LGRB Conversion using Mallicam DS16m Camera

[dpaul]

I'm new to this group and have invested in Pixinsight software. I have a monochrome camera from Mallicam (DS16c) but having problems with stacking individual RGB and L images.
I'm taking individual exposures with a luminance filter then a R, G and then B. These are the correct imaging filters from Baader (not visual ones).
I've learnt how to successfully star align the four separate images and also to dynamically crop them. I've also done a histogram transformation to make them non linear (at least I think I have).

The problem occurs when I use colour space then LRGB combination. I open the four image files but it says 'no compatible images'. I've tried xisf, tif, fits formats but still no success.
If I convert each file to gray scale then it works but then I've obviously not getting the right result (although a washed our colour stacked image surprisingly occurs).

I must be doing something fundamentally wrong. I'm using the Mallincam SKY software to capture the images and tried both the RAW and RGB setting when I save the images (RAW is what I usually do). I love the other thing so far that Pixinsight can do after reading lots of helpful advice on this forum and elsewhere - I just need to get past this fundamental problem.

Thanks

[pfile]

the easiest way to do this is to create 2 masters: an RGB image and an L image. process both to become nonlinear and have similar brightness to one another.

then put only the L master image into the LRGBCombination interface and apply that to the RGB image.

that should work. you can tweak the saturation transfer function and lightness transfer functions to taste, but be aware that smaller numbers = larger effect. so setting the saturation transfer to 0.4, for instance, will result in a more saturated image.

rob

[dpaul]

Hi Rob,

Thanks for the reply - I'm aware that's a possibility but I think there is a fundamental problem of the file format.
The original images I'm capturing are RAW, saved as tif files. Then I've done processing of each one (star alignment, dynamic cropping).
The LRGB combination process won't recognise any files, it says 'no compatible files'. When I change them to grayscale it sees them?

Regards

David

[pfile]

well i'm a bit confused, because LRGBCombination runs on views open on your desktop, not on disk files. so the disk format should not matter as PI is representing any open image in its own internal format.

the LRGBCombination process runs on 4 greyscale images. or you can apply it (with the triangle) to an RGB image with only an L image selected in the user interface. that second usage is what i described in my first reply.

rob

[dpaul]

Hi Rob,

Exactly!

I have all 4 images open but they are not recognised.

You mentioned 'greyscale' - I'm starting with a monochrome camera, taking images that are filtered only red, blue or green and the Luminance filter. So these look monochromatic when first captured. What exactly is a greyscale image, surely this is different?
When I convert the above 4 filtered monochrome images (RGB and L) to greyscale they are recognised by the program, if I don't then they are not.

Regards

David

[pfile]

well, by greyscale i mean "single channel" - monochrome image. so it's the same thing.

i think LRGBCombination is working as intended; it wants mono images as input. but as i was saying, an alternate use is to have your RGB image and L images sitting on the desktop, both processed as nonlinear images. you leave the RGB fields unpopulated, but then select your stretched L master as the L image in the LRGBCombination tool. then, you apply the LRGBCombination process to the RGB image. this can be done either by activating the RGB image (meaning it has a blue window border) and then clicking the square on the LRGBCombination user interface, or by dragging the triangle of the LRGBCombination user interface onto the RGB image.

are you sure that the RGB image and the L image have exactly the same XY pixel dimensions?

rob

[pfile]

also it occurs to me that the TIFF format may not support 1-plane images... i can't remember. it's certainly possible your tif images have 3 planes, each with the same exact data in them, and so they are not actually mono images as far as PI is concerned, even though to the eye they appear grey.

when you have any given window selected, at the bottom of the screen you'll see something like the attached image. if it says "n:3" then your image actually has 3 planes. that could be the problem.

rob

[dpaul]

Hi Rob,

I think you've found the problem -
For some reason I'm getting 3 channel/3 plane results in trigger mode on the mono camera. When I switch to 'snap' mode it goes to 1 channel.

I'll now look into the settings I can.

so bottom line is you found the fundamental issue

thanks!

[pfile]

ah, ok. well as a workaround, if all 3 channels contain identical data, you can use the BatchChannelExtraction script to extract just one of the channels and then you'll have new files which are truly mono.

but its probably better to get it sorted in the capture program.

rob

[dpaul]

Hi Rob,

I did the channel extraction on one of the files and it converts to 3 equal channels and I can delete two of them. The result though is a greyscale images, are these different to a truly 1 channel image?
When I then use LRGB combination, it only lets me use the circle (apply global) and not the square (apply) because with the 'apply' it says 'cannot do on greyscale images.
Also, with channel extraction, presumably I'm only using 1/3 of the data where the whole point was to have the all the pixels for each colour - I'm sure you agree.

At the data capture point I agree is the best way (if I can sort out how).

Thanks

David

[bulrichl]

David,

according to Mallincam, the Panasonic CMOS sensor of your camera has a resolution of 4656 x 3518 pixels. Compare this value with the resolution of the result of the channel extraction. PI shows it in the lower Tool Bar when 'VIEW>Tool Bars>Information' is checked.

Bernd

[pfile]

Hi Rob,

I did the channel extraction on one of the files and it converts to 3 equal channels and I can delete two of them. The result though is a greyscale images, are these different to a truly 1 channel image?
When I then use LRGB combination, it only lets me use the circle (apply global) and not the square (apply) because with the 'apply' it says 'cannot do on greyscale images.
Also, with channel extraction, presumably I'm only using 1/3 of the data where the whole point was to have the all the pixels for each colour - I'm sure you agree.

At the data capture point I agree is the best way (if I can sort out how).

Thanks

David

the target of an LRGBCombination process must be a 3-plane image. that's why it does not work.

again, what you should be doing is:

1) get the per-filter channel masters as 1-plane files, one way or another
2) combine the RGB masters into an RGB image (using ChannelCombination or Pixelmath)
3) process the RGB master (a 3-plane file) - needs to be stretched at some point in the processing.
4) process the L master (a 1-plane file) - needs to be stretched "similarly" to the RGB file during processing
5) enter the L master only in the  LRGBCombination user interface
6) apply the LRGBCombination process to the RGB master.

the Panasonic sensor in your mallincam is mono, right? in which case there is only ever one set of pixel values. the capture program may be simply copying those values into each channel of a 3-channel tiff, but that does mean there's any more data to be had. so you're not losing anything by throwing away 2 channels. on the other hand, i'm not sure i've ever come across a capture program that does what yours is doing, so it bears some investigation.

rob

[dpaul]

hi Bernd,

When I compare each channel after channel subtraction they are equal

[dpaul]

Hi Rob,

I did LRGB combination using 4 separate one-channel files and got a colour image (see attached) - not brilliant because the RGB exposure lengths are probably too small. Does it need a target, seems to work without? I tried RGB combined first with channel combination then used it as a target in LRGB combination and added 1-channel luminance but ended up with no colour at all, even though the target had colour.

Regards

David

[pfile]

were the images stretched beforehand, or were they still linear? the veil is really weak...

i don't use LRGBCombination in "4-image" so i'm not sure if it needs a target; perhaps not (similar to ChannelCombination).

rob