When ever i upload .dng files into SIRIL 1.2 it wont allow me to convert them to fits files.
It always gives me this in the console “Siril cannot open this file in CFA mode: no RAW data available.”
does anyone know how to solve this or will siril 1.2 just not deal with .dng’s
This is because your dng file does not contain RAW data. This is not a RAW file. Without it you cannot process your image with calibration.
it seems to work in siril1.06, so is this just a 1.2 thing?
Nope. In 1.0.6 maybe you had a color image, debayerd. But you can’t have raw data because they don’t exist.
Have an example DNG to provide here for analysis? Also where did this DNG come from?
I have exactly the same issue as you.
I’m using the Galaxy s22 ultra for taking pictures and I’ve been using the same settings all the time but the new pictures won’t open, which I can with older pictures that I’ve processed earlier in 1.0.6.
This picture of the pleiades I had no issues opening in 1.2
20221226_231633.dng (23.2 MB)
But this one of M5 doesn’t work
20230422_012427.dng (22.3 MB)
Tried to download 1.0.6 again and the new files work. So I guess I’ll just have to stay with that version for now
Are these Samsung “Expert RAW” DNGs?
Reading your files with exiftools I have for 20221226_231633.dng
CFA Pattern : [Green,Red][Blue,Green]
Compression : Uncompressed
And for the other (20230422_012427.dng):
Compression : JPEG
So this is perfectly normal Siril cannot open RAW data because they are not RAW anymore as I suggested in my previous message.
Compression alone is not sufficient to determine the DNG type, as lossless JPEG applies perfectly to Bayer CFAs as well (and LibRaw has no problems decoding those). What matters is that these match, e.g. for 20221226_231633.dng
:
PhotometricInterpretation : Color Filter Array
SamplesPerPixel : 1
(so, indeed a “reguler” Bayer CFA raw image), vs 20230422_012427.dng
:
PhotometricInterpretation : Linear Raw
SamplesPerPixel : 3
which is a high bit-depth (BitsPerSample : 12 12 12
) RGB, i.e. debayered one (what Samsung calls “Expert RAW” that has been processed/enhanced by their computational photography engine by combining e.g. a burst of raw images). Btw, LibRaw has no problems decoding these either.
Maybe but Siril needs RAW data, not debayered and not compressed which is not the case here.
Sure, that explains it.
Incidentally, why the limitation of no compression? Are you not using dcraw/LibRaw or RawSpeed for decoding input files? If so, they should take care of it…
“no RAW data available”
P.S. (Losslessly) Compressed raw data is still raw, so the error message could be improved. It seems you really have two cases here: “PhotometricInterpretation is not CFA (print tag value). Only uncompressed CFA is supported.”, or “data is compressed (print tag value). Only uncompressed CFA is supported.”
No but doing calibration on lossy compressed data has no meaning. And the same for debayered data.
doing calibration on lossy compressed data has no meaning
I did mention lossless a few times specifically. It is used in CFA DNGs frequently.
From the DNG spec:
If PhotometricInterpretation = 6 (YCbCr) and BitsPerSample = 8/8/8, or if PhotometricInterpretation = 1 (BlackIsZero) and BitsPerSample = 8, then the JPEG variant must be baseline DCT JPEG.
Otherwise, the JPEG variant must be lossless Huffman JPEG.
So, having PhotometricInterpretation = 32803 (CFA) w/ lossless JPEG (as described in ITU T.81) compression is perfectly valid, and not at all uncommon.
If you’re not using dcraw/LibRaw or RawSpeed, imagecodecs e.g. ships 2 implementations of it (jpegsof3 and liblj92).
OK, so here the blocking point is the already debayerd data.