Adding Ha to an RGB Image

[swag72]

In my IC2162 image I have recently posted, I combined my Ha data into my RGB data. It appears, once the Ha has been processed very roughly on it's own that I have lost detail once the Ha has been mapped into the red channel.

Before you add the Ha into the Red channel, do you process the Ha at all or add totally unprocessed data?

Can anyone suggest why detail was lost when the Ha was added?

For information I was using Pixinsight, and followed Harry's tutorial on adding Ha into RGB data. The principals should be the same regardless of software used though?

Would welcome your thoughts on this as I get to grip with mono CCD imaging!

[ollypenrice]

I don't (yet) use PI for Ha combining into red but I do aim to give it a try.

I think you absolutely must process both RGB and Ha separately before combining. I also process the Ha which I'm going to combine with red a little differently from a nice standalone version. When combined, everything about the Ha - the lovely crisp detail, the sharp contrasts, the sense of 3D relief, will all be diluted. So the trick is to crank them up big time before you apply the image. Lots of sharpening, lots of work on local contrasts.

Also it is is hard to know in advance how to set the brightness of the Ha image. Using my Ps technique I can look at the effect of my Ha layer on the red layer and adjust the Ha accordingly in Curves before flattening. I've found that you can stretch the Ha above its noise limit without this adding visible noise to the HaRGB.

Using PI I suppose you could prepare different Ha stretches and see which one gave the best result.

Olly

[harry page]

Hi

As you know Pi adds the Ha before a stretch

when working with a nebula , my formula may not need much of multiplication factor , try just one to start with

Harry

[steppenwolf]

I don't use PI, I'm afraid but when I blend Ha into my RGB data, I do process the Ha extensively first to obtain a mono image that stands out in its own right. I then blend the Ha into the red iteratively until I get the richness I seek.

Some images also respond well to using the Ha as a Luminance layer after the above process has been completed but only at an opacity of under 50% to avoid muting the colours.

[swag72]

Thanks for reading and answering all - What is most interesting (and I would welcome some thoughts on this Harry if you have time) Olly and Steve, both processing their data in CS, process their Ha data extensively PRIOR to blending it. In PI, you have to add a linear and largely unprocessed image - How much will this then affect the intensity of Ha using PI as opposed to CS?

I can't help thinking if I could add my processed Ha data to the red channel, I'd have much more detail - Is there any way to do this Harry?

[harry page]

Hi

There is a way to do it , but as I have not done it before I need a hour to think about it

Do you want to send me the files to look at , so I can see the problem

Up to you of course

Harry

[swag72]

Of course Harry - What do you want? The stacked Ha and stacked RGB?

[harry page]

Hi

Yes that will do

If you can't email them to me , Let me know and I will send my FTP details

Harry

[ollypenrice]

I'm very intrigued by the idea of using unprocessed Ha at the combining stage. It runs counter to all my instincts since, for instance, I feel I want sharp Ha but not sharp red. I'm not saying my instincts are right and when I get a chance I'll try Harry's PI video and method. I feel that there are qualities in an Ha layer which are unique to that layer and I want to exploit that. However, since my whole approach is 'see what works' I'll be downloading another 'Harry' in the near future!

Olly

[harry page]

Hi

I see you have has a full Juan reply

You can use my way to add together the data non linear wise , but the crunch in Pixinsight is to use the "Linear Fit tool " to match up the frames before adding them together , even if they are non linear .

The Linear fit tool makes sure thing like the background are the same , so you do not get miss balanced images

Regards Harry

[swag72]

Cheers Harry. I will sift my way through Juan's reply and hope that I can extract the pixelmath knowledge to do it. The 'Linear fit tool' (?) you learn something new every day!! I will have a look through that as well.