Photo Simulation

[Sawyer]

I had a client (civil eng) call wanting what he kept refering to as a photo simulation, the city wanted to see how much of the building was showing through the trees 1/4 mile away. The guy I talked with was very implicit on making sure it was exact, adjusting for paralax etc. and pointing out that being off on the parallax would change to roof profile by inches.

 

At first I thought no problem but then I began thinking about it and I wrote to him saying that I would need to stipulate the this is not a "simulation" (I have only heard that term used in legal situations where the 3d company provides a simulated car crash) but an illustration and basically not a legal document. I really haven't heard anything more from them, I thought the work would have been good but it sent up red flags. Anyone done work like this? Does this sent up red flags for others?

 

Cheers

[Chad Warner]

I remember a thread a few months back from someone in the UK offering "verifiable" photo matches that they could back up with data.

 

However, I would definitely never do that, just because I know what I can do is not verifiable, I just eyeball it half the time, and make sure my client knows this.

[oliviercampagne]

yes, this sounds really tricky ! any mistake could be dangerous...

[Nils Norgren]

Back a million years ago I worked for an Engineering Company in their "Advanced Visualization Department", we always called Photo-Montages (3d model matched in and rendered into a photograph) Photo Simulations. I bet this is where he heard this term used (the Engeering firm is a huge, world-wide firm).

 

As for the accurcy, I would tell him that if the model is in the middle of a high -quality photo, taken with good lenses, with information about lens length, shot location, and some context (existing buildings, streets, etc.) then you can provide a "Very" close aproximation to the actual view from that vantage point. If he is looking for an image that a lawyer can stamp as "an exact view" he needs to find an engineer to do the work (one with a professional license) there are no professional standards for such work therefore there is no legal basis for a claim of that kind of accurcy.

 

My 2¢

 

-Nils

[Sawyer]

yes, this sounds really tricky ! any mistake could be dangerous...

 

 

That was my thought at one time the client was talking about the roof line and how it would appear above the trees and that parallax would distort it by inches, & I began thinking of the architects I had worked for and how the exterior is often the least accurate of the drawings. It also just seemed weird to have a perspective need to be so accurate.

 

Again it just sent up red flags for me.

[Jeff Mottle]

They are called certified views in the UK and they are really not that difficult to do. However, you do need a survey site plan of the area and a surveyor to shoot some landmarks, or corners of buildings from the camera location. The camera location and lens specs are required as well. Once you have all that you will be able to easily put it all together in MAX.

[oliviercampagne]

it is also depending on the way they are going to use this "verification"... is it only to check the project, or something more "official". I found it tricky because it seems for me to be situed in the second case...

[oliviercampagne]

there are no professional standards for such work therefore there is no legal basis for a claim of that kind of accurcy.

 

... then i want to say go for it !

[Ernest Burden III]

there are no professional standards for such work therefore there is no legal basis for a claim of that kind of accurcy.

 

Well stated!

 

You can read this the other way, to mean no way to legally back up your product, therefore a liability.

 

I have done a few images like this (drawn, painted, pencil, if you can believe it), that were to be 'legal submissions'. But not under my company, under the architect who hired me, meaning their liability. Yes, I demonstrated to them how I set up the images, but it was their neck on the block. Make sure yours won't be.

 

But obviously it is possible to do such a rendering. Just get an upfront written list of requirements and make them provide whatever is needed, like Jeff listed. Maybe its just a word thing, and they just need a careful rendering. Make them be clear and specific.

[Sawyer]

Thanks-

 

I don't know what the outcome will be.

I don't think that there was that much info to make this really accurate, and I don't know how accurate they needed from me but it was for the city to show how much of the roof was showing through the trees. My thought was I would have to take the civil, combine it with the architects elevation and then piece that into a photo taken from across the river. The inflection seemed to be that this needed to be 100% accurate and that made me uneasy. I let them know that I can be as accurate as the info I am given I would have to say that I would not produce a legal document if that was what they needed. I haven't really heard anything back.

 

Basically I kind of feel like I chickened out. It would have been really tight with the other deadlines I have (they had under 10 days) but the challenge was there. The fear would have been that I hand over the print and then someone says "and you guarentee this is accurate right?"

 

 

Anyway thanks for the perspective

[Chad Warner]

I wouldn't worry about chickening out. Better to do that now than to have someone sue you later for mis-representing their project.

[Sawyer]

Ok but what I was given was the civil plan and sections through the site. The site is 70' setback beyond the top of the hill which is about 200' beyond the river. The view I was asked to show was from the path on the far side of the river. The question I have is how to create a verified picture from a location and to be able to pinpoint that in 3d application space? From a street corner or physical landmark it seems easier but this would be from a trail. I don't have any locator like GPs. All the other photo plate work I have done in the past relyied on measurements from a building and a bit of quess work.

 

Now my comment about exteriors came from working at architects firms and most of the people I worked with created elevations that were not accurate. Exterior elevations were drawn from STUD walls not finish walls. Certainly not everyone works this way but the majority of people I have worked with work this way.

[Sawyer]

Pinpoint accurracy is not as important as a representative view. Are all of the elements accurrately depicted visually so the client can make the decisions they need to make? That's the client's perspective.

 

 

Actually I have eyeballed images many times - the issue here is the accuracy. Thats really the point of the thread.

[tcorbett]

Actually I have eyeballed images many times - the issue here is the accuracy. Thats really the point of the thread.

