Lots of people have been asking: how was this gigapixel panorama created?
I promised to answer everyone’s questions, so here we go…#1 by Rami Saarikorpi on December 18, 2009 – 10:33 am What kind of robotic device did you use? I have Gigapan, but I can not get my Mark II + 300mm to it
I used a robotic device made by my friend Traugott. It allows the use of larger cameras and lenses. It was very helpful and saved a lot of headaches. I have made gigapixel images without this device, and while it worked – – it was much more difficult to shoot, and there were also issues in the stitching that needed to be resolved. Overall, having a robot move the camera for you really makes the process more efficient.#2 by Waph on December 18, 2009 – 10:48 am Did you try other stitching software like the Gigapan and gigapixel plug-in of Autopano Giga? How long did the stitching period last?
I have tried many image stitching programs, yes. The gigapan software allows a degree of automation that is impressive. Autopano Giga is also very powerful. You neglected to mention Hugin, which is a free and open-source panorama stitcher; while the usability takes some getting used to, it is very powerful. I prefer PTGui because of the degree of control it allows in the stitching and blending of the panorama. It is an extremely well-made program. It also allows the possibility of “totally automatic” stitching, and when your images are created with this in mind, it can do perfect job.
The stitching period lasted about 3 days. However it was necessary to do it more than once 🙂 I was using a 4-year old machine; if I was using an up-to-date workstation, it might take half the time. It’s possible that in one or two years, the same image might take a fraction of the time to stitch – many innovations are happening in this little corner of technology!#5 by Maksim on December 18, 2009 – 5:24 pm Did they remove the windows for you in the observation deck? Or which observation deck did you use?
I was situated above the observation deck, in open air. Those windows in the observation deck are unfortunate – a double layer of dirty windows doesn’t really allow a very nice photo to be taken!#7 by Clint on December 18, 2009 – 10:32 pm Amazon EC2 might be a good way to render the Panorama.
Sure, but you’d have to upload the images to render it. And then download it again if you want to open it in photoshop. If I lived in Korea, this might be feasible. However your idea certainly has merit, 360 Cities uses EC2 to convert our panoramas into different formats. It is very economical and convenient.
This panorama was actually shot from 3 different balconies. It is possible to get to the top of the antenna, but this is much farther up, and Prague does not really have any tall buildings anyway, so the view isn’t necessarily better up there.
I was up in the tower for 5 hours. I shot more than one panorama at different resolutions / focal lengths. The first one I made was using a 45mm lens. Only after I had finished that one, I decided to try using the 200mm lens over a few more hours. This second one took more than 3 hours to shoot from beginning to end. I also had some issues with the equipment – the camera was not firing every shot as it was supposed to, and I had to shoot the entire section again from the beginning. Towards the end, my portable hard disk was getting very full. At the very end of the shooting, the battery in the robot finally died – with four shots to spare. So, it was only with a little bit of luck that everything worked ok.
Shadows were not really a problem. If there was much less in the image – say, if it was a panorama of 2 trees in the sunshine with their shadows going across the ground – in that case, shadows would be a problem if there is much time passing between shots. The sun doesn’t move as slowly as you think.#8 by conrick on December 19, 2009 – 4:43 am Did you atacched the camera to a computer or to an external drive?
No, the camera was taking photos directly to its memory card. It wasn’t connected to a computer or hard disk.#13 by HErnestM on December 21, 2009 – 4:49 am A minor point compared to some of the others, but the first thing that came to mind was why was it shot with a zoom instead of a fixed focus 200mm 1.8 or 2.8? On top of that the zoom was extended out to the maximum 200mm. Typically a lens has the most optical aberrations at the extremes (70mm and 200mm in this case). Granted the lens was of decent quality I’m sure, but unless I have to, I try and back off a little from a maximum zoom to get a clearer image. Just curious about the decision to use the zoom.
It was necessary to use the best type of lens possible – the one that is in my possession 😉 That’s the Canon 70-200mm 2.8 which is a wonderful lens. Sure there are better ones out there.Does the tiling process crop out the edges to help with the image quality?
Yes, because the images that were shot overlap by about 25%, the stitching process ends up using only the central portion of each image. So if you are a stickler for fidelity, you can rest easy knowing that the less-sharp edges of the image do not appear in the final stitched panorama.#26 by frank noon on January 11, 2010 – 6:50 pm You must have a lot more patience than me to create this.
Actually, I’m not sure. I don’t know how someone as impatient as myself ever decided to do stuff like this. Life is very surprising 😉
I hope I have answered everyone’s questions. If you have more questions, please leave a comment, and I will answer them as they come.
7 thoughts on “Prague Gigapixel – How was it created?”
Thanks Jeffrey for this answers.
Another question, what was your setting for focus? Did you shoot in autofocus mode or not?
Focus stacking could be interesting with a completely robotic device for pan, tilt and focus.
I focused about halfway to the horizon, and stopped down to about f/16.
Focus stacking is very interesting and will certainly be necessary if using a longer lens, say 400mm – you can see in the 26 gigapixel dresden image that the horizon is out of focus, for example. However focus stacking increases the complexity and potential for error, by a very large amount. This is partly because the camera companies (Canon, Nikon, I’m talking to you) still have very primitive firmware in their cameras with no possibility for user customization – so doing such things as focus bracketing is more of a hack than anything else. Canon and Nikon, are you listening? Why do you allow only 2 stops bracketing? No intervalometer? No focus bracketing in camera?
Can you say more about the tiling process after you created the file. what format? sizes? TIFF has a 4GB max size. And, what viewer?
I used the PSB format (photoshop large) because that’s the only image format which can support such a big file. (OK there is tiff_m or something, but no one supports that in reality)
we are using krpano viewer – http://www.krpano.com. it also has a program for creating tiles called kmaketiles. there is a GUI for creating tiles called Gigatiler which also works well.
Please can you describe it more detailed? As i see, kmaketiles can’t read psb files.
My panorama width is 90.000 pixel, so jpeg is not option due to is limitation of max 64.000 pixel.
Gigatiler works, but the output tile structure is not the same as kmaketiles.
This can help you for focus bracketing