DSS, Is my image B&W ?

[wxsatuser]

On 10/17/2017 at 08:34, SAW said:

Hi,

I've been working on M45 a lot and after DSS stacks all my lights, darks and bias It looks B&W ? It definitely has some colour on my 600D after it takes an image. I am stacking only RAW images. When I open the image in Pixinsight and do the first stretch it goes B&W and not like a green colour I've seen on all the tutorials ?

Is it my 600D or something set wrong in DSS ?

I've not had this problem before, I've always seen some colour in DSS after it's stacked, is it just M45 that's doing it as there isn't much colour ?

Thanks.

This about normal for DSS, and a common gripe from newbies, you have to bring out the colour.

It will be colourful on the camera screen as it's a processed jpg, processing done by the camera.
It's your job in DSS and other software like Photoshop to bring out the colour, sharpness etc,
afterall you are working with RAW files. It's the same for daylight photography if you work with RAW,
you have to do the processing in a RAW converter

There may be not huge amount of colour, looking at your single sub, 180secs with an ED 80 is not a lot.
Bright stars will tend to look white but will have a colourful halo, at 3 minutes you not getting much nebula
so you are missing colour there, longer exposures will help.

[wimvb]

In PixInsight, if your image is RGB, it will say so in the image window titlebar. If the image is gray scale, it will also say so. Gray scale images have the red x on the STF sliders for the green and blue channels. When using bpp script, you check the box that says CFA (right hand panel). With manual calibration, just before image registration, you need to use the batch debayer script.

[Scooot]

7 hours ago, SAW said:

Didn’t have any flats, just stacked lights, bias and darks. 

So the easiest way is to select de-bayer and then re-stack so I get a colour image or 

the other option for a better image is to use the raw option then run de-bayer process ?

Yes using the debayer setting is easiest but i’m not sure using raw is necessarily better it’s just different. If you use raw you have to manually calibrate as wim says. You use the debayer script before stacking and then do the colour calibration routine.The resulting colour and white balance (prior to processing) is certainly different to the image if using the debayer setting prior to stacking.

i used to use the raw setting and manually calibrate my moon images before I knew of the alternative. I was never that happy with the colour balance and my processing skills were not good enough to tweak it to my satisfaction. However now I always use the debayer setting for these and I think the colour is much better. I also use the debayer setting to process some daylight non Astro images occasionally and it’s definitely easier to get the right colour with these. 

I haven’t tried the debayer setting on deep sky images but will do next time I do one to experiment.

[SAW]

1 hour ago, wimvb said:

In PixInsight, if your image is RGB, it will say so in the image window titlebar. If the image is gray scale, it will also say so. Gray scale images have the red x on the STF sliders for the green and blue channels. When using bpp script, you check the box that says CFA (right hand panel). With manual calibration, just before image registration, you need to use the batch debayer script.

I think this might of been the problem, I will try it again this evening after work.

[SAW]

6 hours ago, wxsatuser said:

This about normal for DSS, and a common gripe from newbies, you have to bring out the colour.

It will be colourful on the camera screen as it's a processed jpg, processing done by the camera.
It's your job in DSS and other software like Photoshop to bring out the colour, sharpness etc,
afterall you are working with RAW files. It's the same for daylight photography if you work with RAW,
you have to do the processing in a RAW converter

There may be not huge amount of colour, looking at your single sub, 180secs with an ED 80 is not a lot.
Bright stars will tend to look white but will have a colourful halo, at 3 minutes you not getting much nebula
so you are missing colour there, longer exposures will help.

Yeah M45 is probably not the best image to first be trying in Pixinsight but I've got 2hrs worth of data on it.

[Scooot]

2 hours ago, wimvb said:

In PixInsight, if your image is RGB, it will say so in the image window titlebar. If the image is gray scale, it will also say so. Gray scale images have the red x on the STF sliders for the green and blue channels. When using bpp script, you check the box that says CFA (right hand panel). With manual calibration, just before image registration, you need to use the batch debayer script.

I didn’t know about bpp I’ll have to try that, I’ve always either used imageintegration or FFT registration to stack  

[SAW]

16 hours ago, Rico said:

I find the batch preprocess in pixinisght really temperamental, and as wimvb suggests, lightvortexastronomy has a really nice description of how to manually calibrate, register and stack your images in pixinsight. I can't find a setting in DSS that removes the colour from a raw dslr image. Have you tried converting the .cr2 files to DNG before plugging them into to DSS. I find that can sometimes help. Also, run an automatic background extraction in pixinsightafter the fact. When you save the master file from DSS as a tiff file and you open it in pixinsight, does it say RGB next to the image header?

Good point, I’m not sure I’ll try that later. I know that the image I stacked using PixInsight definitely says grayscale in the image header but I now think this is because I didn’t have cfa checked on bpp script before I ran it. 

I presume I should still have Raw selected in Format Explorer, Dslr-format, edit preferences ? Is this the default ?

[Scooot]

6 hours ago, SAW said:

r but I now think this is because I didn’t have cfa checked on bpp script before I ran it. 

