OK, but I'm still not clear what you are hoping to check. Do you want to check
values in each original image against the output of hdrgen? The easiest way to
do that would be to feed hdrgen a single image along with the response curve
(using the -r option). That's not really a check on hdrgen, though, since you
would only be comparing it to itself for consistency.
You can use the output of ra_xyze piped into pvalue as we discussed earlier,
then pull up individual pixel values in the image in Photoshop or your favorite
tool. The luminance should approximately equal this:
luminance = sample_to_nits * (lin(R)*0.213 + lin(G)*0.715 +
lin(B)*0.072)
lin(p) = ((p+.5)/256)^2.2
R,G,B = pixel components in 0-255 range
The linearization function above doesn't do a perfect job, since each camera
has its own response function, and this is one of the things hdrgen figures out
for you. In particular, I would expect larger errors in the shadow regions of
each exposure.
Does this help?
-Greg
> From: "Morsink, K." <[email protected]>
> Date: July 27, 2017 9:04:45 AM PDT
>
> Hi Greg,
>
> I'm using hdrgen to create the hdr images. I'm asking to understand how the
> EXPOSURE value of the hdr image can be linked back to the different exposure
> times of the original images. This to better understand the origin of this
> value and how it is used to compute the luminance values. I'm not using any
> other software. But from your answer I understand that this is not so
> straightforward to check?
>
> Thank you for your time and clarification.
>
> Kind regards,
> Kars
>
> -------- Original Message --------
> Subject: Re: [HDRI] [Radiance-general] How does Pvalue determine CCT in HDR
> images and how is the EV of a HDR image determined?
> From: Greg Ward
> To: High Dynamic Range Imaging
> CC:
>
> Hi Kars,
>
> It's difficult to answer your question if you don't explain why you are
> asking. If you use the given formulae to compute your EXPOSURE value in some
> kind of manual conversion of each image to HDR format, the values will
> correspond, even if the limits are still standard dynamic range. This is
> sort of what happens if you give Photosphere a single image and tell it to
> build an HDR result. If you give Photosphere multiple images, it knows how
> to apply the formula for you. The same is true of hdrgen.
>
> If you are using different software or writing your own, then you need to
> better explain your expected inputs, or provide a clear example with where
> you are stuck.
>
> Best,
> -Greg
>
> P.S. I am moving the remainder of this thread to the HDRI mailing list.
>
> > From: "Morsink, K." <[email protected]>
> > Date: July 27, 2017 8:22:59 AM PDT
> >
> > Hi Greg,
> >
> > Thank you for your quick response!
> >
> > The formula you provided works for me with a single LDR image and thus a
> > single exposure time, but I'm stuck in how I should insert the multiple
> > exposure times (of the 7 LDR images to form the HDR image) in exposure_secs
> > and get the correct EXPOSURE, could you maybe explain this further?
> >
> > Making use of the equal-energy illuminant E, is it assumed that the white
> > point remains constant (1/3, 1/3)? I'm asking this since I'm taking outdoor
> > photographs, where the CCT (and thereby the white point) changes
> > constantly, assuming a constant white point would therefore affect the
> > accuracy of the conversion to luminance values.
> >
> > Kind regards,
> > Kars
> >
> > -----Original Message-----
> > From: Greg Ward [mailto:[email protected]]
> > Sent: woensdag 26 juli 2017 19:15
> >
> > Hi Kars,
> >
> > I am cross-posting my response to the HDRI mailing list per Chris'
> > suggestion.
> >
> > Specifying the "-o" option of pvalue takes into account the exposure
> > setting, which is determined from the ExIF data if you created the image
> > using Photosphere or hdrgen. The formula below has an empirically derived
> > constant that may not exactly fit your camera, so it is best to add your
> > own calibrating scale factor:
> >
> > sample_to_nits = 87 * (f-stop)^2 / (ISO * exposure_secs)
> >
> > The sample_to_nits is converted to a Radiance picture exposure using:
> >
> > EXPOSURE = 179 / sample_to_nits
> >
> > where 179 lumens/watt is the agreed-upon efficacy of the equal-energy
> > illuminant E over the visible spectrum.
> >
> > Unfortunately, pvalue is not very smart about reporting brightness using
> > the "-b" option. It uses a formula based on the standard Radiance color
> > space, which differs from the CCIR-709 color space produced by Photosphere
> > in both the green primary and the white point. It only makes a small
> > difference, but if you are worrying about such things, you had best use the
> > following to report luminance from your image:
> >
> > ra_xyze image.hdr | pvalue -o -b [other options]
> >
> > This also takes care of the 179 factor, reporting results in
> > candelas/meter^2.
> >
> > Cheers,
> > -Greg
> >
> >> From: "Morsink, K." <[email protected]>
> >> Date: July 26, 2017 7:34:45 AM PDT
> >>
> >> Hello everybody,
> >>
> >> I'm Kars and I'm new to this Radiance forum.
> >>
> >> In the manual of Pvalue
> >> (https://www.radiance-online.org/learning/documentation/manual-pages/pdfs/pvalue.pdf)
> >> it is stated that inputting a file in XYZE format will give you the
> >> luminance values of the image (corresponding to the Y channel). I've
> >> inserted some .hdr images in Pvalue with the following Primaries (taken
> >> from the EXIF data): PRIMARIES= 0.6400 0.3300 0.3000 0.6000 0.1500 0.0600
> >> 0.3127 0.3290, where the first two numbers correspond to the R primary
> >> (x,y), the third and fourth to the G primary (x,y), the fifth and sixth to
> >> the B primary (x,y), and the seventh and eighth to the white point (x,y).
> >> The EXIF data also shows an exposure value of the .hdr image.
> >>
> >> I've got two questions regarding this conversion.
> >>
> >> To my knowledge the white point coordinates can be used to calculate the
> >> CCT, as described by Inanici in Evaluation of High Dynamic Range
> >> Photography as a Luminance Data Acquisition System. The white point
> >> coordinates, as found in the EXIF data, correspond to the CIE standard
> >> illuminant D65. Does this mean Pvalue assumes a constant CCT for all the
> >> .hdr images when converting to luminance? Or is Pvalue making other
> >> assumptions / calculations?
> >>
> >> My second question refers to the exposure value of the .hdr image, as
> >> shown in the EXIF data of the image. How is this value determined? Since a
> >> HDR image consists of multiple images (in my case 7) with different
> >> exposure values. What is this exposure value of the .hdr image based on
> >> and is this used by Pvalue somehow?
> >>
> >> I hope I've made myself clear.
> >>
> >> Kind regards,
> >>
> >> Kars Morsink
>
> _______________________________________________
> HDRI mailing list
> [email protected]
> https://www.radiance-online.org/mailman/listinfo/hdri
> _______________________________________________
> HDRI mailing list
> [email protected]
> https://www.radiance-online.org/mailman/listinfo/hdri
_______________________________________________
HDRI mailing list
[email protected]
https://www.radiance-online.org/mailman/listinfo/hdri
