Moon zoom - go big or go home?

smsaville wrote:

I'm looking for a new zoom lens to use with my Oly O-MD 10 II for occasional sports photography, but more often lunar photography. I already have an Oly 40-150mm lens but the reach isn't really sufficient for lunar photography so I'm looking for something a bit beefier. I was considering either the Oly 75-300 or the Panny 100-300 lenses, but then I found the Panaleica 100-400. Wow.

Question is: Is the extra 100mm (200mm in full frame terms) the Panaleica gives really worth the x3 price tag, or will one of the "more reasonably priced" lenses sufficient for my needs - and if so, which one? I guess the Oly is a better match system-wise, but I've heard the Panny gives better picture quality?

Get a real telescope. Much better suited to lunar photography.  A small refractor or a maksutov are ideal.

100mm extra makes a big difference on the Moon. You can see the Straight Wall (shallow canyon upper part of photo) more clearly in the 100-400mm shot than the 300mm shot. If you take dozens of of shots of the moon, chances are one will show more detail owing to a brief steadying of the atmosphere during that moment. This is critical the longer the focal length and the wider the opening of the front lens as the longer/wider they are, the more susceptible they are to atmospheric irregularities. Those two shots you did are very good for the focal lengths you used.  Here is a shot through a "crude" lens (telescope) with only two elements but during good seeing and at 1000mm though a m4/3rds camera.

BruceRH wrote:

Oh, I know I am limited with what I have, I wanted to let the OP see what can be expected from the PL 100-400. A telescope would be better and cheaper than the PL. I enjoy moon images and at some point I'll get a refractor but for now, I use what I have.

The main benefit of a camera lens (most of them) is portability, even a small telescope needs decent tripod or mounting so it's much more limited unless you are really dedicated, or do most of your shooting from your backyard or balcony.

Barty L wrote:

smsaville wrote:

I'm looking for a new zoom lens to use with my Oly O-MD 10 II for occasional sports photography, but more often lunar photography. I already have an Oly 40-150mm lens but the reach isn't really sufficient for lunar photography so I'm looking for something a bit beefier. I was considering either the Oly 75-300 or the Panny 100-300 lenses, but then I found the Panaleica 100-400. Wow.

Question is: Is the extra 100mm (200mm in full frame terms) the Panaleica gives really worth the x3 price tag, or will one of the "more reasonably priced" lenses sufficient for my needs - and if so, which one? I guess the Oly is a better match system-wise, but I've heard the Panny gives better picture quality?

300mm will give you 3.3° horizontal FOV (2.5° vertical). 400mm will give you a 2.5° horizontal FOV (1.9° vertical FOV).

The moon has an angular size of approximately 0.5°. Assuming a landscape camera orientation, in the case of a 300mm lens the moon will occupy 1/5 of the vertical FOV, or a bit more than 1/4 vertical FOV at 400mm.

Assuming you want to place maximum pixels on the subject, you need to go all the way to 1200mm before the moon will almost fill the frame vertically.

I don't own one, but I read from those that do that the 100-400 is a very nice lens, and I've seen one or two nice images of the moon taken with it - heavy crops. If your principal interest is in moon shots, you might be better directed to start thinking about adapted telescopes.

Filling the field, top of the frame to the bottom (the annoying 3:2 sensor really isn't ideal for shooting round objects) in a 35mm camera requires a 2000mm lens or telescope.  It's an ideal fit at that format.

framedinspirations wrote:

Serguei Palto wrote:

...

This is the average of 3 shots after resampling by 200%, so the effective resolution of the sensor is enhanced.

Can you describe your software/process whatever for how you "averaged" 3 shots? Thanks!

I have used Adobe PS v.8.1.

The general process includes the following steps (I have used the Russian version of the software, so my translation of the menu items can be not accurate).

1. Loading the group of the RAW files as a stack (File->scenarios (script)->load into stack).

2. Crop all the images to the same useful size if necessary (select all the file items in the stack and apply the crop) .

3. Resize all the images by 200%. (image ->image size->200%)

At this step you increase the quantity of the virtual pixels by a factor of 4. PS makes interpolation for each  of the images. You can use different interpolation methods . If you have more than 5 images than the simplest pixel interpolation should be used. In case of small number  (2 - 5) of the images the "Auto" method is better.

Interpolation does not enhance the resolution. But it creates a container for the new information which exists in the different images. The real enhancement of the sensor resolution will appear later - after the averaging.

To do the averaging you continue as follows:

4. Align all the resized images (Edit->Automatically align layers->Auto->OK).

5. Create the smart object from all the aligned images (Layers->Smart Object->Convert to Smart Object)

6. Make  averaging (Layers->Smart Object->Stack Mode->Averaging.

7. Apply sharpening if necessary. Typically if we average many images then the dynamic range is significantly increased, so a strong sharpening can be applied without increasing the noise.

