Medium format resolution

Got a situation here -

 

I need to be able to produce images of archaeological excavation pits. The size on these things is 8

meters by 8 meters. The challenge is to raise a camera above it, photograph downward, and provide

visible resolution down to 1 centimeter. This will require a VERY wide lens. I've found that at a 35MM

camera with an 18MM lens at 9.5 meters away will encompass 8 meters by 8 meters - but the resolution,

no surprise, is useless. So I may need medium format. Cost might not be an issue; I may be able to gain

access to high-end gear ( like a Hasselblad H3 digital setup ). I know that the 6 x 6 format of a Hasselblad

is better suited for that than a 645 like a Pentax or Mamiya, but I will still need to raise the camera pretty

high to do so ( a platform has already been devised to raise a camera to 10 meters. Shorter height would

be better, though ). Assuming a 9.5 meter distance from the surface, do any photo.net people know if

medium format or high end digital can resolve a 1 centimeter increment? ( tack sharp resolution isn't

required here, by the way ).

Your best bet might something like an Alpa 12, a Cambo Wide DS or a Horseman SWD. These

are all dedicated wide-angle cameras. Most of them have some shift capabilities which may

be useful. They take large format lenses which gives you a lot of flexibility. The lenses can go

VERY wide, and the quality will most likely be even better than the best hasselblad lenses.

They take digital backs, or they will shoot 4x5. If you use a fairly large sensor digital back

with the SWD and use a 24mm lens, it is equivalent to about a 17mm lens on 35mm, so

enough for your purposes.

Err... you've just described a 1600x1600 pixel image. Just about any modern camera can handle that. 6x6 MF should be able to handle ~1mm pixels on an 8m x 8m field if the scanner is good enough (ie Nikon 9000).

I would think a superwide would be perfect....

I'm perhaps a little confused here... as someone else mentioned:

<blockquote>

8 m * 100 cm/m = 800 cm length, you say tack sharp resolution is not required - so even

say 4 pixels per cm? to resolve = 3200 * 3200 = 9 megapixels.

</blockquote>

I guess I'm curious why a well shot (stable mounted, mirror lockup) fine grain 35mm

image and well scanned (4000 dpi, no ICE - may blur details, clean neg/slide first) doesn't

almost get you what you want - which implies that a Canon 1Ds MkII might be sufficient?

<p>

Maybe I'm looking at this entirely wrong?

<p>

Medium format would give you more resolution and room for error - but I'm wondering if

there is something wrong with the method of capture.

Do you have an example bad image and can you explain how it was produced (I get the

impression from your message that a digital result in the end is desired).

<p>

I could be completely be screwy in how I am working this out... But I shoot a 4x5 and scan

that at 2040 dpi when I want detail.

theta = approx 1cm/950cm = 1/1000 radian to slide rule accuracy. In military theodolites this arc angle is a mil; with 6400 mils being a full circle. In army gunnery a mil is about 1 yard in 1000 yards. <BR><BR><a href="http://www.liberatorcrew.com/Clinometer/Clinometer.htm"><b>Guns,</b></a> cannons and military sights and theodolites such as the Wild Heerbrugg T2 are often in mils instead of degrees, minutes and seconds. Math is easy using mils when in the heat of battle. If one misses the target by 120 yards to the left at a range of 2400 yards then you move the gun over by 50 clicks ; ie 50 mils; ie 1 mil at 2400 yards is about 2.4 yards.<BR><BR>With a 18mm lens on a full frames Canon 5D 1 mil is 18 microns. The sensor pitch is way less than this; about 8.2 microns. Thus with a decent high quality L series canon lens; or Nikon one with an adapter you probably will resolve a 1 mil object.

What is the 1 cm objects shape? are you trying read type thats 1cm high? or locate rocks, fissures, asps in the raiders of the lost arc pit? <BR><BR>How much "design margin" one needs in a photo capture project radically depends on the contrast and shape of the objects. If one wants to read text with a poor font and ill lighting then one needs alot more design margin. If the 1cm objects are just mice, asps or marbles then one needs less margin, unless the asps are hungry.

Few more details - What needs to be photographed within the dig site is a measuring stick

- a ruler with centimeter increments. My initial results with 100 -speed film and with

digital (6.3 MP) gave me a image that when I zoomed in on the ruler, the result was so

pixelated as to be impossible to accurately gauge the centimeter increments. The other

details to note here are this: If possible, I'm trying to create a setup that "Joe Archaeolgy"

can use - and generally, that knowledge is limited to PHD ( press here, dummy ) cameras.

Gotta be simple. The other tricky factor is that two shots, one with ruler, and one

without, of the same subject might be needed - two shots from 9.5 meters up. That's

going to require an auto-advance film model ( if film is the answer ), with a remote trigger

device.

I know this is difficult, but I'm trying to come up with something very new for the world

of archaeology, and experimentation is needed.

Thanks to all for the answers so far!

Matt, interesting challenge. I shoot film and scan at 5400 ppi. So your 800cm plot will be 7654 pixels across or 10pixels per cm. I've just looked at a techpan 35mm shot done on a tripod with a zeiss lens. The gaps in brickwork at 30m on a 45mm lens (about 1cm across) are quite obvious.

