Go to page:
-
Shanghai, one of the most populous cities in the world, has never been pictured in such enormous detail. Let the world experience this fast growing city and understand China is always my dream. Through my telephoto lens, I wish you, the viewer, can blend into the daily life in Shanghai and enjoy your virtual visit. ...
-
-
Stats
- Favorites
- 41
- Comments
- 84
- Snapshots
- 836
- Total Views
- 323150
- Explore Score
- 176
GigaPan Comments (84)
Toggle Minimize gigapan_commentGerald Donovan (December 23, 2010, 05:23AM )
Hi Alfred - Further to your earlier comparison of the two signs, I'm home now and have access to Google Earth. Was a bit difficult working it all out since your placement of your panormama in Google Earth isn't totally accurate, but looking around, I managed to discover where you took it from. The exact co-ordinates of where you took your pictures from are as follows: Latitude: 31 degrees 12 minutes 14.83 seconds North Longitude: 121 degrees 26 minutes 68.62 seconds East (this is roughly 590 metres at a heading of 305 degrees from where your panorama is currently placed in Google Earth) Now that we know where you took your image from, we can use Google Earth to work out how far away the building is that you snapshotted for your "scientific comparison", and compare to the Dubai example you used. Distance from your camera position to the sign you created a snapshot of in the Shanghai gigapan: 130 metres. (not sure if there's been a glitch, but it would appear that the snapshot you took no longer exists. For reference, it was a sign on an air conditioning unit on the building on the very left hand side of your panorama with the domed roof, 5 floors below the snapshot called "office" created of a chap sitting at his desk, by user gbtbpb). Distance from my camera position to the sign you created a snapshot of in the Dubai gigapan: 750 metres. (again, it would appear that there is some kind of glitch with the snapshots on Gigapan, because this one no longer exists either! For reference, it was on a building called Churchill Tower in Business Bay in Dubai.) We're not quite talking about two things of "about the same distance", are we. Time to move on, I think? Kind regards, Gerald.
Alfred Zhao (December 22, 2010, 07:33PM )
As promised, here is the completely objective comparison of the three largest panorama, due to format issue, you will have to view the image on my flckr page. farm6.static.flickr.com/5168/52843 44733_51a7a5430d_b.jpg
Alfred Zhao (December 22, 2010, 04:57PM )
Again, main question we raised here is what is considered scientific and what is not. Everyone use the same software to stitch image, when there is no standard way to measure the true pixel count of a panorama, then we should not alter nor manipulate the software output, this is a baseline count. State clearly the estimated theoritical analysis is the ethic way to do things in science and technology fields. Shortly I will post an objective comparison of the existing super size panorama. I agree, we need to have a worldwide committee and also a set of standard tools for evaluation. that's something we should propose to the next gigapan conference, fact is, I already raised this issue among serveral key personnel such as Alexandre Jenny (Kolor, Paris 26G), Holger, Schultz (AFB, Dresden 26G). Amount of information is also another important measurement. Before a world wide consensus, everyone in this community should be honest to the tool and emit at 100% and state the theoretical estimation just like I did. Any manual manipulation is a fraud.
Terror Pixels (December 22, 2010, 04:23PM )
Alfred, I do want you to know that this is a very well done image, I like the quality - it's a great stitch! The subject is very interesting and it's an incredible feat! Well done in that respect. I wanted to mention that in case you thought I only had negative things to say. All of us here are part of an emerging field of photography (yes I know I haven't posted much yet), really pioneering this whole thing and, at the moment, we are self governed as far as 'records' go. So I feel rather responsible for the standards (and records) that we set. I really do feel like you are putting megapixel count on an unnecessarily high platform and are sacrificing ethics to do so. For this reason I do wish there was a governing body for records such as this (perhaps Guiness will get on board). Until we have a governing body we are individually charged with being honest about the true megapixel count. I do request as a fellow gigapixel photographer that you stop deliberately misrepresenting the images you shoot. It's obvious that you know how to use Autopano very well, and if I know how to correct for wildly incorrect outputs you do too, so please stop blaming the software for the output. Bottom line is that it's an incredible shot, an amazing feat, and something to be amazingly proud of, but it is NOT anywhere near 272 gigapixels.