8. Save as JPEG.

To get the best results the following conditions must be satisfied:

1. A sensor should be without AA-filter.

2. A lens should be very good (diffraction limited). And the apertures larger than f/8 are highly recommended if the pixel size ~4um or less (the f/8 as in my case is not the best , but  I had no choice; the Pana 100-300 at the long end and at f/5.6 is not the diffraction limited lens;  my copy is sharperst at f/8).

3. In all the images an object should be projected onto different parts of the sensor (my shots are done from hands) .

Below is a testing example. The first image is the single shot. The second image is the high resolution image.

Click "View: Original size" to see the detailes.

Image#1. Single shot

Image #2. The high resolution image. The average of 8 shots after resizing to 64Mpx.

Serguei Palto wrote:

7. Apply sharpening if necessary. Typically if we average many images then the dynamic range is significantly increased, so a strong sharpening can be applied without increasing the noise.

In the context of lunar photography, because noise is random, stacking multiple noisy images cancels the noise out, so you are able to use as much sharpening as you want - whether to make feature more obvious, or to a lesser extent just make the image look more appealing. It is hard to imagine the effectiveness of this technique, the results can be remarkable for noise reduction alone.

For example, shooting a total lunar eclipse is almost pitch black, very dark indeed, and noise is a real problem.

8. Save as JPEG.

To get the best results the following conditions must be satisfied:

1. A sensor should be without AA-filter.

2. A lens should be very good (diffraction limited). And the apertures larger than f/8 are highly recommended if the pixel size ~4um or less (the f/8 as in my case is not the best , but I had no choice; the Pana 100-300 at the long end and at f/5.6 is not the diffraction limited lens; my copy is sharperst at f/8).

3. In all the images an object should be projected onto different parts of the sensor (my shots are done from hands) .

Again, in the context of lunar photography, you can hold the camera rock steady (and you should!) because one of the errors you are trying to average is atmospheric disturbance, which by its nature will cause the image to be projected on different parts of the sensor.

The other factor is the Earth's rotation (moon's apparent movement) will will do the same, and this can be the limiting factor in getting enough images for stacking - the moon moves out of shot quickly. You can't fix that by re-adjusting the tripod head, as that introduces rotation into the shot, which will have to be corrected for.

Below is a testing example. The first image is the single shot. The second image is the high resolution image.

Click "View: Original size" to see the detailes.

Image#1. Single shot

Image #2. The high resolution image. The average of 8 shots after resizing to 64Mpx.

Good explanation of your technique, thanks.

For lunar photography it is probably easier to use PiPP and AutoStakkert!2 which will handle many hundreds of images, including RAW.

Just cause this post got me curious, I figured I would try it out myself.  This was taken with Oly EM1.2 and a Canon FD 300mm f4L lens with a 1.4x extender, on a tripod on March 10.  Stacked 16 out of 20 shots in Registax 6.  So I would think any decent 300-400mm lens could get similar results and legacy lenses in this range are not all that expensive.  Sure I can do better as my learning curve improves (mostly with Registax), but this entire process took about 30-45 minutes.  Once I learn how to use the electronic shutter at 60 fps and a way to fire the camera remotely (I just got the camera on Wed, on about page 45 of the manual!), I will try again and see how it looks with 100+ images stacked, assuming Registax doesn't blow up with that much data, not sure it was designed for large rez still stacking.

mjc1 wrote:

Just cause this post got me curious, I figured I would try it out myself. This was taken with Oly EM1.2 and a Canon FD 300mm f4L lens with a 1.4x extender, on a tripod on March 10. Stacked 16 out of 20 shots in Registax 6. So I would think any decent 300-400mm lens could get similar results and legacy lenses in this range are not all that expensive. Sure I can do better as my learning curve improves (mostly with Registax), but this entire process took about 30-45 minutes. Once I learn how to use the electronic shutter at 60 fps and a way to fire the camera remotely (I just got the camera on Wed, on about page 45 of the manual!), I will try again and see how it looks with 100+ images stacked, assuming Registax doesn't blow up with that much data, not sure it was designed for large rez still stacking.

I use timelapse mode for the capture, and (depending on the focal length) usually get enough images to stack - there are diminishing returns in terms of IQ when using large numbers of images.

The sample you show seems a bit over-sharpened to my eye, is that straight out of registax?

I screwed about a bit in PS as well, not much.  It is oversharpened for sure, not an expert with Registax so have to learn more about the wavelets and other aspects of its processing.  Today was not the day for that, this was pretty much just a fast and dirty test to play with a new camera with no intentions of keeping anything and Registax was crashing enough that my enjoyment factor in the project was rapidly declining!  I did run a single image through Registax with similar settings and the stacked image had a decent amount of better detail, but the bigger difference was that the stacked image had far less noise, so that was a good learning lesson for me (I mean its not a surprise and to be expected, but nice to see the actual difference with your own eyes).  That and apparently Registax doesn't like to run an 8-core processor and only worked when I dropped it down to utilize 2 cores.

Sounds like you need to try PiPP and AutoStakkert! - a little esoteric for some but I've found it utterly stable and gives good results (and both are free).

Will check them out, thanks.  Mostly just need more experience with whatever software I pick, just not something I do everyday.