vertical camera orientaiton

[OPEN_RANGE]

Oliver, in no way am I "trolling" you or anyone and I apologize if I have given you that impression. I'm not sure what your asking for. A max scene with a portrait photo and a cube that would represent a building that is in the photo?

[kosso_olli]

So I just shot a simple picture of a building in portrait mode with my phone. Let me get back to my workstation and I will try to explain.

[theedge]

Here are some places where this was discussed, not only 3ds max related :

http://forums.cgarchitect.com/74895-...-portrait.html
http://forum.vrayforc4d.com/index.php?threads/13810/
http://forums.cgarchitect.com/22309-...gital-slr.html

Definitely not a troll

[kosso_olli]

Okay, first of all: Sorry if I sounded a little rude in my last message.
Alright, here are three images. I still am not sure if I understood the underlying problem correctly. You basically want to match a building in 3dsMax to a photo of a building shot in landscape mode, right?

Original photo:



Quad view:



Render Cam maximized:



I roughly measured the brick building on site, it is 2.5m x 2.5m square, approx 7.5m high to the base of the roof. The cube has the same measurements and matches quite closely. Is this what you are after?

[OPEN_RANGE]

meneke - thanks for the links they do discuss the issue I am asking about.

Oliver I think maybe typo in your message? (I'm asking about a photo shot in portrait orientation, not landscape as you have written above.)


I am going to match into a portrait photo of an existing empty site with buildings on either side. I will not be replacing/masking an existing building as you have shown. Rather I will be adding a 3D model of proposed building that must fit between existing buildings and accurately show the 3D building in context and in relationship to the existing buildings per the parameters of the real world camera/photo. I will need to have the max camera located in the same position, relative to the existing buildings and site, as the real world camera position. The max camera focal length must also match the focal length of the real world camera.

Does your max camera match the position and focal length of the photo you shot with your phone?

Maybe I'm not getting it. I "think" the discussions in the links by meneke and the suggestion to change the film sensor from 36 to 24 answers this. I will have to confirm at a later date when I start this and get my own photography done. Probably be a few weeks. Thank you for your time and help with this!

Photo - by others - attached of the site and similar to what I will shoot under better conditions

[kosso_olli]

Does your max camera match the position and focal length of the photo you shot with your phone?

No idea, the sensor size of my smartphone is unknown. It doesn't matter actually, as long as there is a reference object with known size in the photo and the 3D scene (cardboard box, or wire cube). The rest can be eyeballed afterwards and will be correctly scaled that way. I have done this with automotive clients, and they are very picky with every millimeter that seems off to them.

[OPEN_RANGE]

you can find the sensor size/conversion factor for an iphone by looking at file info in pshop and by searching google.

For arch viz all of these factors must be accounted for. Esp when going before design review boards in national historic districts (as this project). sometimes surveyors are hired to create CAD site plan with camera location and elevation surveyed and included along with existing building footprints, floor/ridge elevations and a benchmark elevation. Some companies even offer "verified" photo match. I will be required to include some backup showing that my camera match is accurate to position, aim point, focal length.

Reference objects, like existing buildings, are critical. If there are no existing reference objects I stand up, usually one each at 3 proposed building corners, 10' lengths of 3" pvc pipe and shoot the base elevation of them and measure their horizontal locations and add that into a site plan which is incorporated into my max scene. Esp on steep hillsides, common to the mountainous region I typically work in. Eyeballing it will not be accepted (and will generally result in an not accurate photo match).

I have actually written a coupl eof blog posts about this for clients and architects to reference:
https://www.openrangeimaging.com/blo...tion/#more-406
https://www.openrangeimaging.com/blo...aphs/#more-187

thanks again!

[kosso_olli]

Anyway, matching 3D objects to photographs in portrait mode was never an issue for me or my clients. Results are close enough to be believable.

[OPEN_RANGE]

your auto renderings are excellent!

be thankful you do not work in arch viz where "believable" is not the same as accurate. If you get it wrong, which is easy to do, and it gets built and it doesn't look like your rendering, trouble ensues and maybe you don't get any more work...or worse. Approval boards and concerned neighbors are requiring visualization that accurately depicts a project in context and in relation to surrounding structures and environment.

[Neilg]

oh my goodness how is this still going on.

basic maths: If you took a photo on a 24mm canon lens the field of view from camera sensor left to right is 74 degrees. If you took that in portrait, that becomes your vertical field of view. you are aware you can type this number in directly to the camera, and switch between typing in the vertical, horizontal, or diagonal field of view?

Like this:

The bottom one produces an identical field of view as rotating the camera

[OPEN_RANGE]

Sure it matches the field of view. aspect ratio. W x H

unfortunately that will not work for purposes of matching real photo unless you 1) change the focal length of the max camera so it is not the same as the real world camera, or 2) change the width of the film sensor, which I think is the best solution.

not trying to be argumentative. read some of the links meneke provided above and you will see what I mean.

[Neilg]

It's also possible to do even more primary school level maths outside of 3d software to do this.

If you shot something at 70mm in landscape at 1920 x 1080 pixels, divide the width by the length to find the ratio (in this case 1.7778 ), then times that by your current lens width in mm to figure out what it should be in portrait. IE: 124.4 mm

[Neilg]

unfortunately that will not work for purposes of matching real photo unless you 1) change the focal length of the max camera so it is not the same as the real world camera,

yes????? this is the correct way to do it???
it is not the same as a real world camera anymore (and is not supposed to be). it is, however, mathematically identical to the results you would get from turning it on it's side.

[Neilg]

Can I also add that I finally looked at the 4 links that have been posted as proof where this has been discussed before, and every single one of them very quickly contains the actual solution and the correct way to do it.
The first link by open range has the solution in post #5, and all 3 cg architect links had the solution in the second post. Why was none of this good enough?

[OPEN_RANGE]

I found the math discussion to be somewhat confusing, but respect that it is primary for you. For me changing the max camera film sensor setting from 36 to 24 seems the best way to go. Then the focal length stays the same as it is in the actual photo and no math required. Thanks to again to mekene for putting that forward.

sorry for any inconvenience or aggravation I may have caused, and that I did not suggest you look at those links and then could have avoided this. my apologies.