I was pointed to this tweet by my twittering colleague today.
As it turns out (as if it would be any surprise), I have been planning also to make a new “world record” image which would surpass the Prague gigapixel I shot last year.
It’s funny, in the very early stages of my addiction to all things panoramic, I remember the fascinating image and story of Max Lyons and his Gigapixel image. Not to belittle the efforts of the people shooting gigapixels these days, but wow, this guy was a real man. He had to actually move the camera by hand, no motorized robotic thing to move the camera with a tolerance of 1/100th of a degree. He had to actually find and record corresponding pixels in each pair of images. No SIFT, no Autopano, or PTGui, none of that. In fact Max started programming himself (creating PTAssembler in the process) only to achieve his goal. Yes, people used to be tougher, as we’ve said over and over, since the times of the ancient Greeks.
Now we have all manner of robotic camera mounts, and we have a great deal of sophisticated and very good panoramic image creation software. It is even possible to create an image ten times larger than Max Lyons’ record breaking gigapixel image with a gigapan unit and a point and shoot camera from your nearest electronics store (Canon G9, say, or equivalent) with no knowledge whatsoever about photography or panoramic image stitching. Whoever says this is a bad thing is an elitist crank. Of course, this is a good thing. Anything that makes previously difficult things easier for people is (with certain exceptions) probably a good thing.
But there are other elements of this “gigapixel race” that are bothering me. Let me get the first one out of the way. How can I say it politely? Some of these huge images are…… DISGUSTING! Or even worse maybe, BORING! Ok, I said it. Maybe I wasn’t polite about it. I’m sorry. But please. Look at the Corcovado 67 Gigapixel image. Does it look nice zoomed out? Not really – it is full of lines because of certain technical reasons (vignetting was not corrected). Now, zoom in. What do you see? Here is a screenshot:
Ok, there is something vaguely recognizable. But let me ask you this – if you bought a digital camera at Best Buy – one of those cameras with a ridiculous megapixel count (more on that later) and you looked at one of your photos at pixel level, and you saw something like this – would you be happy? Probably not. In fact I guess you would immediately return the camera for being defective.
So, how big should the above screenshot be, so that the “pixels” in it are actually “pixels” ?
Well, here is the same screenshot reduced to 1/4 size. I also applied a bit of sharpening, because it was still too soft.
That’s just about right, I guess.Well, it’s better, at least.
So, Luiz Velho, Diego Nehab, Pedro Sander, or Rodolfo Lima – what are you trying to prove? Besides to claim a world record of course! I think what you are proving is not what you intended to prove. You are showing a couple of things.
First, you’ve shown that there is a physical barrier imposed on the resolution of photographs. This resolution is dictated by the earth’s atmosphere. If you look through an 800mm lens as has been done with the above image, you actually won’t see any more detail than if you look through a 400mm lens (unless you’re lucky and you’re shooting on a clear day, minutes after a rain storm has cooled off the surfaces of all distant objects); you will only see how wavy everything looks when you look at such a small detail through so much of the earth’s (definitely non-transparent) air. Even if you had secured your tripod properly, and waited a reasonable length of time for the camera to stop vibrating, and used mirror lockup on your camera (you certainly didn’t do any of these things, because if you did, the image would take a few days to shoot – this appears to have been shot in one day, and the blur on these images suggests the camera was not stabilized so well – yes, time is tricky)
The second thing you’ve shown us is that this “Gigapixel Race” is become silly, the kind of thing that is a domain for hucksters and charlatans – a kind of meaningless benchmark that is impressive only if you don’t know anything about this kind of stuff.
You could think of this gigapixel race as a side-effect of the megapixel race – something that we’ve been benefitting from, certainly, but more often than not, for most humans who buy cameras, is very misleading. The basic idea, if you’re not familiar with it, is that cameras are marketed with ONE specification above all others: the megapixel count of the imaging sensor. People have been led to believe that a 15 megapixel camera makes better photos than a 6 megapixel camera (the truth is in fact usually the opposite). So, while we have been lucky enough to get cameras such as the Canon 5d mk2 or t2i, 21 and 18 megapixels respectively (and “good pixels” at that) most of the time that you see some camera, and its size is megapixels, you’re getting engineering driving by marketing, driven by the human instinct to quantify everything (and presume quality = bigger numbers).
