Re: Photosphere "Cannot solve for response function"
Hi
I am (finally) back from summer holidays and cannot help but add my twopence worth - as the supervisor Peony
refers to.
Peony seems to have narrowed down the problem to a jpeg format issue that makes the image unreadable,
nothing to do with the phone itself.
For example, she is able - using the same HDR software on her tablet to generate a camera response curve.
The software for the phone that she is using to generate the HDR images allows the turning of off auto white
balance and allows the setting up of 3, 5 or 7 images at separate exposure values.
She can clearly create the camera response curve for the $2000 Nikon camera with its $2000 fish eye lens
With her $300 phone and $50 fish eye lens, and the same software as the tablet she cannot create a curve.
The software saves the original jpg images as well as the combined HDR image on each machine.
At first we thought it was inadequate EXIF data in the original jpeg images. However, it is not the EXIF data
because the Nikon camera images with the phone image EXIF data replacing the Nikon's EXIF data still
produce a camera response curve.
She has eliminated the jpg image compression from the question because she has opened each jpg original in
photoshop and re-saved it in a standard jpg format - the Nikon images still work, the phone not.
Personally, I'd like this to work because first year and second year building science student measurement
exercises could be so much more interesting if we were measuring a whole space, not a simple grid... Also:
We can calibrate the sound level meters on the phones. They can use a simple app on the phones to photograph
the horizon line with overlaid sun path diagrams in Augmented Reality mode. The phone as general survey
instrument, not just site camera is becoming a reality. Calibration / trust of each phone, and how to test
it is the issue
At another level, I feel responsible for understanding what the issue might be to try to explain what might
be going wrong so a) Peony can understand, and document this process properly in her thesis Appendix and b)
some recommendations can be formulated as to what camera-phone characteristics to look for.
Thus I return to the question: what might we be unable to control with the phone, but is controlled on the
tablet, that would cause this kind of Photosphere error message? And how might we further test what we are doing?
Thanks in anticipation of any help.
M
------------------------------------
Victoria University of Wellington School of Architecture
Michael Donn
michael.donn@...
PO Box 600
139 Vivian St
Wellington
New Zealand
tel: +64 4 463 6221
fax: +64 4 463 6204
mobile: +64 21 611 280
Skype ID:the_donn
------------------------------------
Today's Topics:
1. Re: Photosphere "Cannot solve for response function"
(Gregory J. Ward)
----------------------------------------------------------------------
Message: 1
Date: Mon, 24 Dec 2012 08:56:34 -0800
From: "Gregory J. Ward" <gregoryjward@...>
To: High Dynamic Range Imaging <hdri@...>
Subject: Re: [HDRI] Photosphere "Cannot solve for response function"
Message-ID: <2B93AA80-729C-406E-88FD-91ADEF27401B@...>
Content-Type: text/plain; charset=us-ascii
Hi Peony,
I guess no one ever responded to this... I can really only repeat my response from August, which is that
getting calibrated output from phone cameras is not possible at this stage, and may never be. Their
application is too different, and repeatability is not one of the design requirements for these modules.
They alter sensitivity, color response, and probably tone response without making any record of the
original data. I really think it is a waste of your time.
Best,
-Greg
> From: "Peony Au" <peony_au@...>
> Date: November 27, 2012 11:13:11 PM PST
>
> Hi everyone,
>
> After much development on my thesis, I have a few queries I am hoping some of you will be able to help me out on.
>
> First of, my supervisor pointed out an application on the iPhone called iPhotoLux which maps the
luminance measurement of an HDR image taken by the iPhone camera. So I am hoping what I am trying to produce
in my thesis does work.
>
> Secondly, I have been using a Nikon DSLR camera to create HDR images in the lighting laboratory using a
scale model, so that my skills in creating a accurate HDR image. I have also used both luminance and
illuminance meters to ensure the photos are accurate.
>
> Therefore, after experimenting in the lighting laboratory, I got my Smartphone camera back out and using
a HDR Camera application available in the Android market. I put both cameras side by side and took images of
the same scene on a tripod. However, I am still receiving the same error message as before, then I found out
that the HDR Camera application does not write the EXIF data required to create a camera response curve.
Therefore, just to try the images produced by the Smartphone out, I coped the EXIF data from the DSLR camera
and inserted into the images produced by the Smartphone camera ensuring that all the exposure value
information are the same. I then try and fuse these Smartphone camera produced images together, but the
same error message reappears.
>
> Therefore, I was just wondering if anyone knows why the camera response curve cannot be created even with
the comparison done between the DSLR camera and the Smartphone camera.
>
> Thank you for all your help.
>
> Kind regards,
> Peony
------------------------------
_______________________________________________
HDRI mailing list
HDRI@...
http://www.radiance-online.org/mailman/listinfo/hdri
End of HDRI Digest, Vol 54, Issue 2
***********************************
RSS Feed