Why shouldn't Hugin trust exif data 100% ?

[ChameleonScales]

As I understand it, metadata is not 100% reliable for Hugin. But what are the main reasons ?

Is it due to:

- the device not knowing the lens you put on it?

- the device simply not having very precise metrics?

- mechanical imprecision in how the shutter and other components move when capturing the photo?

- slight deformations in the device components at manufacturing time or after prolongated use or shocks?

- something else?

Note that this is more out of curiosity than anything else. I don't have a particular problem with it (at the moment).

[ChameleonScales]

Now obviously, this certainly depends on the device's brand and quality but I assume there is a main reason that applies to your average reflex camera, which is my main question.

[Gunter Königsmann]

My guess (it's only a guess) is that for most cellphone cameras the
manufacturer hasn't cared if the metadata it comes with is any correct.

Kind regards,

    Gunter.

[David W. Jones]

That's my experience, too.

--
David W. Jones
gnome...@gmail.com
wandering the landscape of god
http://dancingtreefrog.com

Sent from my Android device with F/LOSS K-9 Mail.

[Harry van der Wolf]

Well,

Earlier my Samsung S5 mini, then my Motorola G5+ and now the Motorola G6+ all do a very fine job when it comes to Exif data. What's more: They also provide GPS data which my camera doesn't.

And I don't understand why it should be worse than on a phone compared to a camera which only sometimes gets new firmware. 

In all those (stupid) commercials nowadays, it seems that the camera is the most important feature on a phone, and those phones go in much higher amounts than cameras nowadays. So I assume the cameras and their software gets a lot of attention and can be more easily upgraded/fixed than firmware in cameras.

The only issue is that they only store Exif data and very limited makernotes. 

No XMP, no composite, no icc_profiles. At least not those 3 phones I used and mentioned.

My panasonic system camera comes with a lot of data in all those areas, where it sometimes is redundant as XMP combines a lot.

Harry

Op do 16 apr. 2020 om 23:36 schreef David W. Jones <gnome...@gmail.com>:

--
A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ
---
You received this message because you are subscribed to the Google Groups "hugin and other free panoramic software" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hugin-ptx+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hugin-ptx/4860DF62-6E08-4784-AF05-85F5265ECB89%40gmail.com.

[FabKzo]

Hello ;-)

Indeed , generally metrics aren't so precise, and metadatas don't take physical/optical errors into account;

to refine all and limit errors you have to precisely calibrate each of your lenses ,as for motion tracking.

Lenses comes generally with a 2% to 5% focal variation.

[Erik Krause]

Am 15.04.2020 um 12:10 schrieb ChameleonScales:
> As I understand it, metadata is not 100% reliable for Hugin. But what
> are the main reasons ?

Not speaking of phones...

First of all hugin originates from panotools which was made in the
1990ies, when there where almost no serious digital cameras. So
thankfully hugin doesn't need EXIF data at all.

Standard EXIF doesn't contain the lens type (fisheye or not). That would
eventually be in the Maker Notes, but not for all lenses and cameras.

Manual lenses don't provide EXIF data since they don't have electrical
contacts.

Even lenses with electrical contacts don't provide exact data, since in
most lenses there is a simple electro-mechanical encoder which works in
only few discret steps for focal length (in a zoom lens) or focus
distance. Cheaper lenses don't even provide focus distance.

EXIF doesn't contain any information about the lens distortion. This
might vary from lens to lens of the same brand and type anyway, so not
even lens databases provide reliable information.

In short: EXIF data isn't reliable at all. Why trust it if all necessary
data can be calculated from the image overlap?

--
Erik Krause

[ChameleonScales]

Wow, thanks for your answers. I didn't know it could be that much unreliable. Very interesting.