leonard_evens

<p>I have only base tilt on my Toho Fc-45X. Here is how I proceed. I choose a high point and a low point I want to be in the depth of field and then try to visualize the exact plane of focus halfway vertically between them. I choose a near point and a far point in that exact plane of focus. <br>

The first object is to get those simultaneously in focus. I make an initial guess, usually about 5 degrees. I focus on the far point and then focus on the near point. If I had to move the standards further apart to accomplish that, I increase the tilt. If I had to move the standards closer together, I decrease the tilt. After a few iterations of this procedure, I find the near point and far point a simultaneously in focus.</p>

<p>The next object is to choose the correct f-stop. I focus on the high point and low point and measure the focus spread in mm along the rail between those points of focus. I multiply that number by 10 and divide the result by 2 to get an initial estimate for the correct f-number. I usually add from half to a full stop beyond that. I then check by stopping down. If the calculated f-stop is not beyond f/16, the gg image is still bright enough to check what is in focus. . Beyond that, the image is too dim to judge. But the following trick sometimes works. Look at what is in focus vertically on the gg at f/16 at some fixed distance from the lens, and measure it in mm. Find the ratio of the deisred f-number to 16. That factor tells you how much the vertical window on the gg would expand if you stopped down to the smaller f-stop, were the image not so dim. If that would include everything in focus that you want, you are okay. Otherwise choose an even smaller f-stop.<br>

The method described above ignored diffraction. If you have to stop down to f/45 or even f/32 in some critical applications, diffraction may be an issue. In that case you should use the Hansma tables described at lfphoto.info to determine the proper f-number.</p>

<p>You can work everything out using the lens equation:</p>

<p>1/u + 1/v = 1/f</p>

<p>where u is the subject distance, v is image distance, and f is the focal length. But when the subject distance u is smaller than the focal length f, the image distance v is negative, meaning the iamge (which is a virtual image as described above) is in front of the lens instead of in back of the lens.</p>

<p>As has been noted, the image goes out of focus if you move the lens closer than the flange focal length. However, if you stop down enough, some distant subjects may still be in acceptable focus provided you don't move more than a few mm closer than the focal length.. The reason is that infinity will still be within the depth of field, which will extend `beyond infinity'.<br>

Generally when the subject is between the lens and the front focal plane, there will be a virtual image. That means if you look at it from in back of the lens you will see a magnified image. That is how magnifying lenses work. But this virtual image cannot be captured on film or any other physical device without use of additional optics. When you use a simple magnifying lens, your eye provides the additional optics and produces an image on your retina which is transmitted to your visual cortex.</p>

Jim,

 

I'm not sure I understand your method. I think I understand why you rounded f/9 down to f/8, but why did you round f/13.5 to f/12.5? Are you rounding down to the nearest 1/3 stop and you happen to know that f/12.5 is about 1/3 stop down from f/11. Does this method require knowing f-numbers in 1/3 stop increments?

 

I use a tape, on which I've indicated the corrections in f-stops for each of my lenses. I just stretch the tape along the rail and read the correction. Of course, I had to do the calculations once in preparing the tape. But, in practice, it requires no calculations at all.

Jim,

 

Your method works pretty well in the situation you describe You have a relatively long lens. Also, you

apparently initially set the aperture at about f/7, presumably because your exposure meter told you that was

right without including the bellows extension. That means the diameter of the aperture is pretty large and can

be measured reasonably accurately.

 

But consider a more typical case. Suppose I am using my 90 mm lens, and my exposure meter tells me that f/22 is

right if bellows extension is ignored. And suppose the reproduction ratio is close to 1:2, which seems to be

the case in your example. That means the bellows extension would be about 135 mm. On the other hand, 90/22 ~

4 mm would be the diameter of the aperture. I have little confidence that I can measure a 4 mm opening in the

diaphragm by holding a scale against the front of the lens at all accurately. Suppose I measured it at closer to

5 mm. I would then decide that the effective f-number were 135/5 or 27, when in fact it would be 1.5 x 22 = 33.

