Archive for the ‘Technical’ Category

Cold Harbour & Myatt’s Fields

Wednesday, May 31st, 2023

My walk on Sunday 9th April 1989 continues in this post, Cold Harbour & Myatt’s Fields. The previous post was Camberwell & Myatt’s Fields.

Church, Shop, Coldharbour Lane, Camberwell, Lambeth, 1989 89-4j-61
Church, Shop, Coldharbour Lane, Camberwell, Lambeth, 1989 89-4j-61

Back in 1989 I was still pretty pressed for cash, still buying film in bulk 100 ft lengths and loading it into cassettes myself in total darkness.

Over the years I’d perfected my method. Two nails on the back of my darkroom door, hang one of the sprocket holes at the end of the roll of film on the top one, unroll it down to the second, cut across, replace film in can. Pick up first spool from a waiting row on the bench, already with a short length of masking tape on it, attach to the bottom end of the hanging length of film, carefully roll it up to the top, remove from nail, pick up cassette body, insert spool with film end though velvet light trap, pick up end cap and pinch cassette to push it into place. Repeat another 18 times until the film roll is finished. Turn on light, trim film ends to fit cameras and put into plastic pots to go into camera bag.

Slow, tedious but then less than half the cost of buying film in 26 exposure cassettes, though I did ocasoinally treat myself – and if I bought Ilford film rather than Kodak I could reuse the cassettes with bulk film. Kodak had crimped on ends which had to be removed with a bottle cap remover destroying them.

I had learnt to be very careful with film in this project to photograph London, working with 35mm cameras much as I would have done with large format camera, carefully considering various viewpoints before deciding on an exposure. But when working with people I had to respond rather more quickly, and seeing these two men in front of the white church door my response was immediate.

Shops, Coldharbour Lane, Camberwell, Lambeth, 1989 89-4j-62
Shops, Coldharbour Lane, Camberwell, Lambeth, 1989 89-4j-62

After that rather snatched image I continued with photographing the two shops which had attracted my interest here, making first a vertical image and then moving back across the road for a wider view. In this (below) you can see the notice for the Celestial Church of Christ and the alley leading to this.

Shops, Coldharbour Lane, Camberwell, Lambeth, 1989 89-4j-63
Shops, Coldharbour Lane, Camberwell, Lambeth, 1989 89-4j-63

As well as these black and white images I also took a couple in colour which you can see in another album. I had two Olympus OM bodies with me and only brought the one with colour negative film out of my bag occasionally, while the black and white camera was usually on a strap around my neck. When I was intending to photograph people rather than buildings I usually went out with a Leica M2 instead.

Coldharbour Lane leads from Camberwell to Brixton and got a very bad reputation after the 1981 clashes between police and locals in Brixton. In 2003 it was called in an article in the London Evening Standard the most dangerous street in the most dangerous borough in London, but that was lagely rabid tabloid journalism. Wikipedia gives several theories about its name none of which seem entirely convincing, but the name seems often to have been associated with the ruins of Roman or Romano-British settlements, The area we know usually call Loughborough Junction around the station on early maps was called Cold Harbour. Coldharbour Lane was then known as Camberwell Lane.

Cafe,  Coldharbour Lane, Camberwell, Lambeth, 1989 89-4j-65
Café, Hinton Rd, Camberwell, Lambeth, 1989 89-4j-65

The S&J Corner Café was on the corner with Wellfit Street, close to Loughborough Junction station. The railway line here is just south of the station. A second bridge can be seen going above this and the cafe which is the line from Brixton to Denmark Hill, now used by London Overground services.

Cyclist, HInton Rd, Camberwell, Lambeth, 1989 89-4j-51
Cyclist, Hinton Rd, Camberwell, Lambeth, 1989 89-4j-51

Another picture I took on the spur of the moment as I saw a cyclist coming towards me under the bridges on Hinton Road. I was standing on the pavement beside the cafe in the previous image, and the cyclist is on the pavement, rather safer than roads like this in London. At the end of the row of shops on the left are the traffic lights and Coldharbour Lane whch I had just begun to walk towards.