I presume I should still have Raw selected in Format Explorer, Dslr-format, edit preferences ? Is this the default ?

Yes if you use the BPP script I think you're right. The Lightvortex tutorial explains it well.

As mentioned I dont use the Bpp script and so far have followed the Lightvortex manual calibration & registration tutorial and you can skip the part where you use the batch debayer script by changing the Dslr format in format explorer.

 Sorry for any confusion

[wimvb]

1 hour ago, Scooot said:

and you can skip the part where you use the batch debayer script by changing the Dslr format in format explorer.

???

Image calibration needs to be done on the raw image files, non-debayered. Image registration needs to be don on debayered images. If you leave out this step in the manual process, when do you get an rgb image. The raw/non-debayered image file is gray scale.

If you let PI debayer the images before calibration, you won't be able to remove hot pixels and fixed pattern noise properly.

The lightvortex tutorial describes the process (look under "workflow" for dslr images)

http://www.lightvortexastronomy.com/tutorial-pre-processing-calibrating-and-stacking-images-in-pixinsight.html

[Scooot]

1 hour ago, wimvb said:

???

Image calibration needs to be done on the raw image files, non-debayered. Image registration needs to be don on debayered images. If you leave out this step in the manual process, when do you get an rgb image. The raw/non-debayered image file is gray scale.

If you let PI debayer the images before calibration, you won't be able to remove hot pixels and fixed pattern noise properly.

The lightvortex tutorial describes the process (look under "workflow" for dslr images)

http://www.lightvortexastronomy.com/tutorial-pre-processing-calibrating-and-stacking-images-in-pixinsight.html

Maybe I’m not being clear. I change the dslr raw setting in format explorer to debayer - rgb, then follow the image calibration and registration process as per the lightvortex tutorial skipping the debayer part. The resulting image is colour not grey scale. 

As mentioned I’ve only done this on moon images. Maybe I shouldn’t but I don’t need to do cosmetic correction on them and there’s not much noise as they’re very short exposures. However the resulting colour balance is much better.

This is the last one I processed this way.

[SAW]

I have a colour stacked image :-)

I'm sure others could get better results after stacking but I'm just pleased I got this far. I've been playing around with a few process after watching loads on YT, just need some practice now.

[wimvb]

Now it's just a matter of colour calibration. Start with holding down the ctrl-key when doing the screen stretch. This will reveal the (almost) natural colour and any vignetting, as it un-links the rgb channels while doing the stretch.

[wimvb]

Btw, this has the potential of a great image. You have vignetting, but dbe will take care of this. The bright nebulosity shows through, even in the non-colour calibrated image.

[SAW]

Thanks, just need to learn the basics of which processes I need to do and which order.

I would of like to of used a master bias and dark for the stacking but just went with stacking all my bias and darks, I found it easier.

[wxsatuser]

On 10/20/2017 at 00:27, SAW said:

I have a colour stacked image :-)

I'm sure others could get better results after stacking but I'm just pleased I got this far. I've been playing around with a few process after watching loads on YT, just need some practice now.

 

Here is a simple set of processes that will get you near a reasonable image.

https://youtu.be/dmy8bsdLejA

[wimvb]

6 hours ago, wxsatuser said:

Here is a simple set of processes that will get you near a reasonable image.

https://youtu.be/dmy8bsdLejA

Sorry, but this video is best avoided. The guy basically doesn't understand what he's doing.

I've watched it until noise reduction (atrous nr isn't even supported any more), and there are faults in every process he uses.

1. Avoid resampling until the very end. It will destroy star shapes from being correctly sampled to being undersampled. And that only makes processing more difficult.

2. An area that has no stars, is empty for a reason. Probably because of nebulosity, which should be avoided for background neutralisation. This sample image is actually one of the more difficult images to process, even for an experienced person.

3. Using only a few selected stars works great with the new photometric calibration tool, but NEVER with the standard tool. It's better to use the stars in the entire image as a white reference.

4a. Dbe: too many and too small samples. He assumes that the dark nebula is supposed to be of uniform intensity. It will be after this.

4b. Don't discard the background model. Always examine it and evaluate its validity. Gradients are smooth.

4c. Even with normalisation checked, dbe can alter colour balance, so do colour calibration AFTER dbe.

4d. If you look closely, you'll see that samples at the top of the image are red. This indicates that they won't be used in calculating the background model. This leaves the image devoid of good samples where they are needed most. At the top.

5a. A trous wavelet noise reduction isn't supported any more. Tgv denoise or multiscale linear transformation is preferred. Both are WAAAAY better at preserving detail.

5b. Noise reduction should ALWAYS be done with a mask. You want to reduce noise only in low snr areas. @ollypenrice has pointed this out on numerous occasions. For a good reason.

5c. The guy selects 5 layers for noise reduction, but only uses three. He sets noise reduction to 0 in layer 4 and doesn't use it at all in layer 5. He can't use more noise reduction, because without a mask he risks smearing out the stars (which he does with his first attempt).

