Nathanael.Lawrence

If you need to contact me, please feel free to write to me at natelawrence@live.com

Cheers!


Mt. Hood Pavilion
Mt. Hood Pavilion
amn.hodge (Over 1 year ago)
Reminds me a lot of the church at fort leonard wood

more...
45th Place NE As It Once Was
45th Place NE As It Once Was
michaeldenis (Over 1 year ago)
Looks good, Nathanael! The overhead view suggests that you have a strong 3D model waiting to be processed here.

amn.hodge (Over 1 year ago)
I agree with and like the idea of moving in arcs down the street. It's actually unneccisary to make a full arc (Three and sometimes two pictures suffice) I've done it myself before.

Nathanael.Lawrence (Over 1 year ago)
I also thought of a different technique for shooting the shots that look straight down the street as I was finishing this shoot that I'll have to try next time. It basically breaks down to a bunch of arcs from one side of the street to the other, keeping focused on one end of the road or block as you do so, rather than the method that I used here of disconnected straight paths up both sidewalks and back as well as straight up and down the middle of the road. It seems like it might provide a more reliable navigation experience as well. Trying to move around in this version is pretty awful as far as the photos are concerned. Free navigation is pretty fun in pointcloud mode, though, especially, as you say, in the Direct3D viewer.

Nathanael.Lawrence (Over 1 year ago)
Thanks, Jim, I appreciate it. =) This is one of my favourites too. I'm waiting for the day when this will automatically link with my 'Orbital Propulsion...' and 'Obligatory Hydrant...' synths (or when the Bing Streetside car drives by and its 3D scanner information is combined with my synth's pointcloud). I do find that focusing on one object at a time leads to far better point clouds but sometimes I just want to see an idea of what everything would look like linked together. It can't hurt to try a big synth, right? There are a fair few photos which didn't synth in which I've left included in this version of the synth, as learning what does or doesn't work is often more valuable and interesting to me than having only 100% synthy versions listed publicly. That's just me, though.

jimcseke (Over 1 year ago)
Wow ! I might be to old to say this, or do they still say this? "Cool....really, really cool". In the Direct 3d viewer, alot of the street scenes really stand out beautifully in the point cloud. Nice work ! This one is a keeper for me.

more...
Memorials at Oregon's Capitol
Memorials at Oregon's Capitol
NateLawrence (Over 1 year ago)
The updated version is here: http://photosynth.net/view.aspx?cid=7993c73b-fefe-4176-9987-24f92df2eab6 I had to sacrifice the photos of the greenery surrounding the Circuit Rider statue to avoid running out of memory, but otherwise I'm pretty happy with it, given the current constraints of how many photos can be placed in a single synth.

more...
2010 May 12 Chemeketa Car Show 3|3
2010 May 12 Chemeketa Car Show 3|3
hotrod69dfw (Over 1 year ago)
Hi Nathanael.....I'm having a big problem with my photosynth and was hoping I could get you to help me.....when you have a chance, will you email me at nivekyesrek@aol.com? You seem to be the expert around here. I'd be happy to compensate you for your time.

more...
2010 May 12 Chemeketa Car Show 1|3
2010 May 12 Chemeketa Car Show 1|3
Nathanael.Lawrence (Over 1 year ago)
Wow. Those quasi panorama shots from the middle of the lawn really sent overhead view for a loop. If you're looking straight down (tap the [T] key) zoom into that little cluster of points in the middle of the screen.

more...
2010 May 12 Chemeketa Car Show: # 31
2010 May 12 Chemeketa Car Show: # 31
Nathanael.Lawrence (Over 1 year ago)
Thanks, Tony. This one fell apart the very worst of the lot, but I'm glad that the dash and hood at least made it into the main synth.

TonyErnst (Over 1 year ago)
I love how you can see the engine and dash through the exterior of the car in point cloud mode. Well done

more...
Oregon Veterans Medal of Honor Memorial
Oregon Veterans Medal of Honor Memorial
don.d.kim (Over 1 year ago)
No problem. I'm just another photo enthusiast here. Its always interesting to hear how other people take photographs, and it seems like you've definitely found an effective method.

Nathanael (Over 1 year ago)
I realised right after I posted this that I was telling all of the above to a Photosynth team member. Whoops.