That would put me off by over half a stop. Suppose on the other hand, I decided it were 3 mm. That would

give me 135/3 = 45 for the effective f-number. That would put me off by almost a full stop.

 

So maybe there is some use for those formulas anyway.

Tak,

 

I've done the analysis to see how the normal fall-off is affected by a tilt. Unless the tilt angle is pretty

large, the tilt shouldn't affect things that much. I suspect that the big difference you see is mainly due to

the fact that one part of the scene is close to the lens axis and the other part is far from the lens axis, and

you are seeing mainly fall-ff from center to edges, with a small contribution from the tilt thrown in. If your

lens follows the normal 4th power of the cosine law that could conceivably result in a two to four stop

difference. Even if your lens has only cube of the cosine fall-off, this could amount to a significant

difference.

 

So your best strategy would be to use a center filter to deal with the lens fall off and make the fall-off

symmetric with respect to the center of the picture by placing the latter roughly where the lens axis intersects

the film plane. The simplest way to do

that is to use back tilt. But you can also do it by using front tilt plus rise/fall or shift of one or both the

standards and repositioning the camera relative to the scene.

Of course, there are limits to such movements, so it may not be possible to do it the second way, depending on

the details of the scene

I should have added that you can download something called Quickdisk, which does it all automatically for you. Try a google search if someone else doesn't provide a link.

If you are doing close-up photography, you also have to compensate for bellows extension. That means you have

to increase exposure to compensate for the fact that the film is further from the lens than would be true for a

distant subject. There are a couple of ways to do this. The simplest is to put a disk of know diameter at the

point of focus and then measure its diameter on the ground glass. The ratio of that measurment to the actual

diameter is called the magnification. You then add 1 to it, square the result and multiply the time of exposure

by that number. For the scene you describe, I would get is would be about 0.5 or 1/2. Suppose that it the case,

then 1 + 0.5 = 1.5 and its square is 2.25. If you exposure meter tells you to expose for 1 sec, you would

expose instead at about 2 1/4 seconds

 

But it is probably easier to adjust the f-stop instead by opening up the appropriate amount. Here is a table to

tell you haw much to open up in fractions of an f-stop

 

magnification / #-stops

 

0.1 / 0.28

 

0.2 / 0.53

 

0.3 / 0.76

 

0.4 / 0.97

 

0.5 / 1.17

 

0.6 / 1.36

 

0.7 / 1.53

 

0.8 / 1.70

 

0.9 / 1.85

 

1.0 / 2.0

 

In using this atable it is fine to round off to the nearest third or half stop.

 

In the example where the magnification was 0.5, if you opened up by about 1 and 1/3 stops you would probably be

okay.

There are three things that you want.

 

1. You want the center of perspective to be off to one side to achieve two point perspective and show part of

the side of the building.

 

2. You want the rear standard to be parallel to the plane of the building facade so that the front of the

building will end up square.

 

3. You want the lens axis passing roughly through the center of the frame to avoid loss of definition at the

limits of the image circle.

 

There is only one way to accomplish 1; you need to position the camera properly. To accomplish 2 and 3, you can

either point the camera in the appropriate direction (3) and then swing the back (2), or equivalently, set up the

camera so the back is parallel to the facade (2), and then swing the front (3) to put the lens axis where you

want it. It doesn't matter how you do it,

the end result will be the same. But you will also introduce one side effect. The rear standard will not be

parallel to the front standard, so the exact plane of focus will not be parallel to building facade because of

the Scheimpflug Rule Exactly where it will be will depend on the geometry of the situation. So the only

remaining question would be whether you can get the entire scene in focus by stopping down far enough. Because

of the wedge shape of the DOF region, that may not be so easy, but because you are using a short focal length

lens, it may be possible.

 

It is possible to calculate in advance if this is possible if you know what all the parameters are, but it makes

much more sense just to try it. One problem with this is that if you need to stop down beyond f/16-22, you

won't be able to see well enough on the ground glass to determine if everything you want in focus is in focus at

the taking aperture.

