24 replies to this topic

[xatamec]

I shot this on March 17th and have been since then mulling over how to process the shot to pull out the Integrated Flux Nebula while avoiding the galaxies and the stars to blow out. Finally I found my way to do it and got a result I like considering the exposure time I could collect.

Camera: Cooled Canon 350D, hardware modified to remove the amp-glow
Telescope: William Optics Megrez 88FD with Borg DG-L reducer/flattener
Mount: SW EQ6 Pro
Images: 55x360"+18x90"+14x30" @ ISO 1600, callibrated with darks, flats and bias
Processing Software: DSS and PixInsight 1.5.8



Link to larger version HERE

Sergi

[Rankinstudio]

That is fantastic! So much faint detail.

[Jeronimo Cruz]

Great job!

[jmasin]

Exceptional work and processing for ~6hrs worth of data! Very nice!

[PeteM]

Amazing Sergi, I have never seen this area processed like this to include the Integrated Flux Nebula. Very well done.

[templec]

Very nice, Sergi! You don't see images like this very often. Well done! I have a question: you say your 350d is "hardware modified to remove the amp-glow"...how did you do that?
- Craig

[sponka]

hi sergi

great picture !!!
you are using PixInsight too :-) great software for great pictures

stefan

[gatsbyiv]

Fantastic image, Sergi. Would you mind sharing your processing techniques for bringing out the flux without blowing out the galaxies? Was it layer masking or something else?

[Illinois]

Wow! Lot of details and faint galaxies! Great job!

[nickatnight]

Outstanding image.

Love to hear your PixInsight workflow.

[Nils_Lars]

Wow great job with the outer flux , I too would love to hear a workflow.

I really like the way you balanced the dusty stuff and the cores and the color , superb shot Sergi.

[Mike Clemens]

This is awesome!

[John Kocijanski]

Great shot. Just looked at this area of the sky the last time I was out. Thanks for sharing.

[xatamec]

Thank you everyone!!

Craig, I made a modification to the camera to eliminate the amp glow. The 350D keeps the sensor amplifier active during the long exposures and this produces the glow, don't know exactly why. We need the amplifier active only at the beginning of the exposure and during the readout, but not during the exposure. I made a circuit which feeds the sensor with 3.5V during the exposure, instead of 5V. This makes the amplifier to be inactive and the glow is not produced. The circuit is activated by a switching signal which is taken from the camera board.
The first to make an amp-off cicuit for a Canon 300D was Mike Kudenov. I asked him about the possibility of doing the mod for the 350D. He gave very valuable information and helped me a lot, although he hasn't finished his mod yet. A bit later I knew about the work of a guy from the Netherlands who had done the amp-off mod for the 350D succesfully, Cas Wilders. He gave me the idea where to take the switching signal from. However, my amp-off circuit is different from Cas' circuit, it is based on Mike's circuit for the 300D, just a bit different. Here you have links with more information about the mod:
http://glogg.jupiter...off/amp-off.asp
http://members.chello.nl/c.wilders/

And here you have a link to a post from a spanish forum where I published my work. I have used google translator to translate it to english (sorry the translation is far from being perfect): http://translate.goo...684&sl=es&tl=en


gatsbyiv, nickatnight, Erik: don't know if you know PixInsight. The processing philosophy and the tools is radically different from PS, so maybe you don't undestand part of my processing workflow. If you need more information please tell me:

1) Callibration and stacking of the exposures with DSS: I got three images: 30", 90" and 360".
Starting of processing with PixInsight:
2) Color calibration: doing a preview taking only M81. I assumed all the flux from M81 must be neutral (it is a good assumption for Sa, Sb, Sc, Scd, SBa, SBb, SBc and SBcd galaxies). With this step I got the coefficients to apply to the G and B channels. I did this for the three images.
3) Background neutralization: this is a PixInsight tool, you need to choose a preview from an area where there is only background. I did this for the three images.
4) HDR Composition of the 3 images: this is another Pixinsight tool. I got an only image with the contribution from the 3 images with different exposure times.
5) Masked Stretch Transform: this tool does an iterative stretch of the histogram masking the image with its inverted luminance at each iteration. This prevents the stars from growing up and the brightest parts from getting saturated.
6) At this stage I created a mask for the galaxies and the stars and their halos. This is the trickiest part. Pixinsight has a powerful tool based on wavelets transformation to create masks. With this tool you can separate parts of the image based on their scales. So I firstly made a mask for the stars, then a mask for the star halos, and then a mask for the galaxies and then combined them all. All this stuff took me several hours, you have the tools but you need to fine tune the paremeters to create a mask with the desired sizes for all the objects to protect in the image.
7) With the mask applied to the image, I did a strong histogram stretch and the IFN came up... and also a lot of noise.
8) Noise reduction with ACDNR. First on the chrominance, and then on the luminance of the image.
9) Final histogram adjust.
10) Soft stars reduction with a star mask and a morphological filter.
11) Final touch, color saturation: first to the stars and after to the galaxies, both steps with masks.


Sergi

[gatsbyiv]

Thanks for the workflow Sergi!

I do have PixInsight and am trying hard to like it, but find it somewhat frustrating. I love things like background neutralization, but then I see what you went through to create a mask and compare that to how easy it would be in Photoshop. I know their philosophy is different, but I also think they underestimate the usefulness and simplicity of Photoshop's philosphy!

