Manual Star Alignment ?

[Eddy Timmermans]

Is there a way to manually align images into a mosaic ?

Yesterday I downloaded 18 images of M104 from the HLA-site.
In the footprints tab they overlap quite a bit so I thought I could merge them into a mosaic.
Even after carefully cropping them there still were several stars that appeared in both images.
But Star Alignment didn't find enough of them.
So is there a way of manually merging such images into a mosaic ?

Eddy

[Alejandro Tombolini]

Hi Eddy, if you search in the forum there are many discussions about mosaics. (Using previews for small overlap).

I think that you can also use Dynamic Alignment to align the frames and after that GradientMergeMosaic Module

Regards, Alejandro.

[Eddy Timmermans]

I retried the StarAlignment tool and this time I checked to see if there were corresponding stars in the intersection.
The first time I tried to include as much as possibe of the galaxy itself.
I succeeded in merging 5 of the six frames in one channel, the last one didn't fit, even though I saw many corresponding stars.
The image so far is really stunning to me. I really would like to finish it even though I think the guys at Hubble did it before me.

It isn't quite clear to me how to use Dynamic Alignment. I tried it on 2 frames and it produces an image with a white rectangle after STF auto stretch with no stars visible at all.
What am I supposed to do after that ?

Eddy

[ruediger]

... it produces an image with a white rectangle after STF auto stretch with no stars visible at all.

Try to configure the STF auto parameters (CMD + Click on 'A'). The default target background value of 0.25 is too high and not reachable on pictures with lots of black area, therefore rendering all other values to white.
You can also drag/drop the STF settings from the unregistered picture the registered one or you manually adjust the sliders.

Rüdiger

[cfranks]

Eddy, 

Steve's video:   http://pixinsight.com/forum/index.php?topic=3369.0   explains those white images.

Charles

[Alejandro Tombolini]

Hi Eddy, I am not sure if I understand your problem, but if I am not wrong the manual alignment could be solved as follow:

Imagine you have two image that are possible to merge with GradientMergeMosaic Module but you need to manualy align. They are name Image1 and Image2



Select Crop tool and expand the image that is going to be used as source image (image1) in the side where image2 is going to be aligned.



Select DynamicAlignment tool and choose image1 as source, and image2 as target.



Select stars in the source and check that the samples are in the same stars in the target. The first and second will need action from you and the following probably will be automaticaly in the correct stars. Always check all samples.

Save the STF from the actual image1 (or 2) dragging the triangle in the workspace.



Execute Dynamic Alignment and after that drag the STF icon over the new image to see it.



Now you have image1 and Image2_registered that must be saved to use with GradientMergeMosaic Module.
Once saved, open GradientMergeMosaic Module and select both images and execute it.



Finally a big thanks to Georg for the fantastic Module.   

[georg.viehoever]

Finally a big thanks to Georg for the fantastic Module.   

[Eddy Timmermans]

Alejandro,

I tried your procedure and it works but now I have a difference in background brightness.
Do I have to change something before the alignment ?

This is the result I get,using only star alignment

Eddy

[Eddy Timmermans]

The image I sent is a mosaic using 4 of the 6 images from Hubble

The last two are not yet in the picture 

Eddy

[Alejandro Tombolini]

Do I have to change something before the alignment ?

Hi Eddy, this is going to be a fantastic image.

I think that is better if you perform DBE first, and have all images without gradient.

Saludos. Alejandro.

[Eddy Timmermans]

Alejandro,

I will try DBE but the last image that doesn't fit in, looks noisier than the other ones.
When I apply a STF with an auto stretch  from another image, the colors are all of. the same applies the otther way around.

So I'll try to work on the background . This challenge will keep me busy for a couple of nights, days, weeks :-).
But my mount broke donw so I have the time.

Eddy

[Eddy Timmermans]

Alejandro,

The leftmost 2 images have a different background than the other 4. These blend beutifully together but when I try to join the leftmost 2, a clear line is visible. Even with DBE I can't seem to fix that.
I suppose I have to keep the mages in linear form to do the mosaic, so how do I fix this ?

Eddy

[georg.viehoever]

It is not necessary for GradientMergeMosaic that the source images are still linear. But so far, it seemed to me that if you do the merging further down the processing pipe, it would be more likely that you introduced differences in the image characteristics that GMM cannot fix (e.g. different color tones, different noise patterns,...). GMM can only fix brightness differences.
Georg

[Eddy Timmermans]

Georg,

I thought as much. That is also why I think that the difference is in the image itself.
I will check the exposure times on the HLA site.

Still, your tool works magnificently, with star alignment and with DynamicAlignment.
It is impossible to detect any seams.

PI is the best software I purchased in years, and I believe I am still only scratching the surface after 4 months of using it.
And the learning curve is not that steep at all, only if you want to produce scientific grade images, and that is not my goal.

Eddy