23 replies to this topic

[Rolf]

Hi,

 

Combining a high-resolution B/W image with a low-resolution color image into a high-res LRGB composite is a standard technique used by deep-sky photographers. The non-trivial step in this process is the precise alignment (registration) of the two images. For deep-sky images this is easy because the stars provide very convenient keypoints. Many standard programs do this registration automatically.

 

For lunar images this is not so easy, because there are no stars to guide the image registration. I did not find any suitable software for this task, so I decided to write my own program. The result is called "PlanetarySystemLRGBAligner". It is open-source software and can be downloaded from Github. The program runs on Windows 7 / 8 / 10 and comes with an automatic installer. This user guide explains the program and its functions in detail.

 

To give you an impression of what can be achieved with the LRGB technique on the moon, please have a look at the following example: This screenshot of the program shows a zoomed-in view of the B/W image:

 

 

The second screenshot shows the color image, after it is precisely registered with the B/W image:

 

 

The two views show exactly the same area on the moon, so the registration seems to have been successful. The resolution of the color image is much lower as compared to the B/W one. The program then produced the following LRGB composite image:

 

 

This color image has the same resolution as the B/W image.

 

The screenshots only show smal sections of the big panoramas. The full view can be downloaded from here:

- B/W image

- Color image

- LRGB composite

 

If you are interested, feel free to download the software for your own experiments. Your feedback would be very welcome.

 

All the best,

 Rolf

 

 

[Lord Baltimore]

Will give it a download and try it when the weather allows me, haven't had a clear sky in months.

[troyt]

Thanks for making it available i'm sure many will find it useful. 

[vilchez]

Thanks, it can be useful.

[PiotrM]

With WinJUPOS I often made X-RGB images (where X was one of luminance channels - shot via L or other longpass filter). That app also corrects for planet rotation.

[Rolf]

Hi Piotr,

 

WinJUPOS and my program deal with completely different problems: WinJUPOS de-rotates planetary images using a deterministic algorithm. If you use it to de-rotate an AVI, the single frames are as much warped as before. Only the rotational effect is removed. Image warping is dealt with by the stacking software (e.g. AutoStakkert!3) after de-rotation.

 

My software is used after stacking. If you want to create an LRGB image, the L and RGB images must be registered precisely, or bad color artefacts will show up. The higher the quotient of image size and the smallest resolved feature, the more difficult this registration process is. For planetary images the quotient is relatively low (small image, relatively few resolved features), so usually it is sufficient to register the L and RGB images by hand. For the moon this is hopeless. It is very large and contains very fine detail. You need to do something special to register the images. That is what my program is used for.

 

Best regards,

 Rolf

[PiotrM]

My software is used after stacking. If you want to create an LRGB image, the L and RGB images must be registered precisely, or bad color artefacts will show up.

That's the part of WinJUPOS I was talking about. Having measurements for the stacked images I can do LRGB images in that app. It works only for planets, while for Moon I had to use Nebulosity and trick it multi-star alignment on Moon surface details and I needed perfect alignment for petrographic images where one image divides another. If your aligner would also be able to save aligned images and not only their LRGB result it would also be handy for this type of imaging

[Rolf]

Hi Piotr,

 

As I wrote in my previous note, LRGB alignment of stacked images is trivial for planets since the smallest structures are relatively large compared to the overall image size. You can do the alignment by hand, or let WinJUPOS do it automatically. WinJUPOS, as far as I can tell, does not even try to de-warp the images.

 

Of course you can abuse a deep-sky software for registering moon images by manually placing "stars" at selected surface locations. Since image warping is a very local phenomenon, for high-resolution lunar panoramas you need many alignment points, and you have to place the "stars" with very high accuracy. That's nothing I want to do by hand.

 

 

If your aligner would also be able to save aligned images and not only their LRGB result it would also be handy for this type of imaging

Did you hava a look at the documentation of my software? You can save the aligned color image (GUI button "Save registered color image") as well as the LRGB result (button "Save LRGB composite image").

