The Largest Digital Photo

Gigapixel writes to point us to what is claimed to be the largest digital photo on the Net, at 8.6 Gigapixel. It is a composite photo of the "Parete Gaudenziana," a fresco painted by Gaudenzio Ferrari, dated 1513. This fresco is in the Church of Santa Maria delle Grazie, in the convent of Varallo Sesia, diocese of Novara and Province of Vercelli, Italy. The site uses Flash to let you explore the fresco over a zoom range of more than 180 to 1. The photo is made up of 1145 images, each 12.2 Mpixel and 16 bits per color channel. Read on for more technical specs of the photo.

Photo Shots: 1,145
Computed Data: 84 Gigabyte
Computed Pixels: 13,982,996,480
Color Depth: 16 bit per channel

Cropped Image Size: 8,604,431,000 (w. 96,679 x h. 89,000) pixel
Image Size before the final crop: 10,293,864,000 pixel (w. 103,560 x h. 99,400) pixel
Size on Hard Disk of the 3x16 bit final image: 51,625,586,000 byte

Size of Photographed Scene: 10.80 m x 9.94 m (35.43 ft x 32.61 ft), corresponding to 107.35 m² (1155.37 ft²).

True Scale Resolution: 227 dpi
Pixel Density: 80 pixel/mm²
Linear Pixel Density: 9 pixel/mm

Hard Disk space dedicated to 16 bit computing: 1.8 Terabyte
Ram: 16 Gigabyte
Processors: 4 x AMD Opteron(TM) 885 Dual Core 64 bit

Shooting on January 30, 2006
Shooting time: 13 hours
Computing time: 3 months
Final Image generated on June 15, 2006

Now we need wall displays.

[headkase]

Now we have wall sized wallpapers we just need a wall display system for them. I can't wait :) Downloading wallpapers for my walls is going to be awesome :)

Re:Talking about google maps...

[nachmore]

I think the point is they took all of these photos and instead of storing them as separate layers somewhere they combined them all into one huge photo:

Size on Hard Disk of the 3x16 bit final image: 51,625,586,000 byte"

Whereas Google Earth and the like, obviously, have more data they are still stored as separate images... (not sure why they needed to connect this one up into one image either, but it must be easier for them to analyse like that)

Re:Wow - worth checking out

[imsabbel]

No, they wont.
Sorry to break it to you, but image sensors arent cpus, so there is no moores law or anything.
There is stuff like "physics" and "optics" that have to be taken into account.

To get that kind of resolution out of a single camera you would neeed lenses that are heavier than you (just to beat the diffraction limit), not to mention that the sensor would need to be HUGE (we are at 2-4 um^2 pixel sensor size today (and thats bad already for various reasons). It should be obvious why getting smaller 500nm or so isnt a good idea (hello wavelenght of light?!). Not to mention that the real bad "noise kills everything" would start quite a bit earlier.
This big detector size would again demand better lenses... (think of large format, but with a need for precission like the best 35mm optics.

The only way to do it, in a handheld camera, would be if some breakthrough would enable negative reflraction index lenses (they can be _perfect_) and then using some ultra cooled detector.
Even then the exposure times would be quite long just because of the quantum efficiency.

Re:Huh?

[klang]

This camera, http://www.gigapxl.org/ [gigapxl.org] takes 4 Giga pixel in one shot .. now, that's cool.

Stitching 40 X 40 pictures together is just a lot of work.

How much of the light spectrum?

[dattaway]

So it maps a large area with fine granularity. Its flash so I can't determine if its a RGB or CYMK photo. Or even if it details bands in the infrared spectrum. Or wavelengths in the ultraviolet?

Re:Wow - worth checking out

[iammaxus]

Have a little more creativity. As the parent (and child) was trying to suggest, there are so many amazing ways that technology has surmounted so many previous "physics" barriers. How about this as a little potential example. You take your 2016 camera which has a measly 10 or 20 megapixels but incredibly processing power and storage and pan it over the fresco back and forth, not very carefully, and it's intelligent algorithms (and maybe built in accelerometers or other motion tracking) patch together what you are imaging into one large image.

Hell, that's a pretty boring extension of todays very real and practical technologies (I know a team at my university that is doing almost precisely that for aerial photography), why not turn the camera around while you are at it and image the room from a few different angles, get some other art work and sculptures and have the camera create an incredibly detailed, textured 3d model of the entire room?

Anyone who has seen the last, incredible 40 years of progress in technology would be pretty close-minded not to see "gigapixel" and more cameras in the next 10 or 15 years.

Soon everyone will be able to make one :-)

[rsargent]

The Gigapan device, being developed by CMU and NASA, is a low-cost way to generate 1-40 gigapixel panoramas using off-the-shelf digital cameras. Soon it will be available to the general public. See some panoramas taken with the device [cmu.edu] or find out more about the commercial version [cmu.edu]. (Disclaimer: I'm part of the Global Connection project, which is developing the device)

Re:Wow - worth checking out

[CrackedButter]

I had read in a UK photography magazine that in the near future, photography will actually be captured on video. Instead of using a camera to get the decisive moment captured in a single frame, photographers might use high resolution video technology to get their shot (lazy maybe?). You can do it now if you wanted, my Fuji S7000 (6 megapixel) shoots in video mode but the resolution sucks (640x480). Put a res of 1600x1200 at least and your original idea is off the ground. All those images shot together, then seamless added together, you got yourself a gigapixel image! The entire movie, Corpse Bride was shot using 2 or 3, 16 megapixel Canon cameras just so you know. (Obviously, those images from the movie are placed after one another instead of being placed next to each other). By the way, in the 2016 most cameras (including the shittest ones) will be well over the 20 megapixel mark. A lot of them are now, the highest you can get (as far as I know is 16 megapixel canon SLR and I read somewhere that Hassleblad have a 32 or 64 megapixel camera for sale.

Re:Talking about google maps...

[Curtman]

Whereas Google Earth and the like, obviously, have more data they are still stored as separate images... (not sure why they needed to connect this one up into one image either, but it must be easier for them to analyse like that)

Maybe they couldn't get their hands on one of these. [nrc-cnrc.gc.ca]