HDRComposition question

[CraigNZ]

Using a DSLR I have recorded sets of 5,10,20,40,80,160 and 320 seconds, each with 20 images and all taken at ISO 100.  The images have been calibrated, registered and integrated, so I now have a single image for each exposure time.

If I now use HDRComposition to combine them it fails after the second image with the message:

** This HDR composition cannot integrate further images - aborting process.

It works okay if I use any two images (e.g., 5 and 40 seconds).  It also works if I use any three images as long as they are less than 160 seconds in length (e.g., 5, 40 and 80).  But I cannot do any 4 images (e.g., 5, 10, 20, 40).

Does this mean there is some sort of math limit to the function?

[TobiasLindemann]

Craig,
may I ask why do you want to combine so many steps? Are all steps really necessary. Are your stars saturated on the 10 sec. exposure at ISO 100? I would try to use only the 320, 40 and maybe 5 sec. steps, if the 10 sec. are overexposed. In my opinion the other steps give no better quality, but I don't know the object and your images. In HDRComp. the lower exposures are only used for the regions, where the longer exposures are saturated. So the short exp. give no better S/N  in the lower regions.
Tobias

[Alejandro Tombolini]

Hi Craig, try lowering the "Binarizing threshold" value.

Saludos, Alejandro.

[CraigNZ]

I tried 320, 40, and 5 and it failed.  I lowered the threshold down to .13 and I get a linear fit error and .14 it gets the cannot integrate error message.

It does work with 320 and 20 (or any other one as long as there are only two being combined).  But with the 320 and 20 I get a red image.  Do I need to color correct each of the subs before I use HDR?  No white balance has been performed on the images. The red channel is dominant in all of the subs and maybe the red is saturating somehow the HDR algorithms?

[Alejandro Tombolini]

if you didn't do, it would be better to apply DBE on each sub before HDR. Also try unchecking the "automatic exposure evaluation" and ordering the images as describe in the tooltips.

[Juan Conejero]

This HDR composition cannot integrate further images

This message means that, with the parameters you are using, the HDR composition process cannot integrate more images. In other words, it has reached a point where the next image in the set cannot replace any pixel with unsaturated data.

As Alejandro has said, decreasing the binarizing threshold parameter may help. However, nothing really useful can be said without looking at the images.

[CraigNZ]

So if I understand correctly, if I select images of 320, 40 and 5 for example, it will start with the longest exposure time (320 seconds) and identify all saturated pixels.  It will then move to the next exposure (40 seconds in this example) and try to resolve those pixels with unsaturated ones.  Then if there are no saturated pixels in the 40 second image it will stop because nothing is added by the 5 second image.  Is this right?

[CraigNZ]

How does HDRComposition know if a pixel is saturated?

[CraigNZ]

Never mind that last post, I found the articles detailing the development of HDRComposition and the use of LinearFit.

I think where part of my problem comes in is the images were taken on different nights with different sky conditions.  Ideally they should all be taken on the same night under the same sky conditions.  When I use blink and scan through the images I can clearly see changes in the brightness due to high level cloud passing through at different times and on different nights.  This may be confusing the LinearFit algorithms.

[vicent_peris]

Hi Craig,

I think it should be the binarizing threshold. Please check at which point the pixels are saturated in your DSLR images because the saturation point can be far from 1 and lower than the default 0.8. If you want, I can check for you the number if you upload the images.

I don't understand why you need to use LinearFit, it's a tool that's not required at all to make HDRComposition work. Moreover, you can shoot your exposure time sets in different nights, no problem at all. Anyway, it would be good to check the images because the clouds can be altering the automatic exposure evaluation. At the end, you could uncheck this uption and sort the exposures manually.


V.

[CraigNZ]

After reading through the explanation of how HDRComposition works and after analyzing my images I decided to try 80, 20, and 5, by adding them in that order and then turning off the automatic exposure setting.  This worked and I got a good HDR.

[vicent_peris]

OK, then it was a problem with the clouds. If you don't have any cloud in your data sets, then it should work well even if the images have been shot in different nights.


Best regards,
Vicent.