 

Best regards,

 Rolf

[PiotrM]

Did you hava a look at the documentation of my software? You can save the aligned color image (GUI button "Save registered color image") as well as the LRGB result (button "Save LRGB composite image").

I was looking for saving each aligned "channel" and not end result RGB or LRGB. I need to apply one image as a flat of another to get that petrographic images.

[Rolf]

Hi Piotr,

 

My program does not care about the internals of the color image. It applies the same transformation on all color channels. Therefore, if you want to register a monochrome image with the reference image, just change it to RGB, run my program with it, and change the registered image back to monochrome.

 

Best regards,

 Rolf

[Euphonia]

Hi Rolf!

Congratulations to what is a fantastic and well needed initiative!

I've been searching for this type of solution for a long long time, and I'm quite eager to give this a try.

 

A few questions.

1# Do you suggest using finished high res mosaics vs individual raw panels?

Individual L+RGB panels might better avoiding unintentional minor warping during the mosaic process, whilst data panels from two different cameras might simply not be practical due to unmatched coverage and FOV

 

2# How do you deal with cropping and overlapping data? Many times two mosaic ends up having rather odd shapes and don't follow a strict rectangle shape. How do you merge two oddly shaped images? It would be nice if any spillover is left alone without being cropped.

 

3# Can the finished registered image be re saved into separate layers for later manipulation in Photo Shop or similar?

 

4# Have you thought about incorporating some sort of mosaic canvas in your app? How do you assemble your mosaics? Which app do you personally prefer?

 

Looking forward hearing your thoughts

 

Cheers!

Tobias

Edited by Euphonia, 22 April 2019 - 01:41 PM.

[Rolf]

Hi Tobias,

 

Thank you for your nice words!

 

Here are my answers to your questions:

Ad 1: The whole idea of my program is to deal with the complications caused by full panoramas. Therefore, I assume that the whole moon is contained in both panoramas. Otherwise the keypoint matching in phase 1 could get in trouble.

 

Ad 2: Since I always capture complete moon panoramas (for automating this tedious process I developed the program "MoonPanoramaMaker"), I don't have those strange mosaics with odd shapes any more. I don't know how my program should handle the case where the two panoramas do not cover the same region. In that case I would crop the two panoramas manually so that they almost match. Then my program should work fine.

 

Ad 3: The monochrome image is not altered by my program, only the colored one. You can save the rectified color picture into an (RGB) file. This color version matches the B/W one in size and orientation. Therefore, you can import both into Photoshop and manipulate them as you like. In particular, you can split the color channels, if that was your question.

 

Ad 4: I never write a program to solve a problem for which satisfactory solutions exist. There are so many good panorama programs around, so adding this functionality in my program would be a waste of effort. I personally prefer "Panorama Studion 3 Professional".

 

All the best,

 Rolf

[james7ca]

Thanks for making this available. In the past I've used PixInsight's FFTRegistration tool to align color channels for lunar images but it usually doesn't work that well with high resolution images because of local distortions. Does your program make any attempt to handle distortions?

 

In any case, I definitely plan on giving it a try and thanks again.

[Rolf]

Thanks for making this available. In the past I've used PixInsight's FFTRegistration tool to align color channels for lunar images but it usually doesn't work that well with high resolution images because of local distortions. Does your program make any attempt to handle distortions?

 

In any case, I definitely plan on giving it a try and thanks again.

Yes, my program handles local distortions (pixelwise). Please have a look at the user guide for details.

[Euphonia]

Rolf,

Ok, that all makes sense.

I'm gonna try your software on a single frame target first, just to get the hang of it.

 

I've looked at your moon panorama maker with envy. I'm stuck with an AltAz CPC1100 HD, so it's not compatible.

I'm wondering if adding a wedge would be enough to do the trick though?

 

Have you ever tried high res lunar imaging with a filter wheel? Im assuming you've made the choice to go with OSC due to the fast moving shadows?

 

I'm setting up my rig for a trip to Barcroft station in early June (Whitemountain California 12500ft elevation) for some high res moon imaging. Due to limited time and to maximize the time at the mountain, Ive chosen to use ZWO ASI178CM and the ASI290MM ( with a few filter options depending on seeing).