As I mentioned in the beginning of this post, I am myself preparing to make another “world record” image. But I don’t want to be too “silly” about it. Maybe I am getting old, and letting honor get in the way of what is otherwise simply a fun and competitive game to be played among photographers of the world. But still. I would like most parts of my image to be in focus. I would like elements of the image to be mostly without errors (no lines or other seams between images). I don’t want to magnify the world so that you can see nothing but blurry, defracted water vapor in the distance. I’d like the image to look nice when you’re zoomed out, and also when you are zoomed in.
Remember, you can take a 1200mm lens (one that is available today!) put a 2x magifier on it, and shoot a panorama that’s more than 1 THOUSAND gigapixels. But, unless you did lots of tricks (things that would increase the already enormous shooting time by a factor of 10 or 100) the resulting image would not have any more actual resolution than an image less than 1/10 the size. There are limits to everything, even photographs – some of these limits are bound by the Earth’s atmosphere, not just technology. But I guess we haven’t gotten to the limit yet. And there will be a lot of very silly claims at new world records, with most of these images being viral marketing vehicles and not much more.
So please, gigapixel photographers of the world – remember that there is more to a photograph (even a world record photograph) than how many pixels it contains. Remember why you started photography in the first place? It was to take a photo OF SOMETHING, probably. Let’s not forget that in our quest to push the limits of knowledge and technology.
23 thoughts on “On Gigapixels, Photographers, and Charlatans”
Interesting article and rant Jeffy, but you have very good point there! 🙂
I have to fully agree with you that the quality of this gigapixel image is not world record breaking, in fact its half a job! I’m sure if put to a vote, the work of other gigapixel photographers (including yours) would be voted by the photographic community as being much better, and worthy of praise.
However, I have to also say that I fear the same is happening to “normal” panos on 360cities.net . In the begining you demanded a very high standard of quality from pano photographers on the 360cities.net system, and many panos were rejected as not meeting the standards. In fact, I remember you helping me out to improve some of my panoramic images a few years ago to meet the high standards required on 360cites.net of the time.
Some panoramic photographers like you (and me I like to think) take this business seriously, and strive to create high quality engaging panos for clients, who they treat with professional respect. But the fascination with panos has also attracted another type of person, the one who is only out to make a “quick buck”, and produce poor quality work, dare I say a rip off! The “Charlatans” of the 360 degree pano world.
I’m sure you agree that there is little we can do about what these people do in their own businesses, to each his own I like to say, but I feel quite strongly that these types of people should not be allowed to publish junk on the 360cities.net platform. This, by association, damages your brand as well as mine.
For example, the minimum pixel size of a pano on 360cites.net is clearly stated on the upload page “Minimum allowed size is 6000×3000 pixels. Images must NOT be resized.” However, this is clearly NOT the case in practice for many of the panos uploaded. Members have either uploaded much lower resolution images (approx 2000x1000px in some cases), or they have upsized the pixel size to achieve the 6000×3000 minimum. This now means that my hard work to create high quality panos is not being published besides low quality rubbish.
If a few panos slip through the cracks, that is fine, but when members who do this then go on to sign up Pro accounts on 360cities.net , and this very low quality is allowed to go on, this is sad to see. Money should be be taken in exchange for turning a blind eye to below par quality, far below the standards in place. That is not even to mention the quality of image capture, stitching errors, post processing, etc.
One of the reasons I joined 360cities.net and chose to promote it in my area is bacuase of the high quality standards which seem to now have been forgotten. I beg you, please bring back the enforcement of standards.
Please take my comments in the light in which they are intended, to help make 360cites.net better, and to grow. As a Pro member with over 300 images on 360cities.net , I hope I have some voice on this. I’m sure that other Pro members will agree with me on these points.