Hinton Rd, Camberwell, Lambeth, 1989 89-4j-52
Hinton Rd, Camberwell, Lambeth, 1989 89-4j-52

The pub at right of this picture is The Green Man on the corner with Coldharbour Lane. According to a post on the Brixton Buzz, this had been on the site since 1881, but that is the date of the current building which replaced an earlier pub on the site important enough to be marked on Stanford’s 1862 map. The Buzz says it was closed in 2003 because of drug dealing and crime, and it quotes from the Urban75 blog that it was “was frequented by dealers (crack, heroin you name it), prosi’s and general madhatters.

The buildings at the left date from around the same time as the area was developed around the railways, and number 6 at left has a barely legible road name ‘Hinton Terrace‘ and I think at the top the illegible name of a builder and decorator whose sign presumably once hung on the bracket beside the street name.

House, Lilford Rd, Camberwell, Lambeth, 1989 89-4j-53
Houses, Lilford Rd, Camberwell, Lambeth, 1989 89-4j-53

I walked up past Loughborough Junction and made my next picture on Lilford Road at the corner of Minet Road, returning to the area I had been earlier on this walk, the Minet Estate around Myatt’s Fields. This is on the corner of a terrace with basement flats with an entrance here under the steps which extends along both streets and this grand entrance is actually for two adjoining houses above the flats, one on each street. The Grade II listing calls these “Early-mid C19″ and describes this a “double prostyle composite porch with fluted composite columns.”

Longfield Hall, Knatchbull Rd, Myatts Fields, Camberwell, Lambeth, 1989 89-4j-54
Longfield Hall, Knatchbull Rd, Camberwell, Lambeth, 1989

William Minet founded this Grade II listed community hall, architect George Hubbard, which opened in 1889, as well as the Minet library opposite which was destroyed by bombing and rebuilt in 1956 in what Pevsner described as ‘a meek replacement’. The Library was a memorial for his late wife. The Hall is still in use for various community activities including three church congregations and from 1969 -1975 was the base of Britain’s first Black Theatre Company, ‘Dark And Light’, recently marked by a Blue Plaque. It was Grade II listed in 1979 and is now run by a charity, The Longfield Hall Trust.

My walk made on on Sunday 9th April 1989 will finish in a later post. The first part from it is at Peckham and East Dulwich 1989.


FlickrFacebookMy London DiaryHull PhotosLea ValleyParis

London’s Industrial HeritageLondon Photos

All photographs on this page are copyright © Peter Marshall. Contact me to buy prints or licence to reproduce.


Wandsworth Panoramas – March 2014

Friday, March 24th, 2023

As a photographer I’ve long been interested in the difference between how we experience the world around us and how the camera records it. Some of those differences are obvious but others less so, and some we are seldom aware of.

Wandsworth Panoramas - March 2014

The camera records an image produced by its lens which follows strict optical rules which I learnt about long ago in my physics lessons, though real lenses deviate slightly from those ideal and perfect specimens in those science texts.

Wandsworth Panoramas - March 2014

The camera holds a film or sensor to record that image – and again does so following strict physical (and chemical for film) processes which may fail to record significant features and distort others to produce an essentially flat two-dimensional image. It may not even record colours but if it does they always to some extent arbitrary, as too are the tones.

Wandsworth Panoramas - March 2014

Those of us who grew up on film are perhaps more aware of this than the digital generations. We had to be aware of the differences in recording of, for example Ilford’s Pan F and Kodak’s Tri-X, and how these were affected by processing and printing, and of the rather unreal but different colour renditions of Kodachrome, Kodacolor, Ektachrome, Agfa, Ferraniacolor and the other colour film films, each with its own qualities. Though perhaps if we ever used Orwo film quality was not the right word for its purplish nature.

Wandsworth Panoramas - March 2014

Of course there are differences in the way digital cameras record colour, but these are rather smaller, and we can make use of software to make them match more closely or exaggerate the difference. Lightroom and Photoshop can make my Fuji files look very similar in terms of colour rendition to those from Nikon.

But our experience of a scene is very different, combining inputs from all of our senses, and it would be impossible to over-emphasise the subjective aspects. But even just visually it is still very different. While the lens cuts out all but a small rectangle in front of us, our eyes send information to the brain from a much wider field, much of it except from a small central section lacking in sharpness. Most of us have binocular vision, gathering this data from two eyes a short but significant distance apart, enabling us to see in depth. And our view is always dynamic, our eyes moving around, and as we swing our head around or up and down we have the sensation of moving through a static universe. Doing the same with a camera has a very different effect.