Now I'm just trying to figure out the smartest way to handle the logistics of the data gathering. Still have some time to plan it all out.

 

Cheers!

Tobias

[Tom Glenn]

Rolf, excellent work!  I tried your software for the first time today.  I didn't want to pollute this thread with a ton of my own stuff, so I created a new post to present a few images that used your software.  

 

https://www.cloudyni...s-new-software/

 

My general assessment is that the software is outstanding, and was incredibly easy to use.  In fact, it is so easy to use that I will probably collect color data more often in the future to test this out with some of my higher resolution images.  I'm also excited to see how your other stacking project progresses.  Keep up the good work!

[Rolf]

Hi Tobias,

 

I've looked at your moon panorama maker with envy. I'm stuck with an AltAz CPC1100 HD, so it's not compatible.

I'm wondering if adding a wedge would be enough to do the trick though?

 

Have you ever tried high res lunar imaging with a filter wheel? Im assuming you've made the choice to go with OSC due to the fast moving shadows?

 

I'm setting up my rig for a trip to Barcroft station in early June (Whitemountain California 12500ft elevation) for some high res moon imaging. Due to limited time and to maximize the time at the mountain, Ive chosen to use ZWO ASI178CM and the ASI290MM ( with a few filter options depending on seeing).

Now I'm just trying to figure out the smartest way to handle the logistics of the data gathering. Still have some time to plan it all out.

 

Cheers!

Tobias

I don't know if adding a wedge solves the problem. MoonPanoramaMaker assumes that the mount controller understands ASCOM commands for "goto RA/DE" and pulse guiding in RA/DE. If you add a wedge to your mount, I guess that you also need to switch the controller to RA/DE mode. Perhaps it then understands the ASCOM commands. I just never did such a thing, so I don't know.

 

No, I never used a filter wheel on lunar imaging. And I don't think that this would make sense. As you wrote, the changing illumination would be a problem. Also, it is not necessary to record the color information with the same detail as the L channel. Actually, I do not think that using an USB camera for color capturing is a good idea. I gave up after many trials with an ASI 178MC on my Celestron 11. The problem: while paonorama programs can adapt brightness and contrast at tile boundaries, they cannot adapt subtle color differences, in particular when you have to raise the color saturarion in the case of the moon. With so many tiles in a whole-moon panorama this results in bad color variations across the lunar disk.

 

It is much better to use a DSLR (in my case a Canon 5D MKII) on an Apo refractor for color capturing, and to combine that with the hi-res B/W image captured with a USB camera on a larger telescope. Then you only have to combine a few color tiles, and the panorama program can handle the problem. Interestingly, the only other person I know to have mastered high-res LRGB moon photos did the same. He used a Canon 40D for color, and a monochrome USB camera for the L channel. Unfortunately, he never wrote about his technique, and he did not publish the program with which he combined the two panoramas. But since he was one of the few people who successfully developed a high-res stacking program, he must have been able to write a program similar to my PlanetarySystemLRGBAligner.

 

I wish you good luck with your expedition to Whitemountain California! To bad you cannot use MoonPanoramaMaker. It really makes the logistics of data capturing much easier.

 

All the best,

 Rolf

[Rolf]

Hi Tom,

 

Rolf, excellent work!  I tried your software for the first time today.  I didn't want to pollute this thread with a ton of my own stuff, so I created a new post to present a few images that used your software.  

 

https://www.cloudyni...s-new-software/

 

My general assessment is that the software is outstanding, and was incredibly easy to use.  In fact, it is so easy to use that I will probably collect color data more often in the future to test this out with some of my higher resolution images.  I'm also excited to see how your other stacking project progresses.  Keep up the good work!

Thank you very much for your positive comments! I'm glad that you like the usability of the user interface of my software on which I spent about 90% of the development time. It's hard to believe, but the pure computing part of the PlanetarySystemLRGBAligner is about 125 lines of Python code, whereas the overall software is more than 2000 lines.

