Examples of Large Format shots that cannot be done with small

[fishbulb]

This is scanned at 1200dpi, so 6k by 4k8, about four times the limit of a top end electric sensor, due to the bayer filter and low pass filter.

Unless I'm doing the math wrong, I'm not sure I understand this statement.

6,000 x 4,800 = 28.8 megapixels... but 36mp, 42mp, and 50mp 35mm-size sensors (without a low-pass filter) are widely available. How do you get to "four times the limit of a top end electric sensor" at only 28.8 megapixels? 28.8/4 = 7.2 ... implying that those 36, 42, and 50mp sensors are only resolving 7.2 megapixels of real resolution?

From my own experience, if I want to get to 4x of what my D800 can produce, I need to scan 4x5 at 3000dpi, at least. Usually 4000.

Since I shoot 4x5 primarily for the resolution advantage, here is an example from me.

The first image is the full-size shot. The second is at 25% zoom, the third at 50%, and the fourth at 100%. The shot was Delta 100, f/16, Nikkor 180mm f/5.6, Sinar F, scanned at 4000 dpi on a drum scanner. The scan is about 300 megapixels.



I took the same shot on digital (although at a high ISO, so it's not directly comparable) and the resolution is not. even. close. If I blow up the digital image to match the last image above, it is just a mush, you can't even tell what is bark and what is dirt.

[Kirk Gittings]

Adam the question is though.........how would you work that image if it really got you excited and you only had digital with you (far from home)-which happens to me on commercial shoots on the road frequently. How would a serious photographer solve this challenge-walk away and let it haunt you? Not me. I'll tell you what I would do and it would be simple and I could get a good print out of it-much better solution than not getting an important image (though I would prefer film always). So I would use a tripod of course, base ISO, primo prime lens and a 4-6 X stitch-all of which would not take any longer than shooting it with a VC and would give me a very good exhibition quality 11x14 to 16x20 print. Apples to apples.

[fishbulb]

^ Yes, that is true, panorama stitching can be done for a lot of images. This image would have been easy - nothing is moving, the light isn't changing, the forms are all organic and can easily be warped in the stitching process to fit together and still look 'natural'. I just picked it for this example since I had it handy.

However, as someone who has done a LOT of panorama stitching from digital images (before I started with large format) I can say that it is not always so easy.

For example, it can be very difficult to get architectural panoramas to line up properly, due to all the tiny lines that must match up, even with the best software and the finest by-hand adjustments. Another example is that easy panorama stitching requires very little moving in the frame. Otherwise, the stitching becomes much more complex, and a lot of by-hand masking is required. Third, some subjects just don't look good stitched, or aren't worth the huge amount of work to do the stitching - portraits of people or animals, for example.

[Kirk Gittings]

I agree mostly, though stitching with a T/S lens can help with architectural lines. For me I have worked hard to figure out (based on my own taste in personal image making) what I can make work. Two examples of "must have" images when I only had digital available.

[djdister]

I have a less scientific comparison between scanned film and a digital shot, a bit like fishbulb's in post #133.

First is a 100% crop from a 20MP digital image. Second is a 100% crop from a scan from 5x7 negative, and last is a scan of the whole negative.
Both shots were taken from the same position, within seconds of each other. For my money, the scan from the negative has considerably better detail. If I applied a lot more sharpening to the digital image, it may appear sharper, but would also display sharpening artifacts.


digital


scan from film

[TXFZ1]

"Can't a few more people post some things that are easy to do the LF, that a small format person would have to buy special hardware, software, do workarounds, or otherwise get a less perfect result?"

OK. Here's a photo that was easy to make with a Sinar P. It was easy to provide movements to compose and maintain focus along the plane of the telephone dial. With a view camera, every lens is a tilt/shift lens, whatever the focal length, whatever the age.

Beautiful image, would you consider a Sinar P as special hardware based on today's market?

David

[goamules]

I'm so glad we're getting some more LF examples. I'm becoming more satisfied with what LF can do 'easier.'

You may notice that I didn't qualify our comparisons or put any parameter on what is "easier". The discussion has gone where it is. So I guess at this point we could discuss things like

- Cost. Depends on who you talk to. I can shoot 4x5 in a Graflex and do small contact prints for probably less than my Fuji X-e1 Digital camera cost. 8x10, more. But you don't use the tools the same way. With LF, you might shoot only 4 shots a session. With a DSLR you might shoot 20. Because you can.

- Required skills. This is harder. 35mm film requires similar skills as LF. Digital totally different. I'd say both have a steep learning curve for someone that want's to do thoughtful photography. When 35mm had local 1 hour shops, both it and digitals are about as easy to "show your pictures." But do do it well, all forms take some skill.