A standard lens – around 40 to 50mm on a full frame digital or 35mm film camera gives a similar idea of depth in its flat images to that we normally experience. With longer lens the effect of depth is reduced and by the time we get to really long lenses the images become flat patterns rather than appearing to represent a three dimensional scene. But what interested me more was what happened when the camera tried to represent a much wider angle of view than the standard, when the rectilinear rendering of normal lenses becomes impossible.

On Monday 14th of March I went for a walk with a painter friend who had brought her sketch book to introduce her to an area I thought she might find interesting. And I wanted to further explore some of the different ways of rendering very wide angles of view with digital cameras. I’d brought two Nikons with me, one fitted with a conventional wide-angle zoom which I used mainly at 16mm, close to the limit for such lenses (and I do have a wider lens which demonstrates this) and the other with a 16mm full-frame fisheye which fills the frame with an image which is 180 degrees across the diagonal.

While my friend stopped to make sketches I had time to make a series of images from similar locations. I kept warmer as I was moving around, but she fairly soon got cold, which was a good excuse to visit the pub which appears in some of these pictures, after which I took her back to the station where we had met and went back to take some more pictures on my own.

Back home I uploaded the images. Those from the conventional wide-angle zoom I’ve use as they were taken, with just the normal adjustments in Lightroom. But the fish-eye images I worked on with my panorama stitching software, PtGui, not to join images but to take the raw image data and process it it various different ways to produce cylindrical projections. If the camera was upright when the picture was taken, this will produce straight vertical lines for all upright elements. There are many different approaches to this which produce visually different results, some of which are common in mapping, such as Mercator.

Those I’ve found most useful are the equirectangular, Vedutismo and Transverse Vedutismo projections used in these examples.

More panoramic images from my walk on My London Diary at Wandsworth Panoramas.


FlickrFacebookMy London DiaryHull PhotosLea ValleyParis

London’s Industrial HeritageLondon Photos

All photographs on this page are copyright © Peter Marshall. Contact me to buy prints or licence to reproduce.


Chinese New Year 2005

Sunday, February 13th, 2022

Chinese New Year 2005

On Sunday 13th 2005, 17 years ago, London was celebrating the Chinese New Year of the Rooster which started the previous Wednesday – it was 4072.

Chinese New Year in Soho is something I’ve avoided in more recent years – as I wrote in 2005: “I used to enjoy the rather anarchic celebrations in Chinatown, but it’s now more of an ordeal, with far too many people coming in to watch and too much organisation.”

Trying to photograph in such crowded situations was a problem, and one I confronted in two main ways in 2005, something reflected in the two pictures above. At the top is a picture taken standing back some distance with a telephoto lens, while the lower picture is taken with a fisheye lens, both on a Nikon D70 DX camera.

De-fished version

Usually now when I use the a fisheye lens like this, I would convert the perspective to give straight verticals – as in the above image. But back in 2005 I didn’t have a good plug-in to do this conversion, and although it was possible with various programmes I was using for making panoramas it was a rather time-consuming process.

For this particular event I rather liked the fisheye effect, at least in some pictures. Although it does clearly misrepresent those faces close to the edges of the picture, for me it pulls the eye towards the centre of the picture and perhaps gives a greater impression of the crowding I was working in.

A small problem is that the image you see in the viewfinder is the fisheye one, and not that in the ‘de-fished’ version. But as you can see, the fisheye image which you see has the same horizontal limits at the centre of both the horizontal and vertical sides, with just a little of the image towards the four corners being lost. It’s still possible to frame accurately when working.

It’s not I think correct to call the effect of the fisheye lens ‘distortion’. It is simply a different way of recording the subject on a flat rectangle. Most fisheyes I’ve used (and I own four different examples, for DX and full-frame Nikon, for Fuji and for micro 4/3) seem actually to have rather less actual distortion than my ultra-wide rectilinear (i.e. ‘normal’) lenses.