 

It's similar with my new PlanetarySystemStacker. I finished the user interface and built an automatic Windows installer. I did not yet write a user guide, though, so I don't want to give the software to others. At the moment I'm experimenting with sharpening filters for postprocessing which I want to integrate in the stacking program.

 

All the best,

 Rolf

[Euphonia]

Rolf,

 

5D mkII gives you 6.41µm and I'm assuming full disk coverage? What size APO do you use?

Im guessing 4 inch APO (?), so that's  900mm x 6.42µm RGGB vs 7000mm x 2.9µm mono, that's a HUGE difference in resolution (!?) I'm honestly surprised by the large difference in resolution, yet getting such amazing results.

 

How many subframes do you usually do per color panel? Do you exposure bracket?

 

Now... Just thinking out loud here in terms of maximizing data, resolution etc

I could totally see a case for using larger sensors to avoid stitching errors (obviously some amazing results have been seen), but what about other factors like noise, mirror slap and simply smaller pixels for higher smoother color resolution? Not to mention wear on your shutter (1000 frames = 1% of total shutter life).

 

I've been using a Sony A7sii (large 8.6µm pixels) via USB and with electronic shutter selected, but during moments of good seeing the total amount of frames feels quite limited compared to what one can accomplish with a smaller sensor. Also, no active guiding with a DSLR ( The active feature tracking in SharpCap is quite awesome for lunar ). With 1min sub sets, I feel you can always get more frames with a USB3 and a laptop vs filling up the buffer on a DSLR.

 

My thinking behind getting the 178mc was to get higher frame rates (more frames = less noise etc) smaller pixels and a 14bit sensor and the ability to use active tracking.  6.4mp seemed like a good balance of speed vs resolution, but perhaps I need to revise that idea and prioritize a larger sensor.

 

Im wondering if the 071mc would be a better solution? APS-C, good dynamic range, great QE, 4.78µm pixels and 8-10fps. 

 

Cheers!

Tobias

[Euphonia]

"Blackmagic Design Pocket Cinema Camera 4K" is another interesting contender for lunar photography

 

https://www.bhphotov..._camera_4k.html

 

IMX294 sensor, 60 fps RAW and you can capture directly to a USB-C drive.

 

60fps with full RAW sensor readout is a LOT of data

 

Obviously you forego cooled sensor and most of the benefits running standard capture software, like tracking etc. But for lunar you don't really need a cooled sensor anyway. Plus you can always tether the BMPCC 4k to your laptop as a focusing aid.

[Rolf]

Hi Tobias,

 

I'm using a three lens apo refractor 130/1200mm (LZOS). I extend the focal length to some 3500mm, using an Baader fluorite FFC. I need two tiles to cover the half moon, and four tiles to cover the full moon. There is no mirror slap because I set the camera to live view mode. Also, because I use a diagonal mirror, the shutter operates in the direction of the telescope axis. Therefore, there is no lateral momentum which could blur the image. With this setup I get diffraction-limited views of the moon (at the 130mm aperture).

 

The difference in resolution, as compared to my B/W panoramas captured with a C11, is about a factor of two. That's perfectly acceptable for LRGB work.

 

I understand your theoretical arguments pro / con DSLR. I can only say that I tried both the Canon on the refractor and the ASI178MC on the C11, and the results with the Canon were highly superior.

 

All the best,

 Rolf

[nth_human]

Rolf,

Thank you so much for this software. This problem has been driving me crazy.

'PlanetarySystemLRGBAligner' is exactly what I've been looking for to match localised lunar surface warping on separate RGB mono stacks acquired with RGB filters and a mono camera.

I just pick one channel as the B/W reference, convert the other channels to RGB images, and individually load them as the 'colour' image and your software registers them perfectly. Then I can convert all the channels back to mono images and combine them for the final colour image. I've been unable to find any other software that can easily do this.

Thanks again.

Aaron

Edited by nth_human, 16 October 2021 - 07:51 PM.

[satheeskr]

Thank you Rolf. I’ll try the software for one panel with 290mm for the mono and 224mc for the colour.

[Asixoid]

Amazing piece of software!!

Thanks a lot!!