Wrong DGP Result

Hi @Clotilde_Pierson,

Generating an HDR image for glare analysis, I get imperceptible glare results even if the glare is intolerable for my eyes. Following your tutorial and comparing the measured Ev with the HDR’s Ev, the difference is 67%.

Everything has been checked to be correct and calibrated like the camera response curve, and luminance calibration; given that, I am wondering what could cause this issue.

Would be great to help me in this regard.


typically there are two reasons for this:
a) incorrect header
b) pixel overflow
maybe you can share the hdr-image and the header (by typing getinfo image.hdr → screenshot)

However, we are currently conducting a study on the performance on different HDR imaging devices/algorithms. We are not yet finished with this study, but it seems using JPEG images and having an extreme high peak in the image (e.g. visible sundisk through a shading) can cause large deviations for the sun disk (>50%) from the real value, when using a self-calibrated method (e.g. hdrgen). Typically this is then an overestimation and very often cannot be seen by the Ev check (because the rest of the image is underestimating the luminances, so the integral gets close). For less high contrasts in the image this problematic seems not or lesser to be present.
Using RAW images seems to improve the situation but absolute calibrated cameras are definitely more robust for such extreme situations.


Thanks @Jan_Wienold for your response. Here is the HDR image and its header:

Here the difference between Ev is about 70%.

Thank you

I guess sth “happened” before you applied the pcomb -h -s command, that is not visible in the header any more. Can you share also the hdrS.hdr image so I can check the header? In general, it would be better NOT to use the -h option with pcomb , so you keep the header information and troubleshooting is easier.

Also, would be important to know what commands you used from image-merging to the final image (especially if they are not visible in the header). My guess is that an exposure entry was removed or treated in the wrong way before you applied the pcomb command.

You apply a factor of 1.22, so you did a spot measurement as well? If yes, is that reference area correct in terms of luminance or you also have a large deviation?

And, another question: was there any “peaky” glare source in the scene ? The image looks like it was overcast (i.e. no sun or reflection of it visible, max luminance in that image is 941 cd/m2). But you said the glare is intolerable in your point of view.


Thanks @Jan_Wienold,
I figured out that some of the images are not being considered in HDR image generation process. I will solve this issue and double-check the results.

No I did not measured it for now

Here it was an intermediate sky; was not intolerable but still, some glare was in my FOV.

Yes, I had the sun in FOV but it was in another scene.

I will double-check the process and will update you with the result.

Thanks again.

Hi @Jan_Wienold,

I did the process again with all the Ldr images the camera captured and now the discrepancy is less than 10% for Ev which seems fine.