In the de-fished image you can see that as well as the verticals of the building being straight, people at the edges of the picture are also shown naturally, unlike in the fisheye version. I was also taking some pictures with an ultra-wide 12-24mm lens (equivalent to 18-36mm full-frame) and with that at its widest faces at the edge would have been rendered a little stretched out horizontally.


I’m not sure what some major agencies would make of conversions using software like this, whether they would regard it as an unacceptable alteration of the image. For me its just one of many acceptable corrections of the image, but clearly it does alter the image as recorded by the camera. It would be possible to design a specialised wide-angle camera which carried out the correction in firmware but the market for this would probably be small. Rather it could be provided into normal digital cameras as an option – far more useful than all those special effects which clutter the menus on many cameras now.

More pictures on My London Diary – scroll down a little from the top of the page.


FlickrFacebookMy London DiaryHull PhotosLea ValleyParis

London’s Industrial HeritageLondon Photos

All photographs on this page are copyright © Peter Marshall. Contact me to buy prints or licence to reproduce.


Digital Panoramas on the Thames Path

Wednesday, January 5th, 2022

Digital Panoramas on the Thames Path
I’ve long had an interest in panoramic photographs, both in taking them and also appreciating the work of well-known photographers who have made panoramic images. From the earliest days some photographers wanted to make pictures with a wider field of view than was possible with a normal camera and lenses, and the first patent for a specialised panoramic camera was filed in Austria in 1843, using a curved Daguerreotype plate and rotating lens.

The earliest existing panoramic photographs appear to be those by Friedrich von Martens made in the early 1840s – such as this example on Wikimedia dated from 1846. There are also paper prints from the same era, presumably made from calotype negatives. As well as making single exposures with an angle of view of around 150°, von Martens and others made panoramas using multiple exposures, often with normal lenses. Martens produced what was probably the first 360° panorama using three curved Daguerreotype plates.

Normally we use cameras with rectilinear lenses to render straight lines in the subject as straight lines in the picture. But as the distance from the lens centre to the film or sensor gets longer towards the edges and corners, the image magnification also increases. This begins to be noticeable with extreme wideangle lenses, although more of a problem with some subject matter than others.

Although I’ve worked with a full-frame lens at 12mm, I’ve found that for general purposes a practical limit is around 15-16mm with 18mm generally more useful, corresponding to an horizontal angle of view of 90°. Beyond that the image stretching usually becomes too noticeable.

The first really popular specialised panoramic film cameras were the 1899 #4 Kodak Panoram and the Circut, patented in 1904 and produced in a range of sizes until 1945. Some were still in use until recently for producing long roll photographs of perhaps 800 pupils sitting in rows on the school field. They rotated slowly enough for some students to run around the back of the group and appear at both ends. Cameras of this type were used to great effect by photographers including Josef Sudek.

Having made several multi-image panoramas and found the process limiting I bought my first rather more modest panoramic camera, a Japanese Widelux taking images on 35mm film in 1991. Later I bought a Russian Horizon which gave similar results, and a 120 format Chinese model. I still have these along with a Hasselblad X-Pan, not really a true panoramic camera, but using a panoramic format – with the standard lens it only gives a similar angle of view to a 28mm lens, and even with the 30mmm wideangle I mainly used only around a 90° angle of view.

These cameras were the main reason I continued using some film after going digital in 2002. But some years later I found a way of working with digital cameras to make panoramic images, using a fisheye lens and then ‘defishing’ this with software to give a similar image to those made with the swing lens cameras.

These pictures were taken seven years ago on a short walk along one of my favourite sections of the Thames Path in London, from Vauxhall to Wandsworth on Sunday 5th January 2014.

I took images handheld with a Nikon D800E using a Nikon 16mm f2.8 fisheye lens, and later converted them using an Equirectangular projection in PTGui software. I now generally use the more convenient Lightroom Export plug-in https://www.imadio.com/products/prodpage_hemi.aspx ‘Fisheye-Hemi’ from Imadio.

You can see larger images and many more from the walk at Thames Path Panoramas on My London Diary.


FlickrFacebookMy London DiaryHull PhotosLea ValleyParis

London’s Industrial HeritageLondon Photos

All photographs on this page are copyright © Peter Marshall. Contact me to buy prints or licence to reproduce.