Nathanael (Over 1 year ago)
Thanks for the kind words. =) Although circling things is nothing new, circling everything is sort of my signature (unless I'm trying to synth something huge) and the pointclouds almost never fail to impress. There are perhaps more quick shortcuts to getting nice pointclouds, but I circle everything so that other people's photos will be able to quickly latch onto whatever I'm synthing in the future when we are able to link our synths together. I like to capture the structure of things from as wide a variety of angles as possible so that others' photos have a nice home to nest in. If I do my job right, the computer is likely to quickly recognise any part of the environment in their photos because I have described it so thoroughly with mine. When I finish with an object, there aren't too many angles left that anyone could shoot it from that would be too dissimilar to mine in order to match.

don.d.kim (Over 1 year ago)
Great job! It very nice that you took the time to carefully circle every object in this scene and it really paid off in the result. Overhead mode and the point cloud are particularly impressive on this one :)

more...
The Golden Pioneer
The Golden Pioneer
Nathanael (Over 1 year ago)
The Photosynth web service (whose address is in every photosynth log file) is also valuable for any would be developers of custom synth viewers or mashups. You can find direct links to all the necessary bits in the following link, if you care to pass them on to someone who can make use of them. http://bit.ly/photosynthdevhelp As to shooting time, the hard drive with the original photos on it is out on the road at the moment, so I can't say for certain, but I would be surprised to find that it was any more than ten minutes for this little synth. There was some indecision about how far to back away from the building when passing the first cluster of trees, but once that was settled, it was straightforward again, apart from waiting for traffic to slow to take the rear photos from across the street and back again to complete the circuit.

michaeldenis (Over 1 year ago)
I agree with you. (good info) Nice synth, by-the-way, the spin around your statue is highly fluid. That's tricky to achieve It must have taken a chunk of time to make.

Nathanael (Over 1 year ago)
Also, Michael, I'm really not sure if we'll see the Photosynth team make a custom viewer for Android, but Seadragon's DZI and DZC formats are both fully documented on MSDN and I'm sure that Christoph_Hausner and BinaryMillenium have both figured out the camera coordinate and metadata system, in addition to how to read the pointcloud coordinates and color information, so a custom viewer should be well within someone's ability to program, rather than waiting for Silverlight to be ported to Android, etc. iSynth on the iPhone uses OpenGL and although the author was an intern with the Photosynth team and has now joined them, the iSynth viewer was essentially a third party viewer. I'm sure the same could be programmed for Android, given the desire and competency.

Nathanael.Lawrence (Over 1 year ago)
Unfortunately it's impossible to get good shots of this with my camera's zoom lens and literally impossible to completely circle the sculpture from the ground due to the surrounding trees. I'd love to fly a drone around this, though.

michaeldenis (Over 1 year ago)
Nice thumbnail, Nathanael, you really made me want to see this one. Good to see you synthing again! (Unfortunately, I'm posting from a smartphone running Android: no viewer yet. :( )

more...
Salem Capitol Building Exterior 1 of 2
Salem Capitol Building Exterior 1 of 2
jimcseke (Over 1 year ago)
Thank you very much Nathanael, You have always come through for me. I want you to realize that I absolutely appreciate your insight and assistance. I added you to my list of good and considerate people from day one. Above and beyond as always. Thank you again.

Nathanael (Over 1 year ago)
I'm not sure how long that particular community will last over there, because of the limitations (no editing, lack of synth embedding, etc.) as well as the fact that it looks like Ning will be moving away from offering free community sites very soon (May 4). Nevertheless, I have spent my spare time the past several days embedding many of the videos of various Photosynth team members talking about Photosynth - how it works and what plans there are for the future. I've also posted a number of videos of related research projects that may be of interest to the more avid photosynther. Assuming that the community doesn't disappear when we decline to pay Ning for hosting, come on over and make some posts. I get the feeling that if Ning goes away as a viable option for us, we'll find another venue, having now gotten a taste of the expanded conversation a Photosynth fansite has to offer.

