8 replies to this topic

[DecepitconFromOrion]

I'm trying to make a timelapse video of the eclipse from start to finish. I shot the photos with a 250 mm focal length lens on a Canon 70D without an equatorial mount, and I had to make adjustments to the static tripod throughout so that the sun wouldn't go out of frame. Thus, the sun is moving around a lot, and I can't apply a simple stabilization algorithm to the the time lapse animation. 

 

I'm sure I'm not the only one with this problem. How can I align my frames, other than extremely time consuming manual labor? It's even more challenging since the appearance of the sun changes so drastically from start to finish, so I doubt something like autostakkert would work, right? 

[Daniel Dance]

You're going to have to do it manually in photoshop or similar program, and its really going to SUCK because now you have to rotate each frame as well to match the sunspots, etc...

 

There is NO easy way out of this one my friend.  There is no software that will do this without it looking like crap.

 

1.  Pick one frame as a reference frame and use that to align ALL your images.

2.  Load say 20 images at a time into photoshop.  FILE - SCRIPT - LOAD STACK (include your reference image)

3.  Usually you would use DIFFERENCE blend mode to align, but since you also need to see sunspots and worry about rotation, just change transparency of each image.

4.  Control-T layer will give you handles for rotation and movement.  Make sure to unselect AUTO SELECT under MOVE TOOL, otherwise, it will drive you crazy.

5.  Good luck.  Having done this manually with several hundred frames for my timelapse without worrying about rotation, this is really going to SUCK! 

[DecepitconFromOrion]

Thanks for feedback ... suppose I'm willing to accept rotation but not translational motion ... does this allow for any extra automation?

 

In case anyone's wondering, I had to travel from Germany to the US for the eclipse and couldn't bring my EQ mount with me  

[Juzwuz]

Have you tried the PIPP (Planetary Imaging PreProcessor) software? I don't think it can handle the rotation but it does a nice job with aligning an "edge in shadow". You can align the partial eclipse images from C1 to C2 with an appropriate selection of the edge in shadow option. Then align the partial eclipse images from C3 to C4 with a different edge in shadow option. Then you would have to manually rotate each image and translate the C3 to C4 images to line up with the other images.

[twinion]

I'm having a similar issue so I figured I'd come with some examples.

 

Here is an example of very good alignment. This is two stacked images showing the difference between them in photoshop. These images were taken very close together, but I did have to do some manual adjustment in photoshop to get them to line up properly. Once aligned, you don't see any corona lines, only outlines of prominences/etc, and only a faint and broken ring of white, all indications that they're lining up well.

 

Here is the same thing, but from much later in the sequence. Again, I manually lined up the images so the circles overlapped as much as possible, but now you can see things like the promineces don't line up perfectly and you can a lot of corona lines as well. I'm not really concerned with promineinces, especially when building a corona shot since they'll be blown out anyways, but I think they're evidence that the alignment isn't perfect. I tried doing some manual rotation on the layers, but it didn't really help. Better aligning one part of image seemed to thrown off another. For example, you can see that the base of the flare (?) at about 2 o'clock is all black in this image, so that point is very well aligned. Meanwhile, the small prominence at the top and the large one around 4 are both slightly off. If I rotate the layer a bit to line one of them up, another feature will be misaligned.

I have to imagine that if there is a good way to line these images up, it would require some kind of manipulation beyond just moving and rotating layers, but I don't really know where to begin. It could be that there really isn't much hope for aligning images from minutes apart on a static tripod.

 

Any thoughts?

[Daniel Dance]

That is a good alignment if you're doing partials.

However, if you're aligning a full stack of images during totality for the intent of processing corona that would be a VERY BAD alignment. 

For corona, again, you want to align using prominences or nearby bright stars as they track almost identically with solar for short durations.

If you align like this using the moon across totality, you're corona will become a blur and you'll lose a lot of detail in it.

[twinion]

That is a good alignment if you're doing partials.

However, if you're aligning a full stack of images during totality for the intent of processing corona that would be a VERY BAD alignment. 

For corona, again, you want to align using prominences or nearby bright stars as they track almost identically with solar for short durations.

If you align like this using the moon across totality, you're corona will become a blur and you'll lose a lot of detail in it.

I assume you're talking about my second image, and I agree it's poorly aligned. The initial tutorials I watched emphasized overlapping the circles, but for corona stacking, should I be aligning them like this?

 

[nimbulan]

I assume you're talking about my second image, and I agree it's poorly aligned. The initial tutorials I watched emphasized overlapping the circles, but for corona stacking, should I be aligning them like this?

I've been wondering about this as well, particularly with regards to prominences.  Is there any way to retain them all or do I just have to pick the beginning or end of totality to base the composite around?

[Starry_Spruce]

I'm no photoshop expert, but what I'm finding is that you have to pick and choose what you want to feature, and/or how realistic you want to be (how much you're willing to cheat). 

 

If you are stacking shots from near the beginning of totality with ones from near the end, there will be pretty significant movement of the moon. I think in that case, you would need to add a single image of the moon back in at the end with masking, and you will need to somehow hide the enlarged black spot the composite created with by having it be white (blown out area of inner corona), or by enlarging the moon (not my personal ethical choice as that messes with scale). 

 

Because the visibility of the prominences changes, again, being realistic dictates choosing one side or the other to feature. Or you can 'cheat' as much as you want with masking, resizes, blending, etc. and add them in all around.

 

In the end, doing an HDR of the corona with shots really far apart in time is not ideal, and some compromise or cheat will have to be made. 

Edited by Starry_Spruce, 16 April 2024 - 09:37 PM.