- Associated hardware. With LF you need plastic pans, a lightbulb, a dark room. With 35mm, you need an enlarger. With digital you need a decent computer, Photoshop/GIMP software. Etc. I'd guess both require some trouble to set up for, but one will cost about $1,000 or more in the computer/software.

- Special hardware or native ability? Lots of people talk about a small format shift/tilt adapter. Few people have ever heard of one or use one. Almost all Large Format people use shift and tilt. "easier."

[barnacle]

Unless I'm doing the math wrong, I'm not sure I understand this statement.

6,000 x 4,800 = 28.8 megapixels... but 36mp, 42mp, and 50mp 35mm-size sensors (without a low-pass filter) are widely available. How do you get to "four times the limit of a top end electric sensor" at only 28.8 megapixels? 28.8/4 = 7.2 ... implying that those 36, 42, and 50mp sensors are only resolving 7.2 megapixels of real resolution?

From my own experience, if I want to get to 4x of what my D800 can produce, I need to scan 4x5 at 3000dpi, at least. Usually 4000.

I may have been flippant, but with an element of truth: the scan from a scanner has either three separate red/green/blue sensor strips, or illuminates a wideband sensor with red, green, and blue light sequentially. A scanned image is (usually) therefore at the specified pixel: each pixel of the image is the detected value in RGB, or the scaled sum of RGB for monochrome. If scanned at less than the maximum resolution of the sensor, it averages a number of pixel sites, so things get complex, but simple for integer multiples.

A 2-d image sensor almost always has a colour filter in front of it - in a pixel square, two green filters, one red, and one blue. You can get away with this because the eye is not so sensitive to colour detail as it is to overall luminance detail (and this is why colour TV works, as well as jpeg and other perceptual coding systems). But... this means that if you are deriving a luminance value for each pixel, as you don't have a luminance value directly available, you can only get a luminance value for the filter block as a whole. It's kind of vague, since the green, to which the eye is most sensitive and which supplies around 60% of the luminance information - but there are only half of many of them as there are pixels: the best you can get is half the resolution you'd expect from the number of sensor elements.

So if you have a 30MP camera - say 6000x5000 pixels, close to the scan I made, you have in effect only 3000 x 2500 image sites: there are differences across the image site, and adjacent pixels won't be the same level, but you're not seeing the true luminance *or* colour in anything greater than a quarter of the quoted resolution.

As for image sensors without low-pass filters: they're broken by design. This is basic signal processing theory: you cannot represent in a sampled system any signal which is greater than half the sampling frequency, and if you do try and sample you get a thing called 'aliasing' in which higher frequency information appears as lower frequencies. In spite of what manufacturers may claim about signal processing, you *cannot* remove aliased signals; you *must* ensure that no frequency information enters the system. Only if the lens is unable to resolve to no better than half the resolution of the sensor can you do without the low-pass optical filter.

Apologies if I'm lecturing to a thousand DSP afficionadoes here; my experience is that on the whole people don't understand digital systems!

Neil

[BBW]

When I do work for others, I use digital BUT when I have the time and motivation, I still gravitate towards my LF (4x5, 5x7, 8x10). I have friends with all the fancy digitals, ask why I bother with such archaic equipment. I shrug my shoulders, smile, and show them my 8x10 contacts. I admit, you can get digital to be close, or even better, depending on your skill set, but for me, the tonal range I get with the contact hasn't been matched by digital yet. Came close once for me, but then I was using an older Epson 4000 Pro printer loaded with a set of Cone BW inks.

FWIW, you can argue for both, but I like the SLOOOOOW pace of LF. I like BW and the tones I get. As for the skill set versus equipment comment a ways up, for me, I find if I have the vision for a picture, hopefully my equipment will allow me to realize in a much more fluid action. No doing it another way just because the equipment doesn't allow (yes, I like TS on the smaller formats but its seems so much more elegant with a large image and bigger knobs . Though it does help justify GAS

[Ken Lee]

Beautiful image, would you consider a Sinar P as special hardware based on today's market?

Yes: I suppose that all analog photo equipment and materials are special today. They are not sold through the usual channels. Much of my equipment is no longer manufactured.

The same could be said about musical instruments made by Stradivarius

As other have pointed out, for 99% of images, newer equipment can be used. It just so happens that for some small percentage of cases, LF makes it easy.

We could easily come up with cases where the reverse is true, like sporting events or wildlife photography: for those applications I'd leave the Sinar P at home.