Bow Creek Panoramas – 1992

Thursday, February 18th, 2021

DLR Viaduct, Bow Creek, Leamouth Rd, Leamouth, Tower Hamlets, Newham, 1982 92-1j61pr_2400

At the end of 1991 I finally bought my first panoramic camera, a Japanese Widelux F8 which I couldn’t really afford. It was a camera that took around 21 pictures on a 36 exposure film, with the film curved around a part of a cylindrical path with the lens pivoting around the centre of the cylinder.

Bow Creek, Bridges, Leamouth Rd, Tower Hamlets, 1992 92-3a42a_2400

The lens is a 26mm f2.8, though it needs to be stopped down to around f8 for most pictures as the camera is fixed focus at around 6 ft and only gets sharp at infinity when stopped down. Winding on the film winds up a clockwork motor and returns the lens to its starting position. On pressing the shutter the lens swings around through about 140 degrees, exposing the film through a slit at its back which swings across close to the film. It has 3 shutter speeds, 1/15, 1/125, and 1/250th, but even at the fastest speed it still takes rather longer to actually complete the exposure.

92-3b38_2400

The design keeps lens to film distant constant – at around 26mm – right to the edge of the film across a negative 24mm x 56mm. If this was flat, the distance to the corner would be more like 40mm and so objects at the edges get stretched to around 1.5 times actual size. This camera eliminates this distortion, but at the expense of introducing its own which you can see in these pictures. This becomes particularly noticeable in the curvature of most straight non-vertical lines.

Pura Foods, Bow Creek, M & J Reuben, London Sawmills, Wharfside Rd, Newham, 1992 92-3b52_2400

In particular, horizons become curved unless the camera is kept absolutely level. The pictures of Bow Creek were made with the camera on a sturdy tripod and with the help of a spirit level. There is one on the top plate of the camera, but I found a larger separate one more useful.

Bow Creek, Orchard Place, River Thames, Lower Lea Crossing, Tower Hamlets, Newham, 1992 92-1n12_leamouth_2400

Although the angle of view is often stated as 140 degrees this is perhaps misleading and I think probably is the angle across the diagonal. Rather more useful is the horizontal angle of view, which I think is just slightly over 120 degrees. Theoretically it would be possible to create a full 360 degree view in three exposures, but practically it needed four, though I don’t think I ever succeeded on the few occasions I tried to make one.

Later I made many more panoramas here and around London, particularly with a similar Russian camera, the Horizon or Horizont which was rather more convenient to use, as well as a few with a much larger medium format version. I also used a Hassleblad X-Pan, a nice camera which was panoramic only in format, with a similar negative size, 58x24mm, but using rectilinear lenses which can’t acheive a really wide angle of view.


Bokeh Bunkum

Friday, November 13th, 2020

I’ve never really understood the hoo-hah about bokeh which has become far too dominant since photography came on the web; it really was not an issue, not even a term photographers had heard of before 1997 and the blame for its introduction can be clearly laid with Mike Johnston, then editor of Photo Techniques (but now author of The Online Photographer blog) who published three articles on it in May of that year, as well as changing its spelling from the Japanes ‘boke’ to reflect its pronunciation for us anglophones. And it was on his blog that I found the link to the video Bokeh Is Overrated by Andrew on the Andrew & Denae channel.

Bokeh is a term to describe the different rendering by different lens designs of out of focus background areas, particularly highlights. The differences become more obvious at wide apertures and with longer focal lengths. Some designs retain a more wiry core, while others are smoother, though the differences can be subtle. The most extreme example of a bokeh that I almost always find unpleasant is the doughnut shaped out of focus highlights produced by all mirror lenses; typically 500mm or more, their mirror design gives these despite the limited apertures – usually f8.

Bokeh has I think become so popular because it gives people a kind of plug-in solution to producing “better pictures”. You don’t need to think but can simply buy a f1.2 lens and use it wide open for your portraits etc. Lazy photography. And now getting lazier still when you can apply it in software after the event or in camera.

Andrew’s experimental study which he describes in his video isn’t really about bokeh, but about out of focus backgrounds and I think all his relevant pictures in it were taken with the same lens but at differing apertures. But its results still show the largely irrelevant nature of the holy grail bokeh quest to actually making better pictures, pictures that other people and not just fellow bokeh-obsessed nerds will prefer.

