Scan size and quality surprise (Epson V500)

I’VE ALWAYS ASSUMED that the quality you get from making a large scan and then downsizing it is going to be much the same as just making a small scan to start with. I now know that it isn’t.

As it takes several minutes to scan a medium format negative at 3,200lpi (with digital ICE dust removal turned on), I’ve generally preferred to use a lower resolution.

Imagine my shock, then, on discovering that the smaller scans make grain FAR, FAR worse than it really is and introduce all sorts of compression artifacts.

Here’s the proof. These scans were made from a single negative at exactly the same settings. Just the LPI was changed. To show just how huge the difference is, took screen shots at three times the “actual pixels” size of the resuling scans.

800lpi scan, at 300%:

3,200lpi scan, downsized to match the dimensions of the 800lpi, then viewed at 300%

And, just for comparison, here is what the original 3,200 scan looked like at 100% without enlarging:

The scan actually looks less grainy straight out of the 3,200lpi scan at 100% than it did from the 800lpi one at 100%.

THE LESSON from this is obvious: in order to get the best quality out of the V500 (and presumably other scanners) one MUST scan big and then downsize. With a 6x6cm negative, a 3,200lpi scan means an output file of around 120MB (40MP), so to keep the memory usage under control I need to downsize all the big scans straight away.

I still can’t believe that reducing the file size increases the grain size – but it does, and I have heaps of files that need rescanning because of that.

****

As I said earlier, when scanning I always pull out the end “input” sliders in the histogram to retain all the info, rather than clipping off the whites (which the scanner does in auto mode). The Output sliders then need to go to zero and 255 to make the blacks black and the whites white. This generally creates a dull image and the midtone input slider needs to be moved down a bit to brighten it, leading to colour distortion.

This is where the midtone grey eyedropper comes in handy: activate the dropper and click it on a neutral grey part of the image and it resets the white balance. If the image is still dull, a little tweaking of the brightness and contrast sliders can help a lot – and for some reason they do not affect the “input” settings on the historgram, so they don’t add noise (which trying to adjust the input can do).

Using this method, one can get results that need very little subsequent editing, other than dealing with dust and scratches.

“f/8, and be there!”

This, supposedly, was the mantra of the Press photographer. f/8 would cover a multitude of sins and as long as you were there you probably got something that would be good enough for the paper. Those who remember the newspapers of the 1960s or 70s will recall how small photos were normally used. It wasn’t exactly challenging to get something that would look OK printed no bigger than the size of a 4×5 negatives.
However, in theory the old American Press cameras are capable of even better results than the famed Hasselblads or Bronicas.
For this shot, I used my Graphlex in “Press camera mode”, hand-held and focusing using the built-in rangefinder, which is set up for a 135mm lens, the Schneider Xenar which presumably came with the camera when it was made.
As the food carts were in the shade, f/8 was appropriate, with a shutter speed of around 1/150 (allowing for the slightly slow shutter). I lost the side of the negative with a slip of the dark slide, and the old Xenar seems to have picked up a bit of flare on the right-hand side but the focus is exactly where the rangefinder put it and the image is as sharp as it gets.
f/8 and be there is better than just about good enough!

Oh, and “foul” is not self-criticism of the food they serve, it’s the name of a popular Egyptian bean soup.

A version of this photo is for sale as an art print here

Souq Waqif, Doha

It’s taken a long time but I guess this image says I’m finally getting up to speed with the large format camera.

Souq Waqif in Qatar, shot on Feb 13, 2013. Graflex Crown Graphic 4×5 camera, Rodenstock APO-Symmar 150 lens (UV filter), Fomapan 200 film, 1/125 @ f/16, developed 6.5mins in Fomadon Excel standard solution.

A toned version of this photo is available for sale as a print here:

Primes, zooms, landscapes and panos

THE question cropped up in a forum discussion of whether primes are better than zooms. This is a fairly common question amd my response is that the best modern zooms are very nearly as good as any prime. My 70-200 f/2.8L is as sharp as anything I’ve got at all focal lengths and wide open.

The main advantage of primes is that they are usually a LOT smaller and easier to carry and they are usually a little bit faster, with apertures of f/2 or less at focal lenghts of less than 100mm, which helps in low-light situations or where you want to isolate your subject from the background.

Today’s lenses all seem to deliver much the same in terms of clinical optical perfection. However, every now and again you may come across and old lens that just has a feel to the pictures that others don’t. I get that with my Leitz Summicron 90, which I featured in my last post. It is a late 1970s lens (and still costs a packet today – in fact, more than it did a few years ago when I got mine)

***
Another question that came up was what would be the best lens for landscape photos?

I’m not sure if there really is a “best”. Perhaps there is, in terms of focal lengths, if the question is about what is most likely to be useful most of the time. But, then again, that depends on an individual’s own artistic style.