Nathanael (Over 1 year ago)
Hi, Jim. No worries. I actually haven't been on the site so much the last half a week. I do hope that you took the time to at least look at the links to the videos that I posted as it took some time to find those timecodes. As to a good place to get the word out, I like vimeo as a great alternative to YouTube, but it doesn't have quite as big of an audience. Gary Mortimer recently started a Photosynth group over at http://photosynth.ning.com and it has a few of the more knowledgable suspects over there, that I'm sure would be up for a discussion. Sadly, the site deletes synth embed codes before it saves your posts (something I'm trying to work out with them at the moment), so the discussion is limited to text and videos, but you can directly message people, so it is a little bit better than our forum, although you still can't edit a post except for the 15 minutes after you first save it.

jimcseke (Over 1 year ago)
Hey Nathanael, I hope I did'nt give the wrong impression by deleting much of the comments. The pages were getting way too cluttered with replies and comments. I thought that a very important item was missed during our conversations, and I felt it sort of got buried under our back and forth. When my methods are used correctly,the captured images are completely editable. Other than possibly a distorted capture, or perhaps messing around with the actual colors(which tends to remove many potential matching references). Having mentioned that, where is a good arena, other than Photosynth or even Youtube, to pass on my information? I feel that, the larger the number of individuals at least trying this, might provide a better overall test under all types of synthing. If you could suggest such a site, it would be much appreciated. Sincerely Jim

more...
Salem Capitol Building Exterior 2 of 2
Salem Capitol Building Exterior 2 of 2
jimcseke (Over 1 year ago)
Well I guess I assumed wrong, sorry about that. I thought that you had started to release the test you discussed, so I examined the photograph to try to determine which was photographed using the level method. 2 of 2 look to be the most level to me, so I assumed that this was the side by side test you were going to perform. Oh well, I guess I'll just wait for the actual test you discussed earlier.

Nathanael.Lawrence (Over 1 year ago)
Actually, no, these were shot last Tuesday (2010 April 20) for PhotoCity. 1 of 2 is front and 2 of 2 is back. I'm hoping that Photocity will be able to fit it all into one model, but I've hit the limits of the 32-bit synther, so have had to split it into two synths for Photosynth. These were shot, however, on a tripod which has a built in surface level which I occasionally borrow from my dad. This was basically because I was abandoning orbits for once and going more with a crawl along the surface in the vein of alFaku's synths, just to see how it would work and to try to cut down on the number of photos. Honestly, though, due to the uneven ground, the tripod actually made my shots less level than they would be if they were just handheld, but at least the vertical sets of four all matched the same angle.

jimcseke (Over 1 year ago)
Oh, by the way, did you use your DSLR camera for this shoot. If you did, I am curious as to how you got past the problem with your view finder and all.

jimcseke (Over 1 year ago)
Hey Nathanael, I am assuming that these two synths are the test. I appreciate you taking the time to do perform these side by side comparisons. I am also going to assume that "2 of 2" was photographed using my level method.Having photographed in this manner for so long now, I can quickly recognize what a "leveled" set of images looks like. Good choice of subject matter. A good representation of the synther's ability to reconstruct an object, given the data provided.I think that you have provided an excellent comparison. At this point my opinion is meaningless. By getting to know you, through our online discussions,I am positive, that you closely examined both final synths. You are a very methodical person. I know that your final opinion will be unbiased and well thought out as always.So, after examining all of the data, as meticulously as I am certain you have,(I hate to quote this line)"what is your final verdict?"

more...
Oregon Capitol Building (Front)
Oregon Capitol Building (Front)
jimcseke (Over 1 year ago)
here is the link if you have not seen this "synth" already" http://photosynth.net/view.aspx?cid=018fd774-d6f8-4beb-ae10-9b275b8c104f

jimcseke (Over 1 year ago)
Hey Nathanael, I was glad to see your comment as part of the discussion in the forum. One of the things I have always admired and respected about you, from almost day one, is your willingness to be open-minded about new possibilities, and you have never ever dismissed any of my suggestions or ideas. your input has always and still is valued. The fact that you are willing to at least "see for yourself" has always been one of the things that I admire about you.I would not expect any less or more from you. You do not blindly dismiss, but you do not blindly accept. And you back up what you think by thinking before you let your opinion be known. In my life, I have not met too many people with that same positive trait. I am going to leave you a link which pictorially shows exactly the materials you need for your up coming test. Thanks for the nice welcome back, as I have missed being away very much. This has not been a joy ride for me, and now I can share all.