While the details of the survey which occupy much of the video are a little tedious, and it clearly – as the video and a disclaimer makes clear – is not a scientifically valid survey, its hard to fault the conclusions and advice Andrew gives in the final section of the video, which I hope will do something to cool the bokeh obsession.

Among his conclusions are that the “strength of a photograph is not measured in terms of background blur” and “what is in focus is always more important than what is not in focus” but there is a lot more that makes sense.

There are, as he says, valid reasons to want fast lenses, mainly to work in low light, where as well as for taking images, with cameras that have optical through the lens viewfinders such as DSLRs they give a brighter image. Of course they are far less necessary than used to be the case with film, when ISO 1600 was about as far as we could push; now we have at least 3 stops more to play with. For static subjects we now have image stabilisation which enables us to use much slower speeds hand-held – and as a last resort there are tripods, though I now seldom need to dust mine off.

I do own a f1.2 lens, a Fuji XF56mm short telephoto, but I’m not sure I’ve ever used it wide open. Usually the depth of field is simply insufficient. Stopped down to F2.8 it gets noticeably sharper too. But usually I’ll trade any advantages of a fast lens for the smaller bulk and lower mass and price of an F2.8 or f4 counterpart – and will take the great majority of my pictures between f5.6 and f11, except by accident.


All photographs on this and my other sites, unless otherwise stated, are taken by and copyright of Peter Marshall, and are available for reproduction or can be bought as prints.


Meridian 2

Wednesday, August 26th, 2020

Continuing with pictures from my walk along the Greenwich Meridian in Greater London in 1984-6.

Stratford Bus Station – Peter Marshall, 1995

My walks took me as close to the line of the Meridian I had pencilled on my 1983 1:25000 OS map as possible, though that line may not have been quite exact. I think it goes through the area at the extreme left of the picture above, here just a few yards east of the roadway. My series of walks kept as close as possible to the pencil line, but it often runs through private property, buildings, across rivers etc and many detours, some quite lengthy were required.

Barge carries contaminated earth from Poplar gasworks site, Peter Marshall, 2011

One of those fairly lengthy detours was north from Poplar, where the line ran through the gas works site and across Bow Creek. It wasn’t until 2011 that I was able to go onto the former gas works site, having been engaged to photograph the use of a barge to carry away the heavily contaminated soil from the site. The line crosses the river here, going through the left end of the large shed close to the opposite bank, near to Cody Dock. This is also part of a private business estate, though you can now walk along the roadways in it. There are several such areas I have been able to photograph in later years, but I won’t add any other later pictures to these posts.

Stratford Station – Peter Marshall, 1995

The line continues through the east end of Stratford Station.

Thinking of the line of the Meridian, I had decided it was appropriate to use a panoramic format, and these pictures were all taken with a swing lens panoramic camera. I think at the time I owned two such cameras, an expensive Japanese model and a cheap Russian one. The Russian was a little more temperamental and it was sometimes difficult to wind on the film, but had a much better viewfinder and I think was probably used for most of these. Both give negatives which are roughly the width of medium format film – 55-58mm – but only 24mm high, the limit of 35mm film, giving a roughly 2.3:1 aspect ratio. There is no discernible difference in image quality.

Langthorne Rd, Leyton – Peter Marshall, 1995

Both used 35mm film and curve it in the horizontal plane around a little over a third of the outside of a circle, with the lens pivoting roughly 130 degrees around the centre of that circle during the exposure. This keeps the distance between the centre of the lens and film constant, avoiding the distortion produced by using flat film, where the edges of the film are further from the lens node. This gives a very noticeable distortion with ultra-wide lenses, limiting them to an angle of view (horizontal) of roughly 100 degrees.

St Patrick’s Cemetery, Leyton

Swing lens cameras are limited in angle of view only by the mechanical limitations and can generally cover 130-140 degrees. But the curvature of the film does produce its own unique view. Assuming you keep the camera upright, straight vertical lines remain straight as the film is not curved vertically, but non-vertical lines show curvature, increasingly so as you move away from the centre of the film. You can see this clearly in the shop window in Langthorne Rd.

