Back to forums

Which imaging acquisition and processing techniques could I learn more about to increase image quality and accuracy?

Processing Deconvolution One-Shot Color Deep Sky
1 reply•106 views
SarahPi avatar
SarahPi avatar

Hello,

I use a full-frame mirrorless camera, I mostly do daylight photography, some night photography and astrophotography too.

I did hightlight in bold keywords to make it easier to scroll thru, for those that don't want to read most of my post.

Before moving to (and investing) in more advanced software using linear image processing, I would like to read more and learn about all the possible image acquisition and image processing techniques which could potentially be applicable (and beneficial) to daylight imaging to improve image quality.

By improving image quality, I mean it in the most purist way, without actually making things out of thin air, hence I thought which better place than an astrophotography forum to ask, since I imagine people want images as sharp and as close as possible to a scientific truth, with no phantom stars nor halos being added to their pictures nor any generative algorithms.

Here is what I tried so far:

  1. I know about the three types of calibration frames (dark, flat, bias) since I became interested in night photography.

  2. Did try focus stacking in landscape and macro photography, to have the benefits of having a sharper but shallower depth of field, while still having multiple planes sharp.

  3. Did become interested in frame averaging to increase SNR in perfectly still subjects, as well as speckle imaging/shift-and-add to try reduce atmospheric turbulence effects on faraway telephoto pictures of still subjects.

  4. Did only read about MMT (Multiscale Median Transform) as a better way of sharpening algorithm than unsharp mask. Not sure how it compares against deconvolution, in terms of giving sharper yet closer-to-truth images.

Is it possible to get PSF from a specific camera/lens pair by say imaging some tiny LEDs in a dark room, so that I can apply camera/lens pair-specific deconvolution? Or would it not be that useful, since even if using that specific camera/lens pair, PSF would change in different imaging conditions?

Any readings and suggestion are welcome

Thanks in advance

andrea tasselli avatar
andrea tasselli avatar
Let's say: it's complicated. Not sure you can apply deconvolution technique in absence of actual in situ samples of the psf even when a theoretical sample exists. And then I'm sure that the same may not apply to field aberrations unless some sofisticated neural network trained AI is applied and then I'd still suspicious of the results. You can try some psf-less deconvolution techniques with arbitrary weights such as Lucy-Richardson and see what happens.