Alfred Zhao (December 22, 2010, 03:48PM )
I like this image ( www.gigapan.org/gigapans/63167/ ) very much. Regarding explorability, The event such like Obama Inauguration ( www.gigapan.org/gigapans/15374/ ) is full of explorable details. Some remote, unfamiliar place with much details are often in the highly explorable category. Also cultural recognition plays very important role. People from Shanghai found this picture very explorable because they are familiar with the buildings, life style, street and city. The current gigapan ranking and showing mechanism exactly buries the chances for the highly explorable pictures from appearing. The life cycle for a typical panorama is only less than a day. As soon as an image disappears from the most recent upload, the chance for that image to come back to life is very low. Even highly explorable, they won't get the exposure unless you seriously promote it via other channels. In most cases, people do not have the kind of energy or capability to do so. On the other hand, an image for whatever reasons become "popular", they will take the same spot for lengthy amount of time. Just like re-enforced learning, people keep seeing them and therefore their position get even more stable. "Rich become more richer". I found this mechanism suffocates the creativity in this community and needs to be toppled. A simple way is to introduce random pick on top of the most "popular" image, that way, an image's life cycle get longer and good things can surface based on their merits (eg; beautiful picture, better detail, explorability, size, advanced technique, etc.)
Ronnie Miranda (December 22, 2010, 12:08PM )
This is a great gigapan for its technical achievement. But creating the largest gigapan doesn't necessarily mean it is interesting to explore. There are many gigapans in this site that are quite the opposite - their goal is not to break any world record, but rather to make a great image that is interesting to explore. The keyword is 'explorability', and that is what the explore (popularity) score and what "gigapanning" is all about.
Tom Nelson (December 22, 2010, 06:38AM )
I have no interest in arguing about pixel counts. Instead, I congratulate you on your huge panorama, full of interesting detail, and I appreciate the significant obstacles you overcame to achieve it. Tom
Gerald Donovan (December 21, 2010, 10:57PM )
Hi Alfred - You don't think it's fair that the "most popular" panorama appears at the top of the list of "most popular" panoramas? I wouldn't worry about it if I were you. Firstly, I don't think many here actually see this as a competition, and it's certainly not as if there's any real material gain to be had by having a panorama at the top of any particular list here (Gigapan wouldn't even give me a spare battery!). Secondly, give it a day or so and I'm certain you'll have a "score" of over 200 and appear at the top of that list. The Dubai pano hasn't been the most popular gigapan on the site for the entire time it's been up - others have had higher scores. Regards, Gerald.
Alfred Zhao (December 21, 2010, 10:02PM )
Regarding the resolution comparison, it was a question raised by someone in the community to me privately. I just share some results in the thread. There is absolutely no intention to demean you. However, I do see people negatively comment on my image that was worth of 9 months of decent work. One more comment for Dubai image, due to the lacking of random rotation mechanism on Gigapan website, the most popular image always get extra boost. This is not a fair game. It buried a lot of great shots in gigapan. This is shared feeling among many gigapaners. I hope you can understand that. No offense though.
Alfred Zhao (December 21, 2010, 09:48PM )
Gerald, The point I am trying to make is: A. Autopano Giga has its reason for interpolation, since we all use the same tool, we should use the same baseline. If one people set to 50%, the next one set to 70%, there is no apple to apple comparison. To show spherical image on a 2D space, the closer you get to the north or south pole, it will have to be extrapolated. That's why the north and south pole in the world map looks so big. This is the scientific reason behind it, so there is really nothing wrong with the Autopano tool. Alter the slider ruin the baseline. That's why I strongly disagree the comments in the thread. In addition, a good panorama should be properly cropped. I am trying to raise the bar here.
Gerald Donovan (December 21, 2010, 09:16PM )
Hi Alfred - Not sure if there's a delay in comments getting posted, or whether my last one didn't get through. If similar comment comes through twice, my apologies in advance. Briefly: (1)Strange that you accuse me of having a "sour feeling" of other people having broken "my record" when I have already in these comments corrected someone who was saying how great the Dubai one was, and pointed out how there are others out there WAY better than it. Such as the Vienna one. It's not my record that's claimed to have been broken here. And I can assure you that on the day someone published a panorama that beat the Dubai one on both size and quality, I was actually delighted. And mightily impressed. (2) If you want to do a "scientific comparison", then do one. Posting two images on flickr doesn't exactly help people now, does it? You could just have easily posted links to the two snapshots here on gigapan. For reference, having done a quick check on Google Maps (don't have Earth at work), the sign in the Dubai panorama that you link to is on a building roughly 900m away from where the camera was. Not sure how far away that building is in your example, but I'm sure you can provide the details. A proper objective comparison of image quality across different panoramas is of course something that is both interesting, and useful. That image quality will of course be impacted by a number of factors not limited to interpolation. But of course, all that is clouding the actual issue here. I recognise that you state you're happy to let Gigapan do whatever it feels. I think I - and others here - are simply questioning why you do this. Regards, Gerald.