Whipps Cross – Peter Marshall, 1985

To be continued…

Against Facial Recognition

Sunday, August 16th, 2020

I’m not sure if you need this. But for some people in some countries it could be very important, assuming that it works. I’ve always been very open on-line, posting only under my real name and everything I post is public. I’ve been careful though only to post things that I don’t mind everyone knowing about me.

As a journalist I’ve had some advice and training on privacy issues, particularly on messaging and e-mail, but haven’t ever felt I was in a situation where I needed to put this into practice. But I do sometimes worry a little about my pictures on line and how these might be used to build up profiles of some of those present by legal or illegal groups, including the police who are already making use of facial recognition in various city environments.

There have been various attempts to block facial recognition, both through the courts and through various subterfuges, including the use of masks and special makeup. Covid-19 has surely added to the problems faced by Dynamic Neural Networks in recognising individuals and whereas wearing a mask was often a criminal offence now you may be fined for not doing so.

What is new about Fawkes (it gets its name from the ‘Anonymous’ mask) developed by a team of students at the SAND Lab at University of Chicago is that it is the first tool to enable us to “protect ourselves against unauthorized third parties building facial recognition models that recognize us wherever we may go” that “gives individuals the ability to limit how their own images can be used to track them”, able to defeat the tools used by systems such as https://www.vice.com/en_us/article/5dmkyq/heres-the-file-clearview-ai-has-been-keeping-on-me-and-probably-on-you-too clearview.ai using deep learning to identify individuals.

The team explain how Fawkes works (and for the technical there is a publication and source code available on the site)

At a high level, Fawkes takes your personal images and makes tiny, pixel-level changes that are invisible to the human eye, in a process we call image cloaking.

They go on to state that “if and when someone tries to use these photos to build a facial recognition model, “cloaked” images will teach the model an highly distorted version of what makes you look like you.”

Original
Cloaked

I’ve downloaded the software (a small file available for Mac and PC) and run it on a picture or two. It was rather slow – but my first files were large. I tried it again on a couple of 600×400 pixel images to post here, and it took around 100s to convert the pair.

The differences are real but pretty subtle – easier to see if you right click to download the files then view them one after the other in your image viewer. The change between the two in each pair then gives me a slightly weird feeling

But these were both images of a single person and I thought I’d try it on something rather more complex but the same size. Although it said it would take about 1 minute, 5 minutes later I was still waiting, and waiting…. I went away and did something else and I think it took around 7-8 minutes. There were small differences to most of the larger faces in the image but many appeared completely unchanged.

Original
Cloaked

The input files were all jpegs, but the output files are png, and have roughly five times the file size in bytes. They had also lost their various keywords and presumably other metadata. The files went back to a similar size to the originals when saved from Photoshop as jpg at an appropriate quality level, and it is these I’ve used here. Saving as jpg perhaps very slightly diminishes the differences.

I have of course no way of knowing whether the ‘cloaked’ files would – as the inventors say their trials show – provide at or near 100% protection “against state of the art facial recognition models from Microsoft Azure, Amazon Rekognition, and Face++”, but can only accept their assurances – and presumably their paper gives more details on their testing.

Fawkes is at the moment more a demonstration of concept rather than usable software, and you would have to be very concerned about your on-line privacy to treat pictures with it. But it does show that there are technical ways to fight back against the increasing abuse of personal data and its commercial exploitation by corporations.

Recently we’ve seen complaints being made by protesters about photographers putting their pictures online, with some arguing that their permission is needed or that they should be pixellated. While photographers rightly argue their right to photograph and publish public behaviour as a matter of freedom of speech – and the idea of claiming privacy seems to negate the whole idea of protest, I can see no objection to minor alterations in images which retain the essential image while frustrating AI-assisted data acquisition. It would I think be rather nice if Adobe could incorporate similar technology as an optional ‘privacy mode’.

Images used above are from My London Diary No War With Iran protest on 4th Jan 2020 opposite Downing St.


All photographs on this and my other sites, unless otherwise stated, are taken by and copyright of Peter Marshall, and are available for reproduction or can be bought as prints.


Fuji or Olympus?