PS: Thanks again for a GREAT system!
Thanks for the comments. Although I don’t agree with the appropriateness of them as a comment to this blog post (it is really a different topic) let me reply to a couple of your points.
The value proposition, strategies, and intricacies of running and maintaining a community website such as 360 Cities has so many factors that it is not fair to compare it to the subject matter of this article. If you don’t believe me, try starting a community website yourself 😉 and going from zero members to hundreds or thousands. Not everything can be done as you wish or suggest (nor, arguably, should it be).
WRT to your concerns about quality, we will promote and encourage the “best and great” images on 360cities above all others, and we will do this more and more as time goes on. At the same time we will proabably continue lower our “quality requirements” in order to reach a truly mass / global audience (but not promote / show such images as a matter of course). There is a place for both, and it can be done. The exact way we are doing this, and the evolution of that, you might not agree with, but by our measures, it is continuing to work very well for us.
Hi Jeff. I think my comment is directly related to your post. We both agree that “quality is not just about the pixel count”.
I have to disagree that 360cities.net is a community website, its more of a 360 panoramic image directory.
You do not have community like features (think of flickr, facebook, etc, these are community websites). 360cities.net does not have such features that allow and encourage users and guests to interact on the media (except on the forum, etc).
I put this to you: Why does the “upload a new image” page clearly state that 6000×3000 pixels is the minimum size, when you are allowing far lower? Either the message should say “min 2000×1000”, or
Thank you for the challenge to create a community website, if fact I have been very involved in a large community website, and the advice I can give based on what we learnt in setting it up was: “be true to what you claim to be”. Do not claim to be high quality, and then allow any level of quality.
Take a look at iStock. They have a very high standard of images allowed onto their website, images that fall short are rejected! They have grown into a massive company, and a great internet brand, which they would never have done if they lowered their standards. I urge you, please bring back the quality standards on 360cities.net , and stick to them, for your sake, and your members, and the future of the website.
Not sure why you said that Max needed 1/100 degree accuracy when shooting that first gigapixel image. With his 280 mm lens on a 1.6 crop camera, the field of view in the horizontal direction is 4.5 degrees and about 3 degrees in the vertical. According to Max’s website, he simply moved the camera enough to get just a bit of overlap between images. So a half degree of accuracy should have been plenty. No?
I was referring to the idea that some motorized panoramic heads these days can be used within 1/100 of a degree precision, and as such, images can be placed together on the computer without any need for optimization and stitching (using control points etc.). that, if i’m not mistaken, is how the gigapan system works, among others.
OK, I tweeted you asking if you were referring to Julian Kalmar as a Charlatan. Originally when I clicked on this, my browser loaded another Twitter user’s profile and Tweets. After reading your blog post I am in complete agreement with your comments. I have been posting comments with recommendations such as not using certian lenses ( Canon 100-400mm f/4.0 ) and capturing photographs early in the morning with a digital calibration device at one side of the panorama to get white balance and exposure information as the light changes. I have recommended using one of the video viewfinders to achieve the best focus as well.
Hi Jeffrey, Hi John,
thank you Jeffrey for starting this discussion and thank you John for your plea for quality.
When I look at the discussions around these monsterpixelpanoramas it’s always the same: Which camera did you use? Which lens? Which machine moved it? Which software? and so on. No questions about photography. It seems people look at panoramic photography as if it is just a technical thing – all you need is the right equipment. And that is how many monstepixelpanoramas look like. Just automated connection of pixels. Nobody ever wanted to know: Why did you choose this place? Why at this time of the day? What do you want to show or express? What is arranged, what is natural? … And the worst thing is, you can see it on many panoramas, that even the photographers didn’t think about these questions.
All they think is: the bigger the better. To me this is not photography anymore, it’s just outdoor photocopy. And sometimes it’s even worse. There are pictures with people whose bodies had been cut off and crippled, because the camera-robot just followed his order and the “photographer” didn’t care. To make the photography of a place natural you need people. Cutting them into pieces or connecting moving bodies to something not natural is just a lack of respect.