Again, great image, though. Very creative and well executed.

[Nils_Lars]

Thanks for the workflow Sergi , I am familiar with PI since I used it for a few projects with the trail version and I havent had the funds to purchase the full version yet but I plan to.

[David Pavlich]

That one needs a frame!

David

[Doubleglaze]

Nice shot! The balance between detail, color, and the overall background field is great.

Mark

[alpal]

That is a very nice photo indeed.

[Tonk]

Hey! thats truely superb!

[xatamec]

Thank you all for your comments!

Thanks for the workflow Sergi!

I do have PixInsight and am trying hard to like it, but find it somewhat frustrating. I love things like background neutralization, but then I see what you went through to create a mask and compare that to how easy it would be in Photoshop. I know their philosophy is different, but I also think they underestimate the usefulness and simplicity of Photoshop's philosphy!

Again, great image, though. Very creative and well executed.

Honestly I can't give you an opinion about processing with PS because I started processing my first astrophotos with the former free version of PixInsight and after some time I moved to the commercial version. I think that if you get used to an application and its tools, changing to a different one is more difficult than having started from the beginning with it. You get used to a certain workflow and you miss you can't do the things the way you know. The key is having the patience to spend enough time to get to the point where you understand how to work and achieve results with your new toolbox, and then you can make a fair comparison between the two applications. I know that the people that have made the effort to understand PI coming from PS, have stayed with PI and haven't come back to PS. Although I agree with you that there should be a tutorial for those users coming from the PS world. Having said that, they have a lot of processing examples and very good video tutorials in their web site.
I don't know which tools PS has for creating masks, but I am almost sure that I wouldn't create the mask in PS in much lesser time than in PI. You can create a star mask in PI in 30 seconds, and a galaxies mask in 60 seconds. But, if you want to make a strong stretch to an image, your mask must be very, very precise, otherwise you will get dark or bright rings around the structures being masked. I spent so much time building the mask because I needed a lot of trial and error to determine the optimal size and the optimal blurring for the elements of the mask to let it me make the strong strech without having these undesired effects.

Sergi

[Tonk]

I am almost sure that I wouldn't create the mask in PS in much lesser time than in PI. You can create a star mask in PI in 30 seconds, and a galaxies mask in 60 seconds. But, if you want to make a strong stretch to an image, your mask must be very, very precise, otherwise you will get dark or bright rings around the structures being masked

This is an excellent point. Absolute care is the essence here.

You can readily perform these highly controlled masking opperations in PS in a highly interactive way with constant feedback.

For those wishing to use PS it can be noted that if you process via layers, masks can be applied retrospectively. You add a mask to the desired layer (a curves layer, colour saturation layer etc etc) and then drop in a copy of the last non-layer image in as the initial mask. (If you apply a curves/saturation etc operation directly to the current image - you can't tweek these iteratively or mask them retrospectively. If its not what you wanted your only recourse is "undo" and start over)

Processing the mask in PS is best done using a combination (and sometime iterating cycles of) the blur filter and the curves tool. The blur filter allows you to control the amount of edge feathering to blend the masked and unmasked regions cleanly. The curves tool allows you to stretch the mask and readily increase/decrease the mask contrast and hence the scope of the "pass through" and "excluded" regions and the shape of the gradient in between. That way you can create beautifully accurate masks fairly rapidly.

Finally you can tile both the mask and the final image windows side by side and watch in real time the effect that altering the mask has on the actual image.

These techniques are very versatile and are the basis of manual HDR processing (see Gerry L's M42 tutorials) as well as the controlled luminance, and colour stretching that Sergi is doing. I've used them when L*A*B processing globular clusters to do selective luminance sharpening and colour blurring.

I've never used PI so it would be interesting to hear how interactive this mask processing can be. How does it compare to the description I have given for PS?

[Nils_Lars]

Actually I have heard people say that layers are the weak point of PI but there is a new version coming out with lots of new things , im not sure if this is one of them.

[xatamec]

Hi Tonk and Nils,

This is very interesting. The use of layers is one of the key differences in the processing philosophy between PI and PS. I've also heard the comment that Erik points from people used to processing in PS. I have never used layers, so I don`t miss them. It's not worse or better, you just do things a different way.
In PI you can also see in realtime the effects that altering the mask will have on the next process that you will apply to the image using Real Time Previews.
Look at the following screenshot:



I am going to apply a histogram transformation to the image in the upper left corner, I've choosen to display a Real Time Preview (lower left corner) to see the effects of this process on my image prior to applying it. If I move one of the sliders on the histogram transfomation process window I will see its effect real time.
Also, the image is masked by a star mask (just for the bigger stars). The mask is on the upper right corner. I've enabled the view of the mask on the actual image, so you can see in red the areas being protected by the mask.
I can enable or disable the mask at any moment and see its effect on the Real Time preview. I can also apply any kind of process to the mask, for example a curves transformation, and I will see its effect on my actual image in real time on the Real Time Histogram Preview, all this prior to actually applying the Histogram Transformation Process to my image.
I can do all this also to small zones of my image to have a close up of, for example, a star. So I can see what is the effect of changing the mask on the image before applying actually any kind of process.

Sergi

[xatamec]

There is a very good post in the PixInsight Forum about the topics we discussed here: layers, masks, etc:

http://pixinsight.co...6;boardseen#new