But you can use the following method instead. After you have set up everything the way you want it, focus first

on the extreme point to the left you want in focus, note the position of the standard on the rail, do the same

thing for the extreme point to the right. Then find the distance on the rail in mm between those extreme

positions, multiply it by 10 and divide it by two. That will give you an estimate of the f-stop you need to

use. You would then set the focus point halfway on the rail between the two extreme positions, and perhaps stop

down from half to one additional stop beyond what the above calculation gives you to account for inevitable

errors. For example, suppose the focus spread between the two positions was 6 mm. 10 x 6/2 = 30, so you would

need to stop down to to something like f/32 to f/45.

 

Another thing you can do is to stop down as far as you can still see and measure the distance on the ground glass

of what is in focus, and find the ratio of the total left right distance to that. Then multiply the f-number you

are using by that ratio to estimate the f-number you actually need. For example, suppose at f/16, about half of

what you want in focus is in focus, so the ratio of what is wanted to what is in focus at f/16 is 2 to 1.

Multiply 16 by 2

to get 32. In principle f/32 should work, but again, you might want to stop down half to a full stop beyond

that to be sure.

 

Finally, there is one potential problem, particularly if you use a format smaller than 4 x 5. If you need to

stop down pretty far, diffraction will begin to be a problem.

 

Taking several shots and bracketing would probably help you avoid going back and redoing the whole thing., but

given the possibly extreme demands you are putting on the scene, that may be necessary in any event.

Michael Briggs gave you good advice. Let me just add one point about using tilt (or swing).

 

Guess a tilt to begin, it doesn't matter much what it is but about five degrees would be right.

 

Select a near point and a far point which should be in the exact subject plane you are aiming for.

 

Focus on the far point and refocus on the near point noting whether the standards move closer together or further apart.

 

If they move further apart, increase the tilt. If they move closer together, decrease the tilt. Stop when both points are in focus.

 

After playing with this a while, you should be able to get the desired plane in focus in two or three iterations.

It should be noted that in the case of negative film, you can overexpose to compensate for the fall-off of illumination and then by means of dodging produce an evenly illuminated image when printing using an enlarger. If you scan, you can do the same thing digitally. But if you are using slide film, since it has relatively little latitude, the extent to which you can apply this method is limited.

 

I used this method to compensate for fall-off with a wide angle lens as a substitute for a center filter. It works reasonably well, but it can require a lot of fiddling, and all told using a center filter is easier. A center filter also works with silde film.

If you live in the Chicago area, you might consider joining the Midwest Large Format Asylum

 

/home.comcast.net/~midwestlf/

 

They are an active group of experienced large format photographers. They go on monthly outings,

and you might benefit by going along some time.

I have the 90 mm f/6.8 Grandagon_N. I am very happy with it, and I use it pretty often. Its coverage is adequate for my purposes, particularly since movements with the bellows so compressed are somewhat limited for my camera..

 

I used it for several years without a center filter. But at one point I got the 75 mm, f/4.5 Grandagon-N, because I found stiutations where the 90 mm lens wouldn't let me get as close as I needed to be. I probably could have managed without a center filter for the 75 mm lens, at least of landscapes, but I found that in architectural photography, my pictures would show color shifs at the periphery because of different degrees of fall of illumination. So, I finally got a center filter. Since the same filter works on the 90 mm lens, I now use it pretty regularly, but if I didn't have the 75 mm lens, I wouldn't have gotten it.

I've been doing 4 x 5 panoramic photography for about a year now. I have a way to go before I will feel I've mastered it, but I have managed to make some pictures.

 

I wanted to photograph building facades that I couldn't get far enough that my widest angle lens, a 75 mm Grandagon-N, could encompass the entire building. So I tried to see if I could do it using two or three exposures and merge them digitally.

 

You have to be somewhat careful about the sort of scene you try to pht ograph this way. If there are clouds in the sky, they may change quickly between exposures. Subject movement, even leaves rustling can create problmes when you merge images.