A fairly normal lens for landscapes would probably be a medium wide-angle – 35mm for full-frame or a 24mm for crop frame (or a good zoom that covered those ranges). However, for some types of landscape, particularly panoramas with little or no motion in the foreground, an excellent alternative is to use 80-90mm (FF) or 50mm (crop frame), hammer off a quick succession of framse and stitch the images together in photoshop.

Like this:

I would have needed a super wide-angle to get all of that into a single frame. It was shot with a 24-70, zoomed to 70, I think. Doing this, you end up with a super-high resolution image, which is super-sharp (but sometimes the frames won’t join together properly and the hoped for image goes to pot).

There may also be times when you want to use a telephoto lens instead of a wide-angle for a landscape. This was shot with a 180mm prime lens on an old Mamiya twin-lens-reflex camera (probably equivalent to something like 100mm on 35mm full frame) and it really pulls in the snow-capped mountain far beyond the town:

And here’s another, also shot with a telephoto (and worth clicking on for a larger view)

Will digital do this?

I recently finished processing this shot, taken with my Leica R4s ModP and the Summicron 90mm lens, and it struck me as a fine example of the sort of quality in the tones that film can deliver and but digital doesn’t seem to match – even with the best digital cameras.

Am I right, or am I imagining it?

Slaving over a hot camera

I’ve been busy this month, slaving over a hot stove while I try to crank up some action on my food blog, http://www.gourmetlens.com.

Unfortunately there just don’t seem to be enough hours in the day to do everything I would like to do but for the moment if you want to keep tabs on the progress of my food photography (and try your hand at some intriguing recipes) you can follow me over there.

Which is not to say that I won’t be back here before long with some more offbeat observations about photography.

Unlucky ’13!

WHAT a start to the year!

I decided to take a set of photos of my old cameras, all shot in black and white with medium format film. I thought they might have a chance of selling as wall art.

To get convenient magnification and depth of field I decided to use the Mamiya C220 with the famously sharp 105mm “blue dot” Sekor lens, which is equivalent to about 80mm in 35mm format. It was to be the first time I had used this with studio flash.

I should have tested it, of course. After flashing cheerfully through 12 frames of Fomapan 120 (carefully setting up the image and then winding the tripod head up 5cm so the taking lens was in the place where the viewing lens had been) and waiting the best part of an hour while it developed and fixed I found myself gazing at a nice, blank sheet of acetate. The X-synch flash must have been out of kilter, probably opening the lens on the old M-flash setting.*

Next stop was the Pentacon Six and my last roll of 120 black and white film. This time, everything went smoothly in the darkroom but when I came to load the film – Ilford FP4 – into the spiral, humidity inside the changing bag from my arms and probably some residual dampness in the spiral meant the film jammed every time I tried to load it.

I’m sure every amateur film developer has met the problem. You slide the film into the spiral (I use a Paterson tank) and then as you start to wind it in, it jams. Keep winding and it starts to buckle and then comes out of the groove.

All I could do, since I had torn part of the backing paper away, was dump the film roll in the tank, put the lid on and then, once it was safely in a dark cupboard, chuck a small sachet of silica gel in with it. To my surprise, the film came out nice and dry the next day and loaded into the spiral as if nothing had ever been wrong.

Developing for 10 minutes in partially used stock solution of Fomadon-P once again delivered a negative that was thinner than it should be. Why? I follow all the instructions exactly for temperature, agitation etc., but I constantly get thin negatives from Foma developers unless I more or less double the recommended times. And Fomadon-P or Fomadon Excel are the only developers I can get at present.

Next to the Epson V500 scanner, which for some reason hasn’t been looking at 120 film properly, slicing the preview up into 35mm shaped bits. I decided to try to correct the problem by reinstalling the driver, which promptly corrupted two start-up files in the computer so that for the second time in about three months I’m stranded with a crippled computer and am writing this in Windows “Safe Mode”. Time to upgrade to a new machine, unfortunately.

Here’s hoping this is not a foretaste of what the rest of 2013 will be like!

Right now, I can’t offer you an up-to-the-minute scan, so here’s a shot of Doha Bay on Boxing Day last year (or last week).

*The difference between X-flash (the name comes from the xenon gas used in flash tubes) and M-flash (for the magnesium powder or wire used in the old days) is that the electronic flash fires faster than the magnesium can ignite. To allow for that, M-synch opens the shutter a fraction of a second later than X-synch, which means the electronic flash is over before the shutter begins to move. So using electronic flash with a shutter set on “M” will leave the film unexposed. It’s possible to get round that in a dark studio by opening the shutter on B and then firing the flash manually.

Stitching up wide-angle

EVER wanted a perfect, distortion free, 10mm lens with a camera that delivers 70MP images? Well, if you’ve got a half-decent DSLR and a reasonable standard or short telephoto lens, then you have all you need to replicate the results you would get from a system like that.