Yes John, we really need a better quality, and yes Jeffrey, gigapixel panoramas need to be more than just an amount of pixels – and luckily there are.
Heiner Straesser, Would you offer some constructive criticism of gigapixel photographs on my web site then. Are you able to articulate some potential improvements? Thank you in advance.
seeing your images of San Francisco and Oregon, I just thought “These are some of the great examples I was thinking of, when I wrote – luckily there are”. Only the image of the footbal stadium shows some stitching errors and partial bodies of people. When I have moving people in an image, I move the camera in smaller angles to enlarge the overlapping area. Each single shot is taken 6, 7, or more times. Before stitching I prepare the single shots in Photoshop (identical shots as layers, deleting and adding people from the layers always with a look at the neighboring image) to avoid cutting people into pieces. After stitching I search for the mistakes in my panorama and use the seperate layers of the panorama for correction. It’s a lot of work. But this work also enables you to create an atmosphere. I tried that for example in this image: http://www.360cities.net/image/yeni-cami-innenhof-istanbul-turkey . When you are in the courtyard of a mosque, you see that this is a place where children play, people meet, relax, and communicate. Here you can find people of all ages, traditional, modern and of all social status. To get that in a single image you need to be very lucky, but staying there to take many shots (90 for this pano), gives you the possibility to do it. Kind regards, Heiner
Thank you for your compliments constructive criticism. I have been pushing for multiple camera imaging to reduce non ideal overlap in moving subjects. I am thinking that I need to capture using multiple 645 format with good digital backs and better lenses to reduce noise, increase dynamic range and increase contrast. The 645 cameras also do not contain anti aliasing filters like the 5D Mark II. I really like your 360 pano of yeni cami innenhof. It is very interesting and has very nice colors and tones. It is an excellent subject for a 360.
Thank you Heiner for backing up what both Jeff and I are saying here. Quality is not just about how many pixels in the image.
I like this quote by Heiner, well put!
“To me this is not photography anymore, it’s just outdoor photocopy.”
As pioneers in the gigapixel world, this sounds like an opportunity to establish (and articulate) an industry standard of what constitutes measurable elements (within the image, or in capturing the image) to judge quality.
Of course, as with all art perfection is subjective, but if there were no data points from which to critique then this discussion would be mute.
You made some good points, some of them we are trying to overcome even before you mentioned them. But let’s not forget that gigapan’s viewer stretches the most detailed image 3x in its highest resolution level, so no matter what panorama you’ll view with it, it’ll appear blurred.
Talking about seams between images, the Corcovado 67GP as stated in gigapan’s website is a *first stitch*. We’re developing some post-processing software solutions to be able to come with a seamless panorama (that’ll be our 2nd stitch).
But, all in all, there’s no need to (negatively) criticize anyone here. We know the difference between a beautifully rendered panorama and ours (so to speak). The way you put seems like we’re “attention whores”, which is obviously not the case. We’re working hard to come with software solutions that escalates well into 60GP or more, with superb quality. The Corcovado 67GP is just the tip of the iceberg.
You can certainly find some of my own work online that is also not great – I’ll send you some links if you want 🙂
I understand that this is a “first stitch” – I read that on the page of your panorama. I look forward to seeing the next version of it, and I”m sure it will look much better on a number of levels. I will be sure to blog your efforts (not as a rant next time!)
If you want to share any tips with shooting or stitching, we could talk by email or skype.
If you are to discuss “real pixels” you need to consider that the Bayer sensors (CMOS/CCD) used in most cameras do not even capture or give you this data to begin with. You could easily scale down 30% or even more, the images produced by these cameras and loose nothing as far as discernible detail is concerned.
Attacking and insulting photographers for pursuing their passion of what they like to do with their time and equipment as you have done with this post of your is pretty lame in my opinion. There are far more constructive things to discuss.
You have many photographers that have posted images through your site. Can they expect this same sort of treatment from you some day should they create something that you deem unnecessary or do not agree with?
