Zero or insignificant signal detected (empty image?)

[dhubbell]

Yeah, all the raw data is in 16 bit FITS format.  Not sure where I went wrong...I'll keep a better eye out during my steps to make sure everything is 32 bit float.  The camera was an Apogee A8050 mono.  Everyone knows not to process in 8 Bit, I had no idea XISF was even capable of 8 bit. 

[jcinpv]

I have the latest PI and I just got this error during processing of my Flats.

My camera is a new modified/cooled Canon 6D. I took the Flats by taking images off an LCD monitor that was displaying a 50% Grey image (full screen, 8" EdgeHD placed against the screen, ISO 100, 2s exposure). All images show the histogram in the middle.

I need an explanation and a clue.

[dhubbell]

Changing my XSIF file format to 32 bit fixed my problem.  Check to make sure your XSIF files are 32 bit.

[jcinpv]

Changing my XSIF file format to 32 bit fixed my problem.  Check to make sure your XSIF files are 32 bit.

Thanks for the suggestion. How is that done?
I'm totally new to PI. Trying to stick with defaults until I can at least get through an easy workflow.

John

[dhubbell]

I'm no expert on PI either, but, when you're ready to SAVE AS select file format XISF, then you'll see XISF OPTIONS, from there, I've used the "32-bit IEEE 754 floating point" option.  When this happened, I opened my original file and did a SAVE AS and saw my failed files where in "8-BIT unsigned integer".  You really can't do much with 8 bit images.  I had no idea how my image was saved as 8-bit, but, it was.  Once I re-created the files as 32 bit XISF then it worked, no problem.

I hope this helps you find your answer.

Doug

[pfile]

the defaults are in the "format explorer" tab on the left - double click XISF and you can make sure its set for 32b.

rob

[jcinpv]

the defaults are in the "format explorer" tab on the left - double click XISF and you can make sure its set for 32b.

rob

I've checked Processes and Scripts, can't find "Format Explorer".

[jcinpv]

I'm no expert on PI either, but, when you're ready to SAVE AS select file format XISF, then you'll see XISF OPTIONS, from there, I've used the "32-bit IEEE 754 floating point" option.  When this happened, I opened my original file and did a SAVE AS and saw my failed files where in "8-BIT unsigned integer".  You really can't do much with 8 bit images.  I had no idea how my image was saved as 8-bit, but, it was.  Once I re-created the files as 32 bit XISF then it worked, no problem.

I hope this helps you find your answer.

Doug

When I did "Save As" on one of them, the 32-bit was already checked.

[pfile]

the defaults are in the "format explorer" tab on the left - double click XISF and you can make sure its set for 32b.

rob

I've checked Processes and Scripts, can't find "Format Explorer".

View > Explorer Windows > Format Explorer

or the "Format Explorer" on the left-hand side tab.

rob

[dhubbell]

I'm no expert on PI either, but, when you're ready to SAVE AS select file format XISF, then you'll see XISF OPTIONS, from there, I've used the "32-bit IEEE 754 floating point" option.  When this happened, I opened my original file and did a SAVE AS and saw my failed files where in "8-BIT unsigned integer".  You really can't do much with 8 bit images.  I had no idea how my image was saved as 8-bit, but, it was.  Once I re-created the files as 32 bit XISF then it worked, no problem.

I hope this helps you find your answer.

Doug

When I did "Save As" on one of them, the 32-bit was already checked.

Are ALL of your calibration and light frames 32 bit?  I have a flat frame that was 8-bit and gave me this error.  If all of your frames are 32 bit, then that's not the same problem I had.  I had 1 8-bit frame and it caused this error.

[jcinpv]

the defaults are in the "format explorer" tab on the left - double click XISF and you can make sure its set for 32b.

rob

I've checked Processes and Scripts, can't find "Format Explorer".

Found it. Looked at it. Can't find anything there that says 32b.


View > Explorer Windows > Format Explorer

or the "Format Explorer" on the left-hand side tab.

rob

[jcinpv]

I'm no expert on PI either, but, when you're ready to SAVE AS select file format XISF, then you'll see XISF OPTIONS, from there, I've used the "32-bit IEEE 754 floating point" option.  When this happened, I opened my original file and did a SAVE AS and saw my failed files where in "8-BIT unsigned integer".  You really can't do much with 8 bit images.  I had no idea how my image was saved as 8-bit, but, it was.  Once I re-created the files as 32 bit XISF then it worked, no problem.

I hope this helps you find your answer.

Doug

When I did "Save As" on one of them, the 32-bit was already checked.

Are ALL of your calibration and light frames 32 bit?  I have a flat frame that was 8-bit and gave me this error.  If all of your frames are 32 bit, then that's not the same problem I had.  I had 1 8-bit frame and it caused this error.

I went through all the xisf files in the Flats. They all show 32-bit selected when I attempt to save them. Would it show anything else if the file isn't 32-bit? Or does it simply default to 32-bit when saving?

[dhubbell]

I went through all the xisf files in the Flats. They all show 32-bit selected when I attempt to save them. Would it show anything else if the file isn't 32-bit? Or does it simply default to 32-bit when saving?

If I remember correctly, I noticed a file size difference.  So try saving one both ways, the 8bit file should be smaller than the 32bit file.  I'm sorry I can't be more helpful.  It sounds like you have 32 bit files.

The only other thing I can think of is trying another data set.  If you want to download another data set and try it...I have another data set here you can try:
https://youtu.be/prZCnqw5RPQ

[pfile]

sorry about that - i thought the default bit depth was set there. i think the original topic of this thread was that there was a bug causing PI to save in 8-bit when it should have been using a deeper bit depth - when you run ImageCalibration on an i16 subexposure with f32 masters, the output *should* automatically be set to the f32 bit depth because the depth of an expression where the right hand side has i16 and f32 has to be f32.

before getting on this tangent i was going to ask what the flat was calibrated with. it's possible to destroy a calibrated image by say, subtracting an 1800s dark from a 5s image without scaling the dark.

rob