topic Re: view cone angle vs. camera lens focal length in Visualization

!Restored: view cone angle vs. camera lens focal length

Anonymous — Thu, 11 May 2023 11:18:59 GMT

I'm trying to align an ac camera view with a photo of the site (set as the 3d window background); the ac align view function is not useful in this specific context as there are no adjacent existing (vertical) elements to use as references. I've been able to get the view relatively aligned manually, but have found the main variable that is affecting the accuracy is the view cone setting; so:

the photo information says it was taken at a 28mm focal length; how does this relate to the view cone figure? that is, what would be the conversion factor?

Re: view cone angle vs. camera lens focal length

Anonymous — Wed, 24 Sep 2008 02:37:51 GMT

Robs chart at the top of THIS THREAD would suggest 28mm equates to approximately 75 degree view cone, but I'm not sure its a s easy as that. Have a go though.

Re: view cone angle vs. camera lens focal length

Dwight — Wed, 24 Sep 2008 02:49:04 GMT

You need to: establish the angular view of your specific camera at 28mm, probably a lens chart at manufacturer site. then: make a diagram as to how this diagonal measure might relate to Archicad's horizontal view angle. [as long as the view taken is horizontal]

It is a challenge.
Remember also to allow for the image being cropped.

Re: view cone angle vs. camera lens focal length

Anonymous — Wed, 24 Sep 2008 05:15:27 GMT

thanks.

Re: view cone angle vs. camera lens focal length

Dwight — Wed, 24 Sep 2008 05:27:26 GMT

I just thought of a tip for getting the convergence right if you know the corners of your lot:
visualize a lot sized slab and match that to the corners in the picture background.

Re: view cone angle vs. camera lens focal length

Karl Ottenstein — Wed, 24 Sep 2008 14:49:17 GMT

Also note - discussed at length in older threads - that 28mm is a wide angle setting, and is pretty much guaranteed to have barrel distortion in the photo. (Were there vertical lines of built elements in the photo, they would appear to be bent.) Without removing the distortion with a Photoshop filter/manipulation - or if the camera had a special lens, you will not be able to align a computer model with an existing structure with any great success. Getting the scale right even after correcting for distortion is an art.

Cheers,
Karl

Re: view cone angle vs. camera lens focal length

Dwight — Wed, 24 Sep 2008 17:49:50 GMT

That's part of the problem: if the file knows its focal length, it is a digital file and probably from a sensor smaller than full frame 35mm. If so, the expressed view angle of 28mm will be considerably narrower that we might expect.

Not that any image from a crapcam won't have distortion - also from the fact it was probably hand held and not made from a tripod.

Re: view cone angle vs. camera lens focal length

Anonymous — Wed, 24 Sep 2008 18:55:47 GMT

Would it not be better to work the other way, i.e. don't calculate it, but tweak until it's right?

The method I generally use -
- Model the extents of the site using a temporary slab.
- Adjust the cameras position on plan to the correct position using either an map/site survey/aerial photo to scale placed on the floorplan.
- Then adjust the relative height of the camera compared to the 0.00m level on the building, remembering to add on 1.7 or 1.5m for the tripod level.
- Set the view cone to around 75 degrees (for a compact camera with 6-17mm focal length try starting at 55 deg).
- Manually reduce/increase this figure until the same amount of site is visible, while also recentreing the view.
- I like to use the 'classic 3d navigation' to subtly tweak my position as I find it gives much finer control than the standard 'explore' mode.
- Remember to keep clicking the 'modify the selected' button to update the camera after each manual tweak.

What would be really useful is for ArchiCAD to read the metadata from an image used as a background, extract the date and time and set the up the sun angle and position correctly all automatically.

Re: view cone angle vs. camera lens focal length

Dwight — Wed, 24 Sep 2008 19:14:08 GMT

Your method is inevitable for success - reliance on formulae and automatic tools being useless - but placing a model in a photo, as authoritative as it appears, can only ever be an approximation.

When Graphisoft made the Align View tool it was a kneejerk to a competitor's tool. And it works, but only in a perfect image, and that never happens with photography.

Re: view cone angle vs. camera lens focal length

Fran_ois Chatelain — Thu, 25 Sep 2008 18:05:20 GMT

Hi guys,
it seems to me that you forget to mention the ratio of the 3D/rendering window which should match the photo you want to align too. If the ratio doesn't match the photo's, whatever camera angle you set will be off.
If the photo is digital, the EXIF will give you a fair bit of information, which can be further investigated by checking the camera specs (crop factor...) on the manufacturer website or dpreview dot com.
Now, as a quick help, here's a few horizontal angle values for some common focal lengths (3:2 ratio):
24mm = 76.3deg
28mm = 67.9deg
35mm = 56.6deg
50mm = 41.3deg
85mm = 25deg
But then again, one should allow for some margin depending on lens manufacturer 😉

Cheers
Francois

Re: view cone angle vs. camera lens focal length

Dwight — Thu, 25 Sep 2008 18:30:10 GMT

But lens angle isn't only the focal length or the proportion - it also relates to the size of the sensor.

Re: view cone angle vs. camera lens focal length

Fran_ois Chatelain — Thu, 25 Sep 2008 18:55:43 GMT

Hi Dwight,
which is why it's important to check the EXIF data of any digital photo, which conveys all (most of) the necessary information to help you make the adequate conversion, that is multiply the focal length by the crop factor.
Once you know the camera make and model, as well as the focal length used for the photo, it's quite easy to do the maths to get the real focal length.
It can prove a tad more fiddly if the photo was shot with a P&S camera 😉
But hey, do as I say but not as I do since I always eyeball my alignments in EIAS. That is until I go blind...
Cheers
Francois

Re: view cone angle vs. camera lens focal length

Dwight — Thu, 25 Sep 2008 19:40:32 GMT

Since you are vastly my better at this imaging stuff, I'm glad that we agree that the eyeball is better than blind focallengthanglecropproportion calculations for establishing camera aspects.

A few minutes spent gently walking a model into place is the best way i know of for matching backgrounds.