While many photographers have probably stitched together a handful of images to create a panorama, this scientist has taken things to a whole other level by using a staggering 8,439 pictures.
Detail is usually an important factor when it comes to image quality and this is especially true for documenting and archiving important works of art. Robert Erdmann, who is a Senior Scientist at the Rijksmuseum has managed to capture Rembrandt's famous painting The Night Watch at 717 gigapixels and recently sat down to explain how he did it. The video presentation is from PyCon which is the largest annual gathering for users of the open-source programming language Python. This programming language is crucial for this particular image project as Erdmann uses many scripts to help him accurately reproduce every facet of the painting digitally.
Using a Hasselblad H6D and an elaborate mortised sled system, Erdmann is able to program his setup to meticulously move and document every inch of the painting in great detail. These 100-megapixel images will then go on to be stitched together to make the final image. The completed stitched image is an eye-watering 925,000 pixels wide and 775,000 pixels tall and has a resolution of 5 microns. If you weren't already impressed with those huge numbers, a resolution of 5 microns puts the pixels of Rembrandt's digitized painting at a smaller size than that of a human red blood cell.
While I appreciate that the level of detail in this project may not be desirable or achievable for many photographers, I do think it's important to see the processes involved in something so sophisticated as it may help to inform and inspire working practices going forward. For those curious about how much detail a 717 billion pixel image has, Erdmann has very kindly uploaded the image to his website in a way that allows the viewer to explore every inch of the painting in microscopic detail and is well worth a look at.
That's super cool and way more advanced than I have ever seen. Paintings are never flat and frames are typically not totally square and have a bow that twist the frame more or less. They use 2 Picolites with Picoboxes and custom barn doors.
The camera is mounted on a special plate that moves on different axis, but the the special one that may not be obvious is the one that allows actual rotation of the camera. That's really cool because if the paint lifts a little up where there is a crack, they can actually fix that. I would imagine They probably don't use it much because that would alter the rendering of the painting, but that's probably a way to control shadows and highlights from the light hitting those imperfections if they decide to. That's just my view of course. But that's probably a cool technique for modern oil paintings that get retouched by the artist before a copy. Oil takes months to years to totally dry and exposed to focused lights on fresh retouches can be a real nightmare. They also show that they use two Scoros on one image and a single on an other. I know it makes sense to use two of the same pack but practically I find that two heads on the same pack even a very advanced one actually yield better control and consistency. The image at 6:30 shows how close the Picos are and at 1600ws at 7.5, that's a lot of light output.
It's a very impressive set up with the big lifting arm and the equipment used, but the programming behind is even more fascinating.