jimcseke (Over 1 year ago)
jimcseke (Over 1 year ago)
Hey Nathanael, it's Jim. You've probably notice that I have been gone for a it. I have started a discussion in the forum, which I think you will find interesting. It is a long comment,but when you read it, you will understand why it had to be as long as it was. I know I have the tendancy to digress, but I think I managed to stay on topic fairly well, for the most part. Take the time to read this comment in it's entirety, as I feel that it is important that you do. Have a great weekend Jim

more...
Duplication Test: 1MP Tiles 50% Overlap -Mid Scale
Duplication Test: 1MP Tiles 50% Overlap -Mid Scale
Nathanael (Over 1 year ago)
Minimal overlap may be inevitable, but unless I am mistaken there is a parameter in Seadragon's tools to specify the amount of overlap between tiles. Time will tell. I tell you what. I really do wish that either the synther just had an option that allowed me to have it analyse the original photos as originally provided for image features, rather than all of this cropping nonsense. We've even seen from sneak peeks at the work that the Photosynth team is doing as well as the GRAIL lab at University of Washington, that shows that dense pointclouds are the next step before a mesh can be created and the features (whose estimated positions are what we currenly see represented by the points) can be textured onto the mesh, so we know that Photosynth of the future will certainly afford us more dense pointclouds but it's all a waiting game as of now. This ugly cropping cheat seems to be the only way to get around it for now without knowing|being a programmer.

Nathanael (Over 1 year ago)
As observed above, the synther seems reluctant to find image features which lie on the edges of tiles. Certainly in my tests I will not (as I did in this synth) have tiles from the same image overlap in the same manner. (Multiple tilesets at the same scale from the same image = a big fat negative for real synthing. We don't want self-verifying, self-sustained features and therefore points from within the same image.) It does, however, make me wonder whether a slight amount of overlap on the edges of each tile within an image's single tileset might be useful. I am still somewhat uncertain as to how well the individual crops will match any other photo's crops as they will all be automated and therefore cut many things in the photos in half that you would keep together if you were cropping by hand. Will finding half a surface in one tile be enough to match it to the third of same part of that same surface in another photo's tile?

Nathanael (Over 1 year ago)
Due to Jim's experiment, though, and due to sustained interest by myself and my fellow pointcloud aficionados and fanatics, I plan to try to nail down some good, specific but generally applicable practices for choosing crop sizes, suggestions for programs to batch create tiles for a given set of images etc. during this weekend. Using the Seadragon team's tools for converting images to Seadragon's Deep Zoom Image format should actually work, as I can specify the tile size and compression quality. Once every image is converted, I'll only need the tiles of the largest (read original) resolution from the image pyramid to combine in the synther with the original images, but that should provide an excellent demonstration of my theory, should all go well.

Nathanael (Over 1 year ago)
I'm still a firm believer that for more *meaningful* points that actually follow things' correct shape you need more distinct perspectives on things, not the same perspective rotated around in the computer (which has its own problems like the fact that pixels can only be rotated in 90 degree increments without making an absolute mess of them), but I was happy to find something I liked about Jim's rotating. The cropping idea has been rolling around in my brain since the first week or two that I used Photosynth, but I hadn't ever done much with it except to glue some wide shots and close ups together, even though I knew that it could make the entire pointcloud more dense if you used crops of the correct size(s). The big holdup was mainly due to the fact that I (to this day) don't know what rule the synther uses to decide whether to downsample an image before inspecting it for features... I don't know the magic number that describes that border.

