You did it again with this splendid image!
Its remarkable in that details are preserved on the bright limb as well as the terminator. And there are so many of those details. Its just beautiful!
Are you capturing in > 8 bit mode? The reason I ask is I found this was needed in order to capture simultaneous details in both disk and prominences in solar imaging.
Thanks for the nice comments, Glenn. I have one more image to process to complete a trilogy spanning April 12-14, but time has been in limited supply. Actually, this image was taken in 8 bit mode. I do have a number of thoughts on this matter as it relates to lunar imaging, based upon many hundreds of hours of imaging and processing in the past few years. In a frame for frame matchup, say by stacking an equal number of 8 bit and 12 bit frames (most cameras are limited at 12 bits, some at 14, when capturing in "16" bit mode), the 16 bit files always allow you to push the shadow detail more in lunar images. No doubt about it. But in anything less than very good seeing, the differences are limited and manageable.
Emil made a post a number of years ago, long before I joined CN, and I see that you have commented on that thread. I have always enjoyed reading that post as a reference about how noise affects bit depth, although it is really more focused on planetary imaging, where the dynamic range is very limited to begin with. With the accessibility of larger frame cameras with small pixels, that can achieve high resolution on the Moon while spanning a huge dynamic range, this topic should really be revisited. I have done some tests with data that I have captured in 12 bits and then downsampled to 8 bits using PIPP, and then run side by side comparisons. I have never posted the results because, quite frankly, it takes a lot of time to get a post like that together, but it is very clear that shadow detail can be pushed further with less noise in the 12 bit stacks than with 8, even when stacking 1000 frames. I believe the only legitimate way to test this is to downsample the 12 bit file, because otherwise you cannot control for the seeing variability from capture to capture. Perhaps eventually I will post those results. But the basic summary is that there is more quantization error at the very edge of the terminator in the 8 bit stack compared to the 12, which necessitates more moderation in processing for the 8 bit file. However, the real world impact is not that great, because there is generally no reason to push the shadows to the extreme in every instance.
That being said, I have found that when capturing lunar images in 8 bits, if you stack 1000 frames, you get a very acceptable result even when pushing shadow detail. If file size and frame rates were not issues, I would always capture in 12+ bits for the Moon. But in the case of the image presented in this post, I could tell immediately upon looking at the live view window that the seeing was relatively poor, and I was going to need to be very selective with my frames. So I captured 5000 frames per file, and stacked 1000. I initially thought I might have to stack even less, but the 1000 frame stack looked pretty good. In situations like this, any benefit in dynamic range that a 12 bit capture may offer is negated by the inability to capture as many frames to combat the seeing. Each 5000 frame file here is already 100GB in size, and this would become 200GB in 12 bits, at a slower frame rate. When seeing is better, I do like to capture in 12 bits, although recently I haven't had those conditions. Hard drive space and processing also become limiting technical issues. Huge files are a pain in many respects.
My thoughts about lunar processing have evolved over the past several years. I'd like to think I keep improving, and looking back at some old images, I think this is the case. In general, my goal is to achieve something that looks like a natural view of the Moon. In many respects, the Moon is an easy standard, because we can all look up and see what it looks like. Yet one would be surprised how many lunar images simply don't look like the Moon. Oversharpening is one common problem, but even if we overlook that issue, the main difficulty with lunar images is creating a balanced image that matches the tonality that we see by eye. The raw images that come from the camera generally do not faithfully capture the tonal distribution that we see by eye. Achieving this typically requires many subtle adjustments to the tone curve in Photoshop, completely independent of any sharpening. In my opinion, the resolution of a lunar image is almost irrelevant, because even low resolution images can be displayed at an appropriate scale so as to look suitable. However, an unnatural tonal distribution is immediately apparent, no matter the image scale, because we all know what the Moon looks like.
One final element to this puzzle is the variability in displays, coupled with the adaptation of the human eye. I found this out in many of my earlier postings, in which I would look back at an image the next day after posting and see that it looks dark. I discovered that I frequently process images and post at night (this hobby is nocturnal after all, but also I have my free time at night). At night, with low level interior lighting and adapted vision, what looks good on the monitor will often look very dark when viewed in daylight conditions. I wasn't paying enough attention to the final histogram. I always process my final images by eye, and purely subjective assessment, rather than by any formula, but I do use the histogram to guide some of the decisions. I try to find a region of the Moon that should probably register as "middle gray", and make sure this is reflected in the histogram by spot checking the region in Photoshop. This is all completely subjective, as is the concept of middle gray itself, but it is worth noting than an 18% gray card, as pioneered by Ansel Adams as representing a neutral gray, registers at about a tone value of 124 (on a 0-255 scale) in sRGB color space (although I have seen values as low as 117 reported). This can be a useful reference when spot checking an image. In the case of the image presented in this post, I increased the highlights at the very last minute because I felt like my penultimate version was coming out somewhat muted along the limb, and not reflective of what the Moon looks like. But ultimately, a lot of this is purely subjective.