 

There are two ways you can go about it. The simplest way is to fix the position of the lens, make one exposure with the back shifted all the way to the right and then another with it shifted all the way to left. In principle, it should be easy to merge the two images digitally in a photoeditor such as Photoshop. (But, I use gimp.) It turned out that this was a bit harder than I had anticipated. For example, it may be hard to scan so the intensities match well. You don't want to just match the two images along a fixed line because it may show up in the picture, so you need to use layers and transparency methods to merger the common section. But other than that sort of thing, there is nothing special required.

 

If you can't encompass the entire system that way, you need to swing the lens first one way and use software designed to merge such images. There is a lot of stuff out there on the web describing how to go about it.

 

First, when taking the picture, you need to roatate about the entrance pupil, so you have to find out where that is. You do it by trial and error. You start by placing the turning axis at the lensboard, look at elements of the scene, one in a direct line behind the other and see if one moves with respect to the other as your rotate. That is called parallax. You adjust the position of the rotation axis until you see no parallax shift. I don't see how to do this easily without a panoramic head. There are several on the market, not all of them that appropriate for a large format camera.

I got mine from www.stereoscopy.com/jasper/panorama-test.html and it works fine with my camera.

 

Finally, you need a software package to merge the individual images. There are several different systems based on Panorama Tools, which can do amazing things. I use something called hugin because I work under Linux. I found the learning curve for mastering this software pretty steep, although there are tutorials which show how to do basic operations. I would start by looking at wiki.panotools.org/Panorama_tools

and following the links.

 

An alternative, do-it-yourself method would be to use perspective correction tools in your photoeditor, such as Photoshop, to straighten out your images and then to merge them. But the software packages designed for this purpose do a mcuh better job, if you can get them to work. People use these tools with digital cameras to make impressive panormas from many, many component packages. There is a lively, very helpful listserve which provides a forum in which to ask question. See the Wiki for links.

Let me do some order of magnitude calculations. Suppose your lens has an image circle of diameter 220 mm, and you are rising to the maximum amount possible. The coverage angle measured from the center to the edge would be about 42 degrees. In general, exposure drops by the fourth power of the cosine of that angle. Doing the calculation, the exposure at the periphery might be as low as 30 percent of that at the center, i.e. requiring as much as 1.75 stops additional exposure It is the necessary exposure at the center that your exposure meter is proposing. So unless there is something I'm getting wrong, that might possibly explain some of what you see. Of course, that reduction would apply only to the part of the image near the periphery of the image circle. The reduction would be considerably less closer to the lens axis.

 

But I have to ask, are you really using a rise of something like 50 mm, which is what would be required? That seems a bit unlikely.

 

I suspect the problem has more to do with how you are using your exposure meter to determine the exposure. If you are doing that right, there might be something wrong with the meter.

 

A center filter is usually not considered necessary for a 120 mm lens, but it is possible it would improve the situation. Or you could scan and adjust digitally. But I would thoroughly explore other avenues before trying those.

I use the near point far point method to place the exact subject plane where I want it. My camera has base tilt and axial swing. I don't see much difference.

 

This is how I tilt. I choose two points, one in the foreground and one in the background, which should lie in the exact subject plane. I choose the tilt mostly by experience, but in tricky situations I may use Wheeler's Rule, which I won't go into here, except to say that it gives you a very good first try for tilt angle. I focus on the rear point and then refocus on the near point. If I have to move the standards further apart to get the near point in focus, I increase the tilt. If I need to bring the standards closer together to get the near point in focus, I decrease the tilt. Usually a few iterations of the procedure gets me right on.

 

In dim light, it is of course more difficult to focus, even without tilts. I haven't found though that tilts are any harder than anything else. If the light is so dim you can't see much of anything, you have to do something else, and I doubt if switching from base tilt to axial tilt will help you much. You would be better advised to get a brighter focusing screen. I got a Maxwell screen for my Toho FC-45X, and I find it much easier to focus in any kind of light.