Nathanael (Over 1 year ago)
Two things were happening at once. Jim was cropping in (allowing the synther to see features it normally wouldn't by using its small versions of the original images) which is good. He was also making Photosynth keep parts of each image that could only be found in itself. These points, although kept in the pointcloud, would never be properly arranged in 3D, because no depth could be calculated from a single image because nothing moves relative to anything else. This was bad. I was a bit unhappy to tell him that I didn't think that the rotating could be much use. Then I got to thinking. Couldn't you use that sort of technique on things that are *meant* to be flat? Things like an information plaque, if photographed head on, could be cropped out and the crop rotated all the way around so that its part of the pointcloud would be super dense. It would be flat as that sort of technique will always give, but since the original object was flat there's no problem.

more...
Duplication Test: Same Image, Compounded Artifacts
Duplication Test: Same Image, Compounded Artifacts
Nathanael.Lawrence (Over 1 year ago)
Contrast this synth with one using identical pixels three different times: http://photosynth.net/view.aspx?cid=36ea0c15-bde3-4427-b783-e491dd41ee8a

more...
Duplication Test: Same Pixels, Different Tags
Duplication Test: Same Pixels, Different Tags
Nathanael.Lawrence (Over 1 year ago)
more...
PitPit Self-Occlusion Test
PitPit Self-Occlusion Test
Nathanael.Lawrence (Over 1 year ago)
Having no bamboo nearby and having noticed this sort of plant several times, I decided to take a stab at it, wondering whether the tufts at the top would be feature rich enough to cluster the photos satisfactorily. In the end, I kept my shoot pretty brief and didn't get too many close shots. I think that the leaves around the base in the majority of the shots certainly helped. 100 photos certainly wasn't enough to capture the individual blades of the grasses but the neighboring dog was not appreciating my interest in the corner of his|her territory, so I cut my visit short. =) To answer your question explicitly, the blades of grass are occluding each other as a group, thus the bunch of grass can be said to be self-occluding.

Nathanael.Lawrence (Over 1 year ago)
Occlusion in an optical or a computer graphics sense simply means that a foreground object, when standing between the eye|camera|viewport and a background object, obstructs the view of the background object - that is to say, the foreground object occludes the background object. In the Photosynth Photography guide we are given a list of synthy and unsynthy types of objects and are given the contrasting examples of a single large tree trunk versus a bamboo grove. The bamboo grove is advised against simply because the stalks are all so small and tightly packed that they each occlude each other very often, leaving the synther with little detail to hold onto. A bamboo grove is not impossible to synth, but it certainly would take a much higher number of photos to capture individual stalks than the average novice Photosynth user is likely to take.

jimcseke (Over 1 year ago)
Oh, and very beautiful images and pointcloud. :)

jimcseke (Over 1 year ago)
Sorry Nathanael, but I tried to find a defintion of the term Self-Occlusion a while back. If it's not to much bother, could you explain the term?

more...
Sandy's Cake
Sandy's Cake
Nathanael (Over 1 year ago)
Frosting is definitely something of a mystery to me still. Why some of the photos which seemed to be equally spaced transitions to me suddenly failed to synth was somewhat mystifying to me. This was the fourth or fifth attempt at getting this to be 100% synthy by subtracting photos which didn't contribute to the primary cluster multiple times. Unfortunately some of them must have been linked in some way still because each time I took the stragglers out, more splintered off. When I can find time to resynth this and get rid of that last orphaned photo, I have five or six shots taken by the baker and cake decorater (the first in the filename series) which I'd like to set as highlights but until I can attain 100% synthy, I'm reluctant to put in the description, etc.

David-Photosynth-Team (Over 1 year ago)
Reminds me of our Photosynth birthday cake!

more...
In Search of Efficient Motion
In Search of Efficient Motion
Nathanael.Lawrence (Over 1 year ago)
The challenge with a structure such as this is to provide coverage that will clearly define each major line in the bridge. They all spend so much time overlapping and there are just so many of them that the # of photos that seems necessary to get a decent approximation of each quickly skyrockets into dangerously high #s for the synther to successfully process. Taking enough photos for closeups of each beam results in an unthinkable # of photos. Add to this that there are parts of the bridge which in structure are symmetrical and you have a full blown mess. The texture on these identical parts should be enough for Photosynth to tell them apart but I'm being forced to stick to wide shots, rather than go in for closeups, so the texture analysed by the time the synther shrinks the photo down to 1.5 megapixels for matching is negligible and I did witness several places where in jumps back and forth in the structure where I continued on a smooth course in real life

