Claudio Branch

It is a dog pile!

But justified I believe...

Images are the heart and soul of this industry, not talk.

Brian Cassil

Not to pile on... but worth mentioning perhaps is that 1/10,000 of an inch would be far, far, FAR less than a single rendered pixel in an image. Even if your image was 10,000 pixels wide.

__________________
"writing notes is quieter."

alexg

This VIS seems like what the guys at Hayes Davidson describes in Accurate
http://www.hayesdavidson.com/html/pu...lications.html
http://www.hayesdavidson.com/html/pr.../planning.html
Smoothe I think is doing similar thing with St.Mary's Axe building.

We have been doing quite several projects like this, but when we asked if we are willing to guarantee 100% accuracy, we always say no.
Not that I cannot model accurate 3d from the architect, but it's how the actual photo (before and after) is taken that has questionable accuracy.

From my point of view, one will need 1 gps coordinate, down to the cm of the exact position of the camera, the lens / camera body / tripod height, and the exact angle of the camera target to be regarded as an accurate reference point submissible by law. A single shake by any vehicles / ground around the camera has been proven capable to alter the angle. During high wind, tall buildings also sways, disturbing the ref.points. If the photo is a stitched panoramic, then the stitching process itself has to be legally documented and will not allow any deviation of the individual photos.

Is there a documented process that will stand against legal lawsuit in VIS ?
Claiming 100% accuracy can be disastrous if facing against the best lawyers.
If 100% is not possible, then what is the % that is considered passable under law? VIS might be expensive, but having hundreds of thousands $$$ of building lawsuit for tens or twenty grand $$ fee is very scary for me.
Will our insurance cover for a mistake of 1/1000 inch difference on the photo match? Will it cover for 1/9000 inch? There has to be a legal threshold.
Can we claim 99% as accurate enough?

Maybe Jeff can help out, I have seen the render of St.Mary's Axe before, but haven't seen the actual photo of the finished building taken on the exact same coordinate. Is it 100% the same?

Unless there is a legal documentation of procedures and has been passed on by law, it is too risky (legally) claiming 100% accurate.
Attempting for at least 80% accuracy might be ok, but 100%? I know some lawyers that are just more than happy to challenge this.

__________________
Alex G
www.3da.net.au/www.3designarchitect.com

alexg

One more thing.

Do you need an expert witness signature when taking the pictures? To ensure the proper documentation process has been properly done?
Can a documented movie of your process of drawing be submitted to court as evidence? or can you just hire an expert witness for their input?

Sorry didn't see your other thread on the camera positioning. But my question still stands.

__________________
Alex G
www.3da.net.au/www.3designarchitect.com

mskin

i wouldn't call it a dogpile... just a lot of interest in the subject.


where did Virgil go?!?!?!?

Claudio Branch

The dog pile has turned into a technical and legal morass about theoretical images thus far.

Are they even interesting to look at? Who knows, who cares any more...

nicnic

What is this guy on about? rambling confusing rubbish.

Alot of the work the company I work for is 'visual impact studies'. We call them 'align views' and use a internally developed and legally verified process and expert evidence to prepare verifiable studies of building in context.

Without going to far into it it is a very involved process that involves surveyors, professional photographers, accurate modelling, materials and lighting. This is done completed in 3Dmax, autocad, and sometimes Archicad using a variety of techniques.

Our images have held up in over 100 VCAT sessions (Victorian civil and administrative tribunal). VCAT is the body that asseses and deals with built proposals and complaints. As far as I know our company is the only visualization company in Australia that currently produces verfiable images.

Align View images are significantly more expensive to produce due to teh extra consultants and documentation produced. However it always gives a 'correct' result and is alot easier than trying to match perspective by eye which is very difficult.

The examples on our website arent the best ones for some reason but give you an idea of the quality. Here is a recent one I completed for a small commercial development.



The before image can be found on the website. I will be updating the align view section shortly to include some bigger more impressive buildings and developments using align view.

http://www.orbitsolutions.com.au/align.htm

nicnic

(OT. wow alex i just checked out your website, very nice images!)
(hayesdavidson stuff is very impressive as well)

alexg

Thx Nic

I am interested to know which professional photographer and expert witness to contact if I am to produce 100% accurate verified VIS. If I am to produce a project that may invite a legal lawsuit, I want to feel safe working on it.

Once we have a project in which requires accurate VIS, and the photographer provides a stitched image. My client require 100% accuracy, but when I ask the photographer how he stitch the image, he replies that he does it by 'eye' only on photoshop. Lots of skewing and distorting to get the image together.
That, I think is pampering with the data. How can I be accurate if the raw data isn't

__________________
Alex G
www.3da.net.au/www.3designarchitect.com

nicnic

You cannot use stitched images with camera matching / align view images. It has to be a single shot, generally more than 20mm lens also produces barrel distortion on the outsides of the image so its best to keep the building as centered as possible in the image to avoid misleading distortion.

Images shot from far away can also be problematic.

Having a sound reproduceable methodology is the key to verifiable images. There are numerous ways to produce verifiable images. It is mostly about eliminating as many variables or areas for error as possible and proving through process that the degree of error in your image is small enough to be called 'accurate'.