There appear to be two additional problems except the need for perspective correction. There is a slight rotation that should be corrected first. In addition, there may be some barrel distortion.<div></div>

If you are interested in making only modest enlargements, a scanner like the Epson V700 or V750 should be adequate for your needs. But you can't get an effective resolution better than about 35 lp/mm with such a scanner, which should in principle allow perhaps a 5 X enlargement, but in practice in might not do as well. (I've produced such enlargements which I found satisfactory using an Epson 3200.) So if you expect to make very large prints and have people view them close up, you would be better off using ahigh quality scanning service. You would also be better off having a good lab make such prints rather than using an RGB inkjet printer.

Why the subject should not be too close in portraiture.

 

The human visual system has a property called size constancy. That means that, within certain bounds, the apparent size of an object doesn't depend on how close to you it is. Ao, if you look at a person's face at 1 meter, 2 meters, and 3 meters, it looks more or less the same. The image produced by a camera doesn't so compensate. It will be much larger for an image closer to the lens. Another related property of vision is that relative dimensions appear to be preserved. When a face is two or three feet from the lens, the distance from the tip of the nose to the ears is a much larger percent of the total distance than if the subject is six to eight feet from the lens. The camera faithfully records this, but your visual system distorts it so you see the same thing. As a result, the faithful camera image looks distorted relative to what the eye/brain sees. When the face is too close, the nose may be exaggerated. As you move the subject further away, you "flatten" the camera image so it better approximates what you see in your mind's eye.

 

This is sometimes called perspective distortion, but that is a slightly different, albeit related, phenomenon. If you put your eye at the same distance proportionately to the print as the lens was to the subject, the image on your retina produced by the print should be the same as that produced by the scene. (But, seeing involves scanning and processing the retinal image, so, as noted above, the retinal image is not what you see.) If you make an 8 x 10 print using a 250 mm lens with 4 x 5 film---so using a 2 X enlargement---, you should in principle view the print from 500 mm, but you would normally view it closer to 250 mm. This won't make a significant difference in what you "see". But, if you use a 90 mm lens, the print should be viewed at 180 mm, which you are highly unlikely to do, and "distortions" at normal print viewing distances should be obvious.

 

Because of considerations like this, you are usually advised to put the subject at some distance to the lens. Just how far the subject should be is a matter of taste. If you like the results when the subject is relatively close, that is fine, but as I said, I prefer having the subject further away, and I believe most people prefer images produced that way.

 

The usual rule of thumb is that the appropriate focal length should be from 1.5 to 2 X the normal focal length for the format. But that is misleading, since it really depends on the format. As I noted previously, if you want to make a head and shoulders portrait, that determines the magnification, once you know the format size. You then have to adjust the focal length and subject distance to get that magnification. If you make the focal length smaller, you will have to place the subject closer to the lens. So you have to decide in advance how far the subject should be from the lens, and then choose the focal length to produce the desired magnification at that distance. I described in some detail above how to do that.

I'm not sure I understand what you are asking for. Large format lenses have relatively little depth of field at the normal distance you would put a subject. A head and shoulders portrait requires a magnification of approximately 1/5.5 for 4 x 5 format. That means the ratio of the subject distance to the focal length of the lens should be about 6.5. I wouldn't want to put the subject closer than about 2 meters (about 6 1/2 feet) from the lens. For that a 300 mm lens would suffice. You would be better off a bit further from the lens, so a 360 mm lens with the subject about 2.3 meters or about 7/12 feet would work pretty well. Using a 178 mm lens, you would have to place the subject at about 3.8 feet, which seems to me to be much too close.

 

In any case, at such subject distances, the depth of field is more or less independent of focal length, depending only on the f-stop, magnification, and maximum acceptable circle of confusion or coc. A reasonable choice for coc for 4 x 5 is 0.1 mm. Both the front and back deptn of field would be roughly equal to the product of the f-number, the coc, and by the reciprocal of the square of the magnification. The last term is 5.5 squared which is 30.25. If you stop down to f/11, the product of the f-number and the coc would be 1.1, and the front and read DOfs would be about 33.275 mm or about 1.3 inches, The total DOf would be about 2.6 inches. Do you really need less DOF than that? If so, you could just open up more. At f/8, it would be 8/11 of the previous number. If you really need less in focus than that, you can try larger formats. For example, for 8 x 10, the magnification would be 1/2.75, but you would use a coc twice as large or 0.2. The net result is that you would reduce the front and rear DOFs at f/11 to 16.4 mm or a bit over 5/8 inch.