There are many much better videos available on youtube. Harry Page on this forum, Kayron Mercieca (lightvortexastronomy, no videos, but great walk throughs), Richard Bloch, Gerald Wechselberger, to name a few.

@wxsatuser, this of course NOT aimed at you. I greatly respect your input and opinions on this forum. I just got really (word better removed) off after watching this video, that I felt I needed to react immediately.

[ollypenrice]

17 minutes ago, wimvb said:

 

4a. Dbe: too many and too small samples. He assumes that the dark nebula is supposed to be of uniform intensity. It will be after this.

 

This raised a chuckle! You are perfectly right, of course.

Olly

[wxsatuser]

3 hours ago, wimvb said:

Sorry, but this video is best avoided. The guy basically doesn't understand what he's doing.

I've watched it until noise reduction (atrous nr isn't even supported any more), and there are faults in every process he uses.

1. Avoid resampling until the very end. It will destroy star shapes from being correctly sampled to being undersampled. And that only makes processing more difficult.

2. An area that has no stars, is empty for a reason. Probably because of nebulosity, which should be avoided for background neutralisation. This sample image is actually one of the more difficult images to process, even for an experienced person.

3. Using only a few selected stars works great with the new photometric calibration tool, but NEVER with the standard tool. It's better to use the stars in the entire image as a white reference.

4a. Dbe: too many and too small samples. He assumes that the dark nebula is supposed to be of uniform intensity. It will be after this.

4b. Don't discard the background model. Always examine it and evaluate its validity. Gradients are smooth.

4c. Even with normalisation checked, dbe can alter colour balance, so do colour calibration AFTER dbe.

4d. If you look closely, you'll see that samples at the top of the image are red. This indicates that they won't be used in calculating the background model. This leaves the image devoid of good samples where they are needed most. At the top.

5a. A trous wavelet noise reduction isn't supported any more. Tgv denoise or multiscale linear transformation is preferred. Both are WAAAAY better at preserving detail.

5b. Noise reduction should ALWAYS be done with a mask. You want to reduce noise only in low snr areas. @ollypenrice has pointed this out on numerous occasions. For a good reason.

5c. The guy selects 5 layers for noise reduction, but only uses three. He sets noise reduction to 0 in layer 4 and doesn't use it at all in layer 5. He can't use more noise reduction, because without a mask he risks smearing out the stars (which he does with his first attempt).

There are many much better videos available on youtube. Harry Page on this forum, Kayron Mercieca (lightvortexastronomy, no videos, but great walk throughs), Richard Bloch, Gerald Wechselberger, to name a few.

@wxsatuser, this of course NOT aimed at you. I greatly respect your input and opinions on this forum. I just got really (word better removed) off after watching this video, that I felt I needed to react immediately.

I beg to differ, it's pretty obvious it's aimed at me considering I was only trying to help.

Well it's a pity you never jumped in and gave the OP a set of procedures to try in the first place instead of, 'it will be a great image'.

It was'nt meant to be a perfect  example, just something he could try and move on from there.

I will leave it there and you experts can get him on his way, but give him more than, it will be a great image.

[SAW]

Ok this may be a bad tutorial to some but for me this is what I’m looking it basically shows what processes and what order and where to click, the really basics of which I was looking for. It will give me something to use with my stacked images that I got from the BPP script because at the moment I’m just trying all sorts to of processes with no order in anything and not really getting anywhere, it feels like I’m back using LR and just moving sliders hoping for it to look good lol  

While the PixInsight book is good I feel it’s a little advanced for my level of use in software but watching tutorials then going back to the book things start to click and make more sense  

 

[wxsatuser]

The chap does say it's a down and dirty way of doing it.

You can try other videos and pickup bits and pieces here and there.
Some other tutorials were mentioned, try them all. It takes time to pickup everything PI can do
and just take your time and eventually it will come.

I don't claim to be an expert at PI or any other software and know how hard the learning curve is.

[SAW]

I've noticed in a lot of the video's that the images people are working with are very faint and I think the trouble I'm having processing that M45 image is because I used 1600iso it is just so noisy, maybe I should of gone with 800 and had a fainter image to work with ?

[ollypenrice]

No cause for a spat here. I know little about PI but I've spent thousands of hours in Ps and, on occasion, have advised people to ignore tutorials recommended on here. This isn't an attack on the perfectly sincere person who found the bad tutorial useful and recommended it. It is an attack on the bad tutorial. So I think Wim was right to say what he did. He recommended a number of competent sources in an entirely non-confrontational way. I'd call that good forum behaviour. I don't see a problem.

Olly

 

[SAW]

Well this is about the best I can do in PI for now, suppose it's not too bad but going to image with a lower iso from now as there is loads of noise. 

[wxsatuser]

Don't worry you are doing ok, learning curves are quite steep in this game.

Noise is always a problem for dslr cameras, if you can get guiding, dithering will certainly help with the noise.

[SAW]

Ran the SCNR as someone recommended on FB which has definitely helped with the green removal.

Cooled camera is next on my list. Probably stick with a colour otherwise mono and filters will add in even more learning !