Alfred Zhao (December 21, 2010, 07:15PM )
Gerald, first of all, the flckr snapshot is removed, if you can not tolerate people doing some simple scientific comparison, that's fine. You said you intentionally manipulate the Autopano result by adjusting the slider, I don't think it is the right way to do things. That means you can control how much pixels you want to have. I think the best way is to emit at 100% and clearly state the estimated optical pixels. That is what I did. If you have sour feeling about other people break your record, that's understandable. In any case, my image set has 12,000 images, it is something very few people ever did.
Gerald Donovan (December 21, 2010, 06:58PM )
Hi Alfred - I'm not sure what you're trying to prove (apart from infringing my copyright) with your flickr images. The question here is not one of whether or not "Can we safely conclude, the Shanghai image has at least the same resolution if not higher [than the Dubai image]?". Ignoring the fact that of course it should be - it was taken with double the focal length - the issue here is that you appear to have knowingly and deliberately allowed a known "flaw" in the Autopano Giga software to massively extrapolate YOUR image. Comparing it with other images is irrelevant. You have repeatedly done this with your large panoramas, despite it having the issue pointed out to you on a number of occasions. I simply don't understand why you continue to do this. Autopano Giga/Pro has a simple little slider that allows you to adjust the size of the output panorama. I've used it myself on several of my more recent gigapans in order to "correct" what was clearly a ridiculous output size for the image taken. As you point out, this tends to happen where there is a large variance between above and below horizon FoV's (couple of my examples - latest Burj Khalifa gigapan and the Sheikh Zayed Mosque one). I'd much rather be discussing the clearly impressive feat of taking and stitching 12,000 images, and the subsequent tiling and uploading by the way, but if you want to keep going down this path, that's entirely up to you. Kind regards, Gerald.
Raymond Goldfield (December 21, 2010, 05:55PM )
So much detail! Congratulations! New world record, you should apply for Guinness.
Alfred Zhao (December 21, 2010, 12:43PM )
At about the same distance. Let's compare two snapshots. Dubai Snapshot (estimated 4000 mm x 1330 mm) farm6.static.flickr.com/5002/52815 54082_eb5f9ebf71_b.jpg
, Shanghai Snapshot
(estimated 200 mm x 67 mm) farm6.static.flickr.com/5284/52809
52347_5153ce9a26_b.jpg 
. Shanghai image is 6
times larger in pixel size and only 2.5 times
larger in terms of optical pixel size. Yet, to
achieve the same resolution, Dubai image requires
400 times larger area. Can we safely conclude, the
Shanghai image has at least the same resolution if
not higher?
Greg Downing (December 21, 2010, 09:51AM )
The comparison between a 1 gigapixel image and the full res you made here www.gigapan.org/gigapans/60750/ was very useful. I would be interested in seeing a similar comparison between the "optical resolution" and the interpolated image.
Greg Downing (December 21, 2010, 08:56AM )
Hi Alfred, We discussed the issue with Autopano outputting the wrong resolution in the comments here: www.gigapan.org/gigapans/59267/ and I see you list the correct resolution for what you shot under "optical resolution" What is the reasoning for doubling the resolution with interpolation?
Alfred Zhao (December 21, 2010, 07:49AM )
To answer Karim from Dubai. When you look at the 1/100 ( www.gigapan.org/gigapans/60750/ ) comparison and the 1/1000 ( www.gigapan.org/gigapans/61595/ ) comparison, the levels of detail is significantly different. In terms of picture overview, people can argue, picture taken using super wide angle lens can capture the same scene with the same amount of information as the gigapixel panorama. The best way to see the difference is to zoom into the same location and compare side by side to see the detail being captured. In the community, we were discussing have an information measure attached to each gigapixel panorama. Entropy measure can be a quantitative way to do so. In this case, empty sky and water surface etc should discount the amount of information in the picture. Blur detail should also discount the total amount of information. I am actively thinking and practicing how to improve the image quality for future super large panorama. It is a very hard task, also, cost is another limiting factor, Will Canon 1DS plus 800mm F5.6 prime lens do a much better job? I think it will show difference. Then, we will have to buy the Clauss Radeon HD to hold the monster equipment. But it is one way to do it.
Gerald Donovan (December 20, 2010, 11:31PM )
@karimsaad Thanks for the kind comment, but I feel I should jump in and offer a different opinion. The Dubai 45 Gigapixel has got its own flaws, and is by no means the best multi-gigapixel panorama out there. There are a few others that have been done, but not hosted here, which I am more than happy to recognise as being of much higher quality. The 50 gigapixel image of Vienna immediately springs to mind as being much, much better.
Karim Saad (December 20, 2010, 10:29PM )
Very good quality but no interest in going so big to see a piece of window. I saw the 1/1000 version which is less than 1 Gigapixel and it delivers the same info as this one!. Still Dubai 45 Gigapixel is the best in terms of size, quality, information, subject, composition and worldwide impact! Still the most explored! But finally well done Alfred!