Wednesday, December 4th, 2019

This is the question I’ve been asking myself for some weeks or months. For a year or so I’ve been finding a camera bag full of Nikon gear too heavy to carry for the length of time needed to cover events in London. It’s mainly standing around that I find a problem so far as my health is concerned, and I have to remember to either sit down or to keep moving to stop my ageing veins becoming inflamed. Walking is a little better, though I do get tired much more quickly, and while I used to walk for the length of a working day and perhaps cover ten or a dozen miles, now I get tired and give up in half the time.

I can still run when I need to, though not quite as far or as fast as when young. Last Saturday when I saw a march going down Whitehall from in front of the National Gallery I ran to catch up with the front of it, around 600 yards in the fastest time I’ve done for some years. But still slow compared to my youth, when before smoking took its toll I recorded some decent but not outstanding times. I once won a quarter mile at the local youth sports in a world record time and at least fifty yards ahead of the next runner. The timekeepers ran up to me pointing at the time on their watches, and in a perhaps stupid fit of honesty I told them that the race officials had put the finishing tape in the wrong place. I was very annoyed as the conditions had been perfect and I would surely have recorded a personal best on the day over the full distance.

But no I feel a great need to cut down the weight I carry, and the Nikons are only for special occasions (the D810 is now my slide scanner – more about that one day in another post.)

For some years my holiday cameras have been Fujis. I started with the fixed lens Fuji X100, then went on to an X-E2, followed before too long by an X-E3. I swapped my Leica M8 with a friend for an X-Pro1 because I wanted to work in colour without all the fuss that the M8 needed. All of these Fujis were good in their way – and if I could be satisfied with just a say 28, 35 and 50mm equivalent lenses I would have been happy with the X Pro1. But I really got serious with Fuji with the X-T1.

I tried working with the X-T1 and one of the Nikons. It was still a fairly heavy combination, but the X-T1 was pretty good (if occasionally mystifying.) Its 10-24mm wideangle zoom was an improvement optically than the Nikon 18-35mm that I’d bought when the 16-36mm gave up the ghost (it remains on my desk with an equally almost certainly beyond economic repair D700 as an expensive paperweight) though sometimes a little slow to focus. It was good to have the extra wide angle that its 15-36mm equivalent provided – I sometimes found the Nikon’s 18mm not quite wide enough.

But things were still too heavy. And when I saw an Olympus OMD M5 II selling new for just over £400, Micro Four Thirds seemed to be the answer (as one of my colleagues had been telling me whenever we met.) Along with the body I bought the absurdly small and light Olympus 18- 150mm, also going cheap. Just over 3 inches long and only 10 oz. I don’t own the Nikon equivalent, but it is half as long again, weighs almost three times as much and costs over twice what I paid for the OM lens.

And using the M5 II usually turned out to be a great experience, except for a few quirks – the most serious of which was perhaps the ease with which the main control dial could be inadvertantly moved. Working in shutter priority it is far too easy to find yourself taking pictures at 1/8th rather than the 1/250th you have consciously selected. Though with its effective in-camera stabilisation the pictures were still usually sharp unless anyone moved.

I don’t make a great deal of use of long lenses, but this August I spent some time testing the Nikon telephotos I do have, an elderly 70-300 and a couple of shorter zooms (one a DX) against the Olympus. Despite the much smaller 4/3 sensor, this gave the sharpest images and I could see no difference in the amount of detail.

For the past months I’ve been working almost all the time with the Fuji X-T1 and the Olympus M5 II. I’ve bought an expensive Panasonic Leica wide angle zoom for the Olympus, and can chose either camera for wide-angle or telephoto use, and can’t quite decide which I prefer. Both cameras have their quirks and neither is as straightforward to use as the Nikons. And winter weather and working in poor light have made some limitations felt, particularly with the noise in Olympus images at ISO over 3200. The D750 gives noticeably better results at ISO 6400 and focuses better in low light.

Of course the X-T1 is quite an old model by now – and the M5 II is now being updated as the M5 III. It would be easier to work with two cameras from the same marque, and I’ve been wondering which way to go. The M5 III seems only a minor upgrade on the II, and annoyingly takes slightly different batteries. I’ve been thinking of getting a second M5 II instead of waiting for the III, and the price is now even slightly lower. The X-T30 looks much more of an upgrade on the XT1, and is even lighter than the Olympus, but is not weatherproof, and I have more Fuji lenses… With some special offers and rebates the difference in cost isn’t great…