »
ReplyReply

very interesting. The practical sense of this investigation is very admirable, and often lacking in other reports. Thanks for doing this. Makes you think - what do you really need?
Logged

Geoff
ced
Sr. Member
****
Offline Offline

Posts: 266


« Reply #10 on: June 07, 2012, 03:43:35 AM »
ReplyReply

It is not so much the focussing accuracy which the camera manufacturers claim are critical but the accurate positioning of the sensor relative to the focal length of the lens.
Some backs now offer tools that assist in verification of focus accuracy before nabbing the image finally.
Logged
wentbackward
Newbie
*
Offline Offline

Posts: 9


« Reply #11 on: June 08, 2012, 07:41:24 AM »
ReplyReply

If you use liveview on your PC/Mac you can verify for certain if there is a discrepancy between the sensor plane and GG.
Logged
Pages: [1]   Top of Page
Print
Jump to:  

Ad
Ad
Ad