There are two issues in printing a large print which you have to deal with.

 

The first is pixelization. Clearly, you don't want to see individual pixels in the print, but there can still be subtle effects if the pixels in the print are large even if you can't see them with your naked eye. That is where the 300 ppi figure comes from. But that assumes you are looking at the print from relatively close up, say 25 to 30 cm. If your viewers are going to get further back, then you can relax some. Roughly speaking, if you divide the viewer distance in cm by 30, you should be able to divide 300 ppi by that factor and still get acceptable results. But since this is only a rough rule of thumb and since some viewers will be more critical than others, you should be conservative about such reductions. If you expect viewers to get close, stay with 300 ppi (or possible a higher resolution).

 

The other issue is how much information about fine detail shows up in the print, which of course is limited by how much such information is in the digital image. This can be complex and is related to how the digital image was produced and what you do with it afterwards. If you scan a film image, it depends on the sampling rate in ppi and the quality of the scanner. Personally, I don't believe that you can make a print that large from a 4 x 5 negative or transparency which won't begin to show loss of fine detail on close viewing, but it would be minor if you had top quality equipment. Again this is much less of an issue if you don't assume viewers are going to be "grain sniffers" who get as close as they can and examine very fine detail in the print.

I run into this problem regularly. I am trying to photograph buildings in my local downtown before they are torn down by developers. I usually can't get back far enough, and my 75 mm lens is not nearly wide enough for some of the building facades. so I decided to investigate panoramic photography. This involves making several images, scanning them and merging the results digitally. The software is called Pnoramic Tools, and there are several gui interfaces for it. See

panotools.sourceforge.net

for information about the software. I work under Linux and use hugin, but there are other interfaces for use under windows and Macs.

 

With such software there are two ways you can proceed. (1) Center the camera on the scene with the lenx axix perpendicular to the facade. Then shift to the right and to the left and make separate exposures, with considerable overlap in between. Then you can fairly easily merge the two two images in a photoeditor or using Pano Tools. This may work if you can encompass the whole scene that way. (2) Use a panoramic head and rotate the camera to the right and to the left to make separate exposures and then merge them using the software.

 

Pano Tools is quite powerful, but it does have a steep learning curve. You can use it with a digital camera and merge many separate images. Roger Clark (www.clarkvision.com/photoinfo/large_mosaics/index.html) has produced high resolution pictures comparable to large format this way. I prefer using it with large format images because you gain the advantage of both worlds. See the attached example. It was done by merging two images using a 75 mm lens on 4 x 5 film and rotating both the left and to the right.<div></div>

I have a Toho FC-45X. It has a fixed bellows, but they make an eccentric lensboard which to some extent substitutes for a bag bellows. I use it with a 75 mm f/4.5 Rodenstock Grandagon. The eccentric lens board plus limited rise allow me to make full use of the lens's 195 mm image circle. But I couldn't come close to using the larger image circle of the 72 mm Super Angulon XL.

The formula for hyperfocal distance is wrong, but the example does it correctly. The numerator should be the focal length squared. I've notified the designer of the web page.

I am also an old timer. I like focusing by eye. Much of my photography is done with a 4 x 5 view camera where it is all done that way. But I've experimented with my Nikon D80 comparing auto focus to manual focusing. Auto focusing almost always did as well and often did better. I don't have a split image screen, but I did go back and look at my old Minolta SLR to see if it really was that much better. I found that I couldn't do that well with it as I thought I had, and I think the auto focusing with my D80 did just as well or better.

 

So before you start fiddling with changing the focusing screen, I would suggest learning to use the auto focusing system on your DSLR. There are some common errors to avoid and some more subtle ones. But with experience you can learn how to avoid them. When the auto system chooses the right point to focus on, it almost always does a better job than you could do by manual focusing with a split image.