 

Moving up to 6x6 gives you 20 pixels per cm, if you can scan at the same resolution. This sort of scanner however is big money. If you scan at 2700 then you're no better than 35mm (although grain will be better but lenses will be worse).

 

So I find that I can get the same resolution with 35mm and 6x6 if I use the right film and are scrupulous with technique - not phd cameras though. Ideally you need a large format scanning back which would be ideal for this. Just have to mortgage your home to pay for it.

Hassy SWC, which gives you a digi-back option.

<blockquote>

My initial results with 100 -speed film and with digital (6.3 MP) gave me a image that

when I zoomed in on the ruler, the result was so pixelated as to be impossible to

accurately gauge the centimeter increments.

</blockquote>

Whoa, whoa, whoa! You don't want to distinguish 1 cm wide objects! You want to

distinguish objects MUCH smaller (the lines on a ruler essentially - and perhaps the

numbers with those lines). Eeeeeek!

<p>

Now I'm sitting here thinking Medium Format may be challenged also.

<p>

How much budget do you have? Is immediate feedback necessary? The costs escalate

going to medium format and then to large format (either via a film path or digital path).

The large format scanning back suggestion is a possibility that *may* be cheaper than a

39 megapixel medium format back?

<p>

Can you go to the site and do some MF test shots?

<p>

Good luck.

Matt; usually with archaeological and accident photos folks use special rulers, rod and tapes with large tickmarks, major and minor to radically make the image easier to read. Often sections of a metric surveyors level rod are used; with ticks 1/2cm on some; and 1cm on others. Or a Keson surveyors pocket rod is used that reads in inches, tenths of a foot, or metric. In surveying these rods are made to be read at great distances, with clear major and minor markerings of great contrast. Using a simple ruler is just asking for massive trouble; or making the requirements of the image capture device an order of magnitude more costly. Survey targets and level rods are purposely made to be readable. In metric a 1cm rod has a 1cm black, then a 1cm white bar; with majors at 10cm marks and 1 meter marks that jump out. Thus there is no guessing.

The rod in the center is a metric level rod with a 1cm interval. Pros use rods like this for images, instead of farting around with a wimpy walmart metric ruler. The red and black zones are each 1/10 meter intervals; ie 10cm each.<BR><BR><a href="http://www.mytoolstore.com/berger/rods1.html"><img src="http://www.mytoolstore.com/berger/06fbrods.jpg"></a>

Opps; the red interval is 5cm; the black is 5cm; the red & black combined is 10cm!

What our customers do is often make a "survey rod" like pattern on metric graph paper; with the red and black "centimeter E's" as a master. Then its color copied and many made; then they are laminated. These are placed at the scene for location and scale; with notes added if required in LARGE BLOCK TYPE.

A survey rod is like a high contrast lens test target; its going to give one a radically better system resolution results than a thin lined wimpy ruler, yardstick or grey scale pictorial object. <BR><BR>A great reference target allows one to have radically better design margin with a reference in image capture with aerial, archaeological, scientific and crime scene work.

That is some very good advice Kelly. Better and cheaper than a scanning digital back...

Thanks so far - It's not the lines on a standard ruler that need to be resolved. Using

Photoshop, I was able to construct a ruler with 1cm wide bars, and the fellow I work with

confirmed its accuracy. That's the target. As Kelly suggests, perhaps one of those levels may

be better. I think I can track one of those down. And I did not know that the SWC had a

digital option.

The adventure continues.....I know Indiana Jones never had this trouble - just a bunch of

Nazis.

Thanks again!

Look into the Mamiya 7II with a 43mm lens. Lens is arguably the sharpest out there,

21mm equivalent. The Camera is a rangefinder (focusing won't be a problem this wide),

lightweight, takes 220 film (20 exposures) and is 6x7, the best step before going to

cumbersome large format. Best of all it has aperture priority, making it a huge PHD camera

if it's stable enough at low shutter speeds.

Before I get dogpiled, make that 220 gets 24 exposures ;)

Only problem with the Mamiya 7 series is that it can't be auto-advanced. It would be

difficult to advance a manual crank from 9.5 meters off the ground. The process may very

well require two shots in a row - one with measuring stick, and one without.

After some consultations with a photo store ( semi-pro clientele ), I think a multiple

camera setup with Photoshop stitching may be best ( D-SLRs, 35mm film ). An

archaeologist can't get medium format film developed everywhere, and then scanned.

After all, I'm trying to develop something that's universal. High end and medium format

solutions may be developed later down the road.

Thanks to all!

Matt

"I think a multiple camera setup with Photoshop stitching may be best"

 

Take a look at the open source Hugin program. This will give effortless and seamless digital stitches.

 

Multiple cameras probably won't be all that easy to do. Parallax will be an immediate issue unless you're doing linear mosaics. At the least, I think you'll have to custom build a jig ensure setup uniformity from site to site.

I second the multiple exposure option. There was an artist, British I think, that made portraits of tress be scaling them and shooting them, say, a meter at a time. He later made huge composite prints in which you could make out the details in the individual leaves.