Yes, you make a very good point about Bayer sensors, I have thought about this also. I remember the “70% rule” from Ken Turkowski, ( http://tinyurl.com/c9q23u ) which says that it is ideal to reduce the size of an image to 70% of its original size, because you’ll keep most (or all) of the detail, but reduce the pixel size (and thus file size, download speed) by HALF! But I guess most photographers (myself included, maybe you too) don’t do this…
I’m sorry if the post seemed so negative – yes, it was a rant 🙂
Photography on 360cities is a different matter than the rant above. We provide tips and advice to any photographers who have obvious technical problems with their panoramas (panos not being level is the most common problem by far – here is Willy’s nice tutorial! http://forum.360cities.org/viewtopic.php?f=28&t=370 ) Part of 360cities’ mission is for panoramic photographers to know what they’re doing, and I hope we encourage that.
Actually, I do reduce the sizes of my images in such a manner. Not always as much as 70%, but I do reduce them.
Since I also have been using a scanning back camera (whose tri-linear sensor has no Bayer pattern) for many years, I am a bit more aware than others I guess of the distinct difference between the output of that and my CMOS/CCD cameras.
I wanted to add that several of the technical issues that effect/afflict this type of photography are already/or have been addressed within other disciplines. The astro-photographers for example have been struggling with atmospheric issues for decades, the “birders” can shed a bunch of light on various long lens issues and examples etc. I’m sure there are manny more. One just needs to sift through all of this and see what can apply or assist us with the issues we run into within this particular aspect of photography.
Actually I apply the 70% reduction rule too. 🙂
“We provide tips and advice to any photographers who have obvious technical problems with their panoramas”
This is understandable, and great! But this is not what I am referring to. I’m not talking about the new guys who need some help, but can still publish their below par work, because often their next published works are improvements of the last, which is always great to see how people develop in their skills.
However, I’m talking about the growing number of members of 360cities.net who choose to blatantly ignore the minimum requirements of quality, and who do not put much effort into their work at all. They are blatantly publishing rubbish onto 360cities.net website, decreasing the quality standard of the website.
This is not a skills issue, these people often do have the skills to do a better job, but choose not to put in the effort, and its sad to see that 360cities.net supports theses people, and does not encourage them to “get up to the standards” that 360cities.net claims to require from members.
If a photographer, over a period of time, deliberately plans to spam 360cities.net with poor quality panos, rather than spend the effort to create high quality panos, his panos should not be unpublished. Its that simple.
My tweet (World Largest Photo Contest – now on Wikipedia http://bit.ly/hecmF The Race to 100 GigaPixels is On. Paris26,Dubai45, Rio67, now Budapest70!) did not speak to quality, and looking at it now, I can see why Jeff took such exception to it. It was intended to do exactly what it did here, which was start the conversation about what has been going on over the course of 2010 regarding massive panoramas. I think it is not simple coincidence that each megaimage has upped the gigapixel ante and that the gigapixel count has been highly promoted as a defining point of the image (including your beautiful Prague image which is still promoted on this very page as the “World’s Largest 360 Photo” and the gigapixel count stated in the first line.) Calling the work of other photographers “disgusting” seems a bit harsh, and as Rodolfo points out, it was only the first stitch.
The fact is that there is very good reason to promote the pixel count on these images as part of promoting the work and I don’t fault anyone for it. The attention being received from each of the contributions is increasing awareness for high end panoramic photography and will inevitably lead to more work for all pros involved and will lead to more amateurs like myself getting involved. Of course, quality matters, but the public is not picking apart the images for stitching errors or lack of vignetting correction, and they don’t seem to find the images boring. They are exploring them and sharing them with their friends, and for good reason, because imaging on this scale is so new, the ability to zoom in at such distances is mesmerizing.
It seems to me that what is called for here is organization and sponsorship. I suggest a formalization of the contest that is obviously happening anyway over the next year. Let anyone who is interested participate to create an image over 100 gigapixels and allow the world (not just photographers) to vote on which one they like better. Let sponsors market the contest and award a worthwhile prize and let the chips fall where they may as to the subjective quality. In the end, awareness would be heightened and a few standards just might be settled upon.
