My father, Harry Rowed, was a prolific editorial and commercial Canadian photographer who left me with a large number of high-quality historical negatives dating mostly from 1936 through the 1960s. I’m the process of organizing and scanning batches of these and organizing exhibitions and sales of prints. After using regular scanners for years I switched to photographing the negatives using a DSLR (mostly Nikon D800) with a macro lens. My workflow is entirely in Linux, using mostly RawTherapee, Gimp and Hugin.
Lately I’ve become interested in working on “then and now” photos, taking shots today from the same locations as the originals. One example is this shot of my daughter in 2017 beside my mother in 1947 – a grandmother – granddaughter shot separated by 70 years.
To find the location I had my father’s shot on my phone and hiked over the mountain until the vertical and horizontal alignments matched with the original. I then fine tuned it by moving the camera and adjusting the focal length until the images approximately matched. The original was a 3x4 inch B&W negative and the “now” shot was with a D800 and a 24-70 2.8 Nikkor, which seemed to match fairly well set to about 35mm.
I now want to do a series of “then and now” shots of his gorgeous 6x6 cm negatives of Puerto Vallarta taken from 1953 to 1956. My goal is to have the alignment as accurate as possible, hopefully to produce a video with slow transitions from the “then” to the “now”, and perhaps to produce a book of these images. I think that lens distortions can be corrected with Hugin.
The question is how to get a achieve such a close match to make this possible. My ideas are:
Put the “then” shot onto the camera’s SD card and bring it up on the LCD. Put a piece of clear acetate transparency film over the LCD, draw the outline of buildings etc onto the acetate, then match the “now” image using live view.
Print the “then” image on clear acetate, hold it up and move around until the “now” scene is aligned with the “then”.
Have a laptop while shooting so the images can be checked for accuracy using transpaency in layers on Gimp. This could be done as a final step with other methods to make sure they will blend okay.
Some cameras have a double exposure feature where you can load an image and add a second exposure. I am not sure how this exactly works or how it is visualized, but maybe it’s worth checking if it could help with your task.
I think I’ve figured out a possible solution. I can print out the “then” images on transparent film (laser or ink jet printer) in the size to fit onto the camera LCD screen. Then, using live view, I should be able to move the camera around to get a good alignment. My D800 even has a clip on screen protector that should allow me to hold the film in position without using any tape. I’ll do some experiments on this.
I may need to print the “then” images lighter than normal, in order to get a good view of the live view images.
Here’s a sample of one of the photos taken around 1954 in Vallarta. Since there is a lot of foreground as well as background, it would be really difficult to get the exact positioning and focal length without having real time live view, at least to get the degree of accuracy needed to seamlessly transition from the “then” to the “now” on video.
I’m not familiar with that feature. Many camera will do multiple exposures which combine images to make a single final exposure, but being able to display an image while displaying live view would be really a cool feature. Does anybody know which cameras would have this?
FUJI XQ1. In the advance setting (that means put the control dial to Adv. mode), one can take a photo with multiple exposures. In that mode, one takes the first photo, and that photo is displayed transparently while taking the second photo.
It is a fixed lens compact camera. So you might want to know if the photo taken by your father is by lens whose focal length is within the 24-96mm range
IMHO what really counts it the actual position of the camera, the focal lens only determines the width of the field of view, you can shoot a bit wide for safety.
To position the camera accurately, I would threshold the original picture and print it on transparent slides (like the one used with overhead projectors of yore) . Then holding the slide at arm’s length I would try to align it with the landscape (this could require appropriate scaling).
PS: I just made a swift experiment, using an X-T1.
I set the camera in double exposure mode.
I placed a blue USB stick in a bookshelf,
and took image 1. I removed the blue USB stick
and put a black USB stick in the exact place,
and took image 2 (which became a multiple exp).
Without changing the setup I took image 3 (which
would be the result you are after: a photo of the
black USB stick in the same position as the old blue
one).
This is more difficult to describe than to perform!!!
Addendum: As far as I understand, you will have to bring
the “old” photos along, since I cannot figure out another
way to bring the old photo into the camera and then use
it for positional purposes
This is a technique I’ll try. One of the difficulties I see with this is ensuring that the transparent slide is exactly in the same plane as the sensor. Even a slight tilt will throw off the registration. Attaching a transparent slide to the back of the camera LCD would solve this issue, but of course I’d be working with the small screen.
I like this idea a lot. I’ve been thinking about buying an external monitor like this for video work, but it would be great for this purpose as well. Not only are these monitors larger than the LCD screen, but brighter as well, making them easier to work with on bright sunny days. I’ll research which ones will work best with the Nikon D800. In the meantime I’ll run some tests with printing off the “then” pictures onto transparent film to the size of the camera LCD to see how well this approach works and if there are any glitches I haven’t thought of.
Lovely pic. That’s Vallarta Mexico?
Love the relax of the guy on top of the horse.
BTW, in the above “minimalistic add in camouflage” you can read Mejora{l} dolor de cabeza ~ improves headache =)
you can shoot a bit wide for safety.
what the nut said; overimposed transparent prints might work. A zoom for room for maneuver (?)
Did some experiments today. I took a photo in the nearby woods, then printed it on transparent film with my B&W laser printer in the size of my camera’s LCD screen. I taped it to the screen and was able to get a decent match by using live view and moving around until the images aligned. Some points:
The small screen is difficult to see properly, and in bright sun would be almost impossible. This would be a lot easier with a larger, brighter external monitor.
A monopod or tripod is needed.
It might also be helpful to have a few memory cards, one for each location. This way, the “then” and “now” shots are next to each other, so you can easily see any alignment errors while toggling back and forth. This may not be necessary if using a 7 inch external monitor with an overlay transparent print.
I suggest: when shooting, worry about the camera position, not the focal length. The focal length affects cropping, which is trivial.
In the OP shot, consider the peak of the small rock the women are leaning against. Compare that to the peaks in the background landscape. We can see that in the right-hand photo, the camera should be positioned slightly to the left. This would also correct the horizontal positioning of the women’s heads.
When shooting, the scale of the image is irrelevant. Only the camera position is important. So an acetate over the camera screen wouldn’t work for me, because the difference in scale would distract me and I would move the camera back and forth to get that right, but that’s the wrong approach. I would take a decent sized print or acetate or even a hand-made tracing with me, with important features highlighted: significant bumps in hills at the skyline, mid-foreground and foreground. Move around to find the position where the same relationships hold, and objects are the correct relative size.
Then get out the camera and put it in the same position. Ensure the lens is wide enough to capture all the original scene.
As you mention, the camera position is really the key, and be sure to be wide enough. Another issue is correcting for lens distortions from different lenses. I think that image-align-stack in Hugin, using manual control points, should correct for this if the camera position is correct.
I still think there’s a lot of promise in the acetate overlay. It is difficult with the small camera LCD and would be almost impossible in sunlight. I’m looking at either an external monitor such as used for video, or using a phone or tablet as an external monitor. It looks like the app “Camera Connect and Control” will work for this purpose with a USB OTG cable to connect. It would be wonderful to use a 10" tablet for this.
If the focal lengths were radically different, say the original was shot on a 24mm lens and the new shot on a 200 mm lens, you get different perceptions of the objects in their space, right? The longer focal length compresses the space, while the wider lens makes things seem further away.
