Repairing an image by cloning from another

Spread the love

MANY photographers habitually use layers from everything. I don’t! In fact, I try to minimise my time spent on post-processing shots for stock library sale, and work very quickly. If it needs complex setup or demands working using layers to be able to go back and change things, I’ve probably already wasted too much time. Here’s an example of an Alpha 700 shot created from two slightly different versions, and how it was done.

From pillar to post

From pillar to post

This is an example of a situation that has stock image sales potential; couple on holiday, using a payphone in Benidorm, Spain – right next to another means of contacting home, the yellow post box used to send postcards. Two forms of communication in one scene. That kind of storytelling juxtaposition is what makes saleable editorial photography.

But those roadside posts are in the way, there’s an unattractive guy in a wrinkly sports top in the background, and the post box is rusty with litter round the base.

I was sitting having lunch, and did not intend to move much from my position – certainly not to get up, move close and shoot a picture. That would only have puzzled the couple at the phone box (why would I want to photograph them?) and even worse, they might have looked at the camera and stopped just being people going about their everyday business.

So I continued to observe the scene without moving, and when the background guy got up and went, took a second shot from a slightly shifted angle (just a foot or so to one side with the camera):

Still not right...

Still not right...

My freedom to move was limited (there were other obstacles in the way) and immediately this shot was taken, my subjects moved on. Back in the editing stage, I could see that by using Photoshop‘s Clone Stamp tool, data from one shot could be used to repair the other and remove the offending bollard entirely.

To make life easy, I did not work from a full copy of the first shot. The second shot contained the parts I could not ‘make up’ using random or fractal-nature detail – the girl’s leg. Stonework can be invented in seconds from other bits of stone. It is easy to clone out anything on a wall, street, grass, beach, sea, or other relatively consistent texture. A very soft-edged stamp is used, and care taken never to have repeated details (the eye sees these straight off in the final picture) but to ‘damage’ each cloned area with bits of another. The process is very fast once learned. I clone in small, fast bites using a mouse and never work with a graphic pad; I find the mouse far more controllable and intuitive, and I don’t even look at my mouse hand, just at the screen.

Using a small clip (above) of just the area I wanted to repair from shot 1, floating it over shot 2, I aligned the source with the target by picking a definite geometric feature (the corner of one of the wall stones). The bollard was removed in a couple of minutes, and data from the ‘master’ picture was also used to repair the corner of the dress and tidy up detail. It is much easier to work from a small clip like this, and reduces memory load by having just one large image open. If you use Layers, Photoshop‘s appetite for memory use and scratchdisk space increases greatly (even the History function uses plenty) and I prefer to keep things running as fast as possible on a relatively underpowered iMac 24″ machine.

Finally, yellow from the clean parts of the postbox was used to repair the rusty bits, but with care not to destroy some of the ageing. The soft-edge tool was allowed to go just close enough to the edge of the rust area to leave the outline of the patch. This made the box look naturally weathered still.

Here is the result:

It’s cropped too, which creates a much larger thumbnail when seen in a photo library website (panoramics have skinny thumbnails, square images have the biggest and the most impact). The shadows are not fully repaired but that doesn’t matter – they are just shadows and uninformed viewers will not see anything wrong.

The actual image is 55.5MB. Using the Sony Alpha 700 and CZ 16-80mm lens (an extremely sharp lens) there is no problem exporting the raw file using ACR 4.5 to the 25 megapixel fixed interpolation provided by Bridge/ACR (a 70MB+ filesize). Even at this scaled-up size very fine detail is retained, and images pass the quality control process of photo libraries. It is frustrating to read about some libraries/agencies which will not accept A700 images, but accept Canon 400D and Nikon D80 shots. I have used both of these, and I know where the A700 is positioned relative to them – and it certainly is not ‘below’ in any sense, especially final image quality.

The total time taken on a quick repair job like this is under 10 minutes. I’ll do something similar with maybe one in every 100 shots, sometimes using detail from a second shot taken on purpose, sometimes using detail from another area of a single shot.

Such pictures are marked as ‘digitally altered’ for library sales purposes. Are they dishonest? I don’t think so. Photographers used to use retouching inks, scalpels and even cut up layers of emulsion from one print to float on to another. Images were being combined to overcome the limitations of dynamic range by the 1860s (Gustave le Gray’s wonderful seascapes with dramatic skies) and by 1900 companies like Bamforth in the UK were creating postcards of scenes which never existed at all!

– David Kilpatrick


4 Comments

  • You did a real nice job on that one – especially keeping some of the “aging.”

    We often use this technique in group photos from cycling trips – use the head w/smile to replace a frown, etc. Once we cloned in a closeup of a crawling turtle resized to look the correct size and placed it on horizontal tree limb passing through the group shot. The photo was sent to about 20 participants. No one ever noticed the turtle – or, if they did, never mentioned it 😉

  • Well, I really think it on one hand depends on the purpose, but on the other hands has to acknowledge the specifics of the medium, too. A landscape photographer, showing prototypes or “icons” of situations of light and media, is for sure in another position as someone depicting a certain geographic location. Both versions of your Luxor shots are in their own rights. While in a story about balloon adventures in egypt the cleaned up version would be acceptable, in a story about Luxor it would be not.

  • Agreed. But, as an example, we got some very good sunrise shots over the Nile and Luxor town with balloons in the sky:
    http://www.alamy.com/thumbs/6/%7B2CEE6A7D-3375-41E6-ACAA-DF46F622D83C%7D/ANBABR.jpg
    http://www.alamy.com/thumbs/6/%7B8FF07EB2-2827-4C63-B3E4-8DD3A53169AA%7D/ANBABW.jpg
    One has had some of the messy building concrete reinforcement rods tidied up. It was tempting to completely retouch the example with loads of these unfinished building columns. In the end I decided not to, but I may go back to it one day and produce a ‘clean’ city skyline with the mosque and balloon.

    David

  • “Such pictures are marked as ‘digitally altered’ for library sales purposes. Are they dishonest?” My guts feeling says that it depends on the purpose the pictures are used for. In this example it’s an editorial type shot, an illustration. From my point of view this is completely o.k.
    But last week another photographer showed me postcard type shots of my hometown where he had increased the height of the clocktower and the mountain in the background, making the whole scenery definitely more impressive. If I would get this postcard see this as an incentive to travel there, I would be disappointed, and when the manipulation would get uncovered, I definitely would become angry.
    Unreflected manipulations that do change documentary style pictures – and even a postcard with a landscape scenery is more a documentation than a pure artifact – will badly affect the reputation of photography. So marking altered pictures as such is definitely a good way of handling this.

Leave a Reply