Thanks to all of you who are creating these amazing images and developing exploratory software. I, for one, am enjoying them all.
Shameless Plug: Your work inspired me to attempt to take advantage of a rare opportunity to run the Grand Canyon and attempt to create a virtual tour with very limited resources. I brought a GigaPan Epic 100 and managed to take 28 images at various locations along the river corridor. Not as many as I had hoped, but an 18 day river trip is a lot of work, and exhaustion played a role. The images are now available in Bing Maps via the Photosynth App by zooming in on the Grand Canyon in northern Arizona and are collectively called Grand Canyon GigaView. They can also be loaded into Google Earth via my Tumblr blog http://grandcanyongigaview.tumblr.com/ or explored through the embeds there.
I would very much like to get these images into 360 Cities, but the uploader doesn’t seem to recognize them. Also, I would appreciate any advice on other platforms that my showcase the images better, including Microsoft’s HD View. Thanks in advance for your input.
I would like to make some comments about your note.
First, let me say that, on one hand, I totally agree with you in many respects.
1) To begin with, I also suspect that we are reaching the limits of resolution in digital photography. As a matter of fact, I was already preparing a talk on this subject to be presented in our graphics seminar at IMPA.
2) I share your opinion that the “megapixel” trend of digital camera manufacturers is a bit misguided. More specifically, the goal shouldn’t be ‘more’ pixels, but ‘better’ pixels instead – and surely there is a balance for the optimal solution in the design of image sensors.
The same observation that you made about the recent evolution of Canon cameras was pointed out in a seminar lecture at IMPA on the Future of Photography by our colleague Ricardo Motta from Pixim .
3) I think so too that quality is much more important than quantity. As a consequence, in my opinion, the “gigapixel race” has mostly a relative technical value.
Now, on the other hand, I feel that some of your comments are somewhat unfair.
1) By the title of your note, you seemed to imply that we are charlatans. This is a very strong word, and if you really meant that, I do not understand why. We just posted a panoramic image on our page at the gigapan.org website – and with this we didn’t try to fool anyone. In truth, regarding the record, we only included a comment saying that we believe it was the world record – nothing more that that.
2) You also seemed to suggest that our panorama “Corcovado 67 GP”, is not only BORING, but DISGUSTING (yes, in capital letters). While I respect your opinion, I do not understand why you had to sound so offensive. Actually, we never made any claims that the image was beautiful or even interesting…
3) Your comments also assumed that the posted image was a finished piece, like a picture submitted to an Art exhibit. But to us, that image is more like a “work in progress”, as you can deduce from its complete title Corcovado 67GP (first stitch) and also by the companion descriptive notes.
4) You also mentioned, that the image does not look nice when zoomed out, because of vignetting effects. We cannot deny that, but it is something we hope to fix in a future version of the image, as explicitly mentioned in the description note, which you probably didn’t read.
5) Moreover, you said that the contents are barely recognizable when the image is zoomed in, and you produced a demonstration that the ‘right’ magnification should be about 1/4 of that size. Your experiment is perceptually correct, but I imagine that you are not aware of the fact that the gigapan viewer magnifies the image by a factor of 3x.
You asked what we are trying to prove (PS: this is a very serious thing for a Mathematician). I would like to assure you that we do not want to prove anything – not even claim a world record.
Our endeavor in this project is mostly a technical exercise to experiment with gigapixel panoramas. We strive to understand the issues and limitations of capturing and processing very large images, to confront the challenges and eventually solve the associated problems.
I guess, an unexpected consequence of our project was to spark a discussion about this topic, which I believe is timely and could be beneficial to the whole community if taken open hearted, powered by constructive criticism. Based on your reply to Rodolfo’s comments, I am glad to see that you are likely to be heading in that direction…
Pretty great blog. I just stumbled upon your website and wanted to say that I’ve really liked browsing your web-site posts. I’ll opt-in for your feed and I hope you write again soon!