Nathanael.Lawrence (Over 1 year ago)
Shooting both sides should provide opportunity to flow far more smoothly with the coverage with a few less shots as well. I am also currently debating whether I should continue the method of getting a complete vertical pan at each position before moving sideways to capture that or whether I might meet with better success by simply using a continuous horizontal pass across the entire length of the bridge at each vertical angle, starting looking from eye level forward, moving to looking slightly upward, further upward, and straight up. One thing I know for certain is that I'll be packing a tripod this time. It would be amazing to be able to get shots from a boat on the river, providing semi-circular routes around each side of the bridge as well as full orbits of the supporting pillars, but I don't know if I can find a boat in time. The ultimate would be getting helicopter coverage, but that is extremely unlikely to happen as I am flat broke. =]

Nathanael.Lawrence (Over 1 year ago)
Hmm. There are large chunks of perfect reconstruction and other chunks of absolutely false positives. This bridge is being shut down to be repainted the beginning of next week. I'd like to capture it as it was before that happens and compare it to afterwards but I'm still searching for a way to systematically cover it efficiently but thoroughly. As it stands I only stopped shooting the day before yesterday because my battery ran out. My swivel technique provides great results but unfortunately is also very expensive in terms of numbers of photos for both my memory card as well as image features for scene reconstruction. As things stand I would have run out of space on my card soon anyway and I had only shot half the bridge, if that. Later today should provide enough data to tell me whether I can get a high enough quality reconstruction purely from the strafing parallax that I started experimenting with towards the end of this shoot.

more...
Tadmor Lake's Shoreline Circuit: Expanded
Tadmor Lake's Shoreline Circuit: Expanded
Nathanael.Lawrence (Over 1 year ago)
I actually am currently struggling to synth the same number of images together with another synth (Deep Wood House: South Lawn) because they overlap so much and each image has many features. The only real success I have had as far as getting the entire length of the shoot represented is to split the entire sequence in half by selecting only the odd numbered photos and the even numbered photos. All other attempts to synth even half of that set have crashed and burned for me so far. The problem with using every other photo is that there are some parts which snap apart because you are missing crucial transitional shots, so I'm planning on coming back to that synth later today - still using mostly just the odd numbered images (the more synthy of the two halves) but adding a few evens in weak points where the point cloud fractured before.

Nathanael.Lawrence (Over 1 year ago)
Oh, in answer to the last part of your question: there are plenty of problems with how it is currently synthed. =] there are at least 2 different reconstructions in the point cloud and rather badly misaligned as well. But... as far as running out of memory and such not really. The thing to remember is that it's *not* the number of photos that will run you out of memory or how big each file is. The synther only uses 1.5 or 2 megapixel versions of your photos to find 'image features' in to then match. My photos of the water in the lake or lots of sky have drastically fewer image features than the photos of a grassy hill or the like. What's more, unless the synther can match features to other features, you will not run out of memory in a synth where you cover lots of ground since each feature is only matched to itself in other images a limited number of times.

Nathanael.Lawrence (Over 1 year ago)
In any case, the coverage I got this past time was just not going about things the right way at all. I'll need to do some more musing and pondering before I head out to this location again.

Nathanael.Lawrence (Over 1 year ago)
Hi, Schn828, Yep, this was an overnighter. =] I'm using the cheapest iMac of the current lineup (the 20 inch display) upgraded to 4GB of RAM. Running Windows 7 RC (Build 7100) 64Bit Intel Core2 Duo @ 2.66GHz - 64Bit Processor (Does me no good since the synther is only a 32bit app) 3.37GB Usable RAM out of physical 4.00GB installed (32 Bit Synther can only utilise 2GB of RAM, I believe) Temp folder configured to an external 1TB drive 14.5GB available Virtual Memory (Apparently synther can only allocate 4095MB of it if the log file is anything to go by.) I really didn't know how to handle such a large subject as this lakeshore. There is no object that one can track continuously - things are always obscuring one's view. I could have tried just straight ahead shots the entire way around but I was concerned that I would run out of space on my memory cards and not have space to finish this: http://photosynth.net/view.aspx?cid=8e0c916c-307a-4af6-8e03-69d299050036

Schn828 (Over 1 year ago)
Nicely done. Let me guess, you let your computer synth over night? What are your specs, and did you ever run into the problems synthing it? Thanks.

more...