 

With the parameters that you listed above, I think it would be hard to provide any reasonable promise of accuracy, given that you don't really know the exact position and direction of the camera. And since you probably don't have accurate measurements of the tree positions, that makes it even more unlikely that you could pinpoint a roofline. (Not to mention, trees lose their leaves in the winter, they grow, they get cut down...)

 

This raises a very good question about what level of responsibility do we hold when it comes to our images - and not just sightlines. What expectation is there for materials and colors? Lighting? Shadows?

[Sawyer]

I am sorry about going round and round on this honestly I didn't want the job but it raised questions. I see that being able to pinpoint a balloon in a phot knowing that that balloon was floated to a given height would help. But how could I find my location on a plat or civil cad file when there are no landmarks just a trail? It seems that if I cannot locate the camera location that was used at the site in Max to a really accurate figure within inches then I am eyeballing it.

[Ernest Burden III]

But how could I find my location on a plat or civil cad file when there are no landmarks just a trail? It seems that if I cannot locate the camera location that was used at the site in Max to a really accurate figure within inches then I am eyeballing it.

 

Without ANY landmarks you are just guessing. But a GPS device will locate you to within a meter or so (I think only military can get the inch accuracy). Maybe they have gotten more accurate? Anyway, GPS devices are not expensive anymore, so a job like this one would be your excuse to buy one.

 

But if you look out from the trail and see some structures that are also shown on surveyed drawings, you can look for lines of relationship. Like ' the edge of bldg A hitting the edge of bldg B' which will give you a plan-based line you can draw back to your trail. so you make a line at your feet in the dirt, move to where your photo is from and tapemeasure from the known point. This would be accurate to within only a few meters, but as an angle from the subject its probably enough.

 

Especially in the US there are ortho-corrected aerial photos available via the US Geological Society, many to 2 or 3 meter resolution, some down to .66 meter resolution. That is enough to locate not only your trail but individual trees on it, from which you can tapemeasure your photo location, key that to USGS topo maps and you have a highly accurate camera location. Add USGS Digital Elevation Model (DEMs) for the regions physical features as another measure of accuracy and photo-alignment and your cookin'.

[Sawyer]

Thank you everyone for the info, Ernest, Daniel. If this ever comes up again I will have a better understanding of how to do it.

[mskin]

whay not call a lawyer? im sure this is a no brainer, no fee, two second answer from on of those guys. who by the way are worth every penny they earn.....

[Andrew Schroeder]

I read this post on two days, so please forgive me if I'm re-stating what someone here has already mentioned.

 

Your civil files should be drawn in some sort of coordinate system. Likely (in the US) a state plane system. In a case like this, I have a surveyor pick up positions of reference/calibration points and camera locations. The cad file or DXF that the surveyor provides you with should be in the same coord-sys as the original civil files.

 

Which raises another question, which coordinate system is best to use in photo-simulation? Is a cartesian grid-based one better than a spherical LL with a geoid? Tough questions when you're talking about viewing an object at a distance. In the case of near-camera objects a planar cartesian system is fine, but what about simulating the view of mountains at a distance of +20km? Error at those distances is tough remove.

[ShaunDon]

Talk to a lawyer. Have him draw up a clause to add to the fine print of your contract that explains your work is a digital aproximation based on information from multiple third parties, and therefore you cannot be held liable for any inaccuracies that may subsequently be discovered. Hopefully you've got a multi-page contract to bury that in somewhere.

 

Yeah, it's sleezy as hell, but if you're confident enough to even attempt the task then your best effort will likely be well within human visual tolerance. And if he comes back in a year or two to sue you, you've got the contract he signed. It's still a gamble on your reputation, but it would suffer just the same along with your bank account otherwise.

 

This is all assuming the job pays enough to be worth the lawyer and the added risk. If it were me, and the client really was that anal, I'd probably pass on it just for my own piece of mind.

 

Good luck!

 

Shaun

[lrose]

I remember a thread a few months back from someone in the UK offering "verifiable" photo matches that they could back up with data.

 

However, I would definitely never do that, just because I know what I can do is not verifiable, I just eyeball it half the time, and make sure my client knows this.

 

http://www.hayesdavidson.com/html/home/home.html

 

This is probably the "verifiable" site you had in mind.

[Sawyer]

This is an old post but I didn't get the job. I was underbid - I think I came up with a really simple solution & I gave a qoute of 700 I think. It was really simple not "verifiable" & I was told I was underbid by a photographer who charged "Hundreds less". Not a loss to me.

[derijones]

Hi

My take on it is that as long as you have a paper trail that would allow anyone querying your work to go back and recreate the steps you took to put the image together, you're covering yourself. If you state in your support documentation the likely errors and factors that effect the accuracy of the image, you should be able to argue the validity of the image with anyone who queries it (which is what it boils down to at the end of the day). As someone else stated, there are no standards for this sort of work, so putting the image together in a "best practice" fashion with lots of paperwork should cover it.

I've done a couple of jobs like this for planning apps - neither of them has come back to bite me yet (both are rural developments with lots of opposition and the "other side" are also doing photomontages).

The biggest practical nightmare is getting the elevation of your viewpoint relative to the site - you can zero in the X,Y coordinates pretty accurately using bearings from identifiable locations, but getting the Z coord to even a metre is a pain - unless you've got one of the fancy surveyors GPS.

Cheers

Deri