Here’s one example of what the result might look like (though I had to downsample this to about 25MP in order for the file to be small enough to be accepted on Fine Art America – hardly anywhere will accept 70MP files since they are way too big for almost any practical use)

My Doha, Qatar, panoramic photos are available for commercial use on Alamy, here: http://tinyurl.com/apjvhol

The trick to making huge panoramas is very simple. All you have to do is put your camera on manual settings (including manual focus), set the aperture and shutter speed to get a correct exposure and then, standing in one place quickly take a series of overlapping photos. For best results, have your camera mounted on a tripod.

Once the pictures are on your computer just open them in Photoshop and use the File>Automate>Photomerge command to stitch them together. Or, better still, download the free (yes, I said FREE) Hugin program from Sourceforge and use that. There’s a fairly sharp learning curve to Hugin (which I don’t plan to go into here – there are probably YouTube videos out there somewhere to explain it) but the results it gives are far superior to Photoshop CS4, which is the most recent version of PS that I have.

I find that a short telephoto lens is best for stitched panoramas. The immediate foreground is the part that is likely to cause most problems in stitching because, being closest to the camera, it suffers most from parallax effects. A telephoto cuts off the foreground and encourages you to stand further back from your subject.

Take your pictures in quick succession – as quickly as you can. Movement in the scene is the enemy of stitching, especially movement in the foreground, so don’t give waves or boats or people time to move very far. And it is vital not to change the exposure or focus during the shots (which a camera set on auto will do).

After stitching, check closely for poor blending – lines down the sky, for example, where two shots have been lined up – or for blurring where something that moved has been blended together in an awkward way. It seems that Photoshop minimizes the area that it blends, which can lead to lines in the sky, for example, whereas Hugin does skies much better by blending over a wide area but may do odd things to moving objects like flags because of the wide blending (whereas PS seems to choose one image or another and just use the flag from that). At least, that is the impression I have got from using both programs.

Sometimes, straight lines – such as corners of buildings – may not line up correctly, particularly if you were close to it – leading to obvious flaws in the final image with a wall ending in mid-air. It may be possible to overcome that by using different stitching projections – fisheye, or rectangular, or barrel, PS has a few to choose from, Hugin has a pile of them. Unfortunately, sometimes stitching just won’t work perfectly (but if you do have a 30 pixel stitching flaw visible on a 100MP image, that will only be half a dozen pixels out by the time it is reduced to 20MP, which may not matter).

Overall, however, a stitched panorama made with ten shots from a cheap, short telephoto lens will be of a far higher quality than one made with the finest Zeiss Distagon that costs thousands. Why? Simply because all the lens flaws are diminished in the stitching process.

If you have 30% overlap on 10 images, you end up with a stitched image that is about five times wider than a single shot taken with a superb wide angle lens. Say your cheap lens resolves 30 line-pairs per MM and the Zeiss resolves 80 lp/mm (which is about all any digital sensor can manage, anyway), by the time you shrink your stitched panorama to the width of the Zeiss shot, the resolution becomes equivalent to about 150 lp/mm.

In other words, the huge size of the stitched image more than compensates for the inferior quality of the lens. This is precisely the way large format photography works: The lenses are not as sharp as the best 35mm camera lenses, but the huge negatives overcome that.

Lens edge distortion is also overcome with stitching. Wide angle lens designers have to struggle with all the problems that come from bending the light rays to fit onto the sensor. Soft edges and (especially) blurry corners have always been a problem with such lenses. But with stitching, you just point your camera straight on at the bit that will be on the edge of the frame, so distortion is not an issue (it may be a good idea to shoot a frame or two more than you want to give the stitching program more to work with on the sides).

So that’s it: “Photomerge” stitching is not a poor-quality alternative to having an expensive lens. Quite the opposite, it is a way of overcoming the quality problems inherent in all wide-angle lenses and producing higher-quality images. It’s a technique that should appeal particularly to digital landscape photographers, whose subjects are extensive and static. Of course, stitching works to make images with ordinary aspect ratios as well as wide strips. All you have to do is shoot the scene in several layers.

Ringing the changes

IT CAME as a shock when these giant bells suddenly launched into a deafening clangour while I was photographing from the tower in St Mark’s Square, Venice, but it immediately struck me that this was a rare opportunity to represent sound in the photographic medium.
The lighting was awful, with enormous contrast between the windows and the gloom of the belfry and I wanted a slow enough shutter speed to leave some motion blur in the swinging clapper.
The final result is a bit less refined than my normal studio or landscape shots, but it has immediacy so I am happy with it.

The Best of 2012

IT’S hard to say which of this year’s crop of photos are my best, particularly as I am usually most excited by what I have done most recently – and if my work is steadily improving, that is as it should be. Anyway, here are a few, I hope you enjoy them (they are all my copyright and not to be reproduced without permission, of course):

It’s a reflection of my eclectic approach to photography that this selection includes black and white and colour; small, medium and large format film photos as well as digital; a flower macro and “stitched” panoramas.