Pseudo-flats and Full-Disk Solar Images

06/07/2024. Things just go along for weeks, months, or years, maybe with incremental improvements and sometimes without even that. Sometimes, as time goes along you just get clearer about what you need, and eventually you understand well enough what you cannot do that when you run across the right tool or the right method, you know to say, there it is!, and things change. The last few days of solar imaging have been like that.

Yeah. Click it. It won't be much bigger, but you may see it better. Details below.

Following my come-to-Jesus moment about excessive image scale, a 2x2 binned workflow has cost me nothing in actual resolution but saved huge blocks of time and storage, encouraging lots of experiments that have paid off made way for three key insights this week.

 

The first thing is the pseudo-flat. I've done something similar before, but "similar" does not do what this method does. I can see from testing against old and new data that detail and tones are the best I've ever achieved. As a bonus, this workflow enables routine, full-disk imaging with the TMB92SS / Quark / ASI178MM. Finally! More about that down below.

Here's an image from today of the flare factory sunspot group (it will always be "3664" to me) exiting the stage for the second time. This is not the most impressive example of pseudo-flat rejuvenation I've seen (I'll select some examples shortly), but it is timely and will do for a start:

 

Yes, you should click it. Else you will not know what I am on about.

The pseudo-flat workflow was described some time ago on CloudyNights by "AstroTruck." Here's my account of the method. I implemented the first seven steps as a Photoshop Action:

1. Make a copy of the original frame.
2. Paste it over the original; keep attention on the copied layer.
3. Execute Filter / Noise / Median using a large value for Median (50+)
4. Do Curves / Blend with factory preset darken.
   
5. Repeat step (4) again.
6. Perform step (4) a third time.
7. Blend layer, mode difference.
   [end of action]
8. Adjust opacity of the layer (60-100%) to moderate the effect as needed.
9. Flatten the image.
10. Continue with curves or levels, sharpening, etc.

SmartSharpen often works wonders with the pseudo-flatted files, so much so that I often elect not to use ImPPG and more often have to consult "history" to see if I have.

 

Second, full-disk solar imaging: the example up above is a mosaic of data taken this morning. It uses the best 50 frames from four, 500-frame clips, acquired using full bit-depth, 2x2 binning, no gamma, and 800 microsecond exposures. A simple Antares 0.5x achromatic reducer mounted inside a 1.25-inch snout and spaced 41mm in front of the sensor produced a decently reduced image scale. Since the Sun does not fill the frame, I used a double-layer of thin translucent plastic to prepare an in-camera flatfield. I applied the pseudo-flat process defined above to each finished subframe. Aligning and blending the frames was still tedious and not especially successful. I spent an hour or so getting a barely presentable full-disk image.

While walking Gemma, I pondered keywords to use in searching for specialized software to assist un assembling multi-row solar mosaics. Somewhere along the trail, I remembered the "Photomerge" process already built in to Photoshop (File / Automate / Photomerge). Without expectation, I tried it out on four identically processed (but not yet pseudo-flatted) frames. The result was not bad at all -- mostly (but not entirely) well aligned and with invisible seams. Then I ran the pseudo-flat action against the four component files and tried again. The new result was excellent! The full-disk image up above has been tweaked (cropped, sharpened, histogram adjusted, resized) for presentation. The order in which the files are presented to Photomerge determines (or at least influences) the Z-depth of particular details, so some nimble naming may be required to, for example, hide featureless spaces that sometimes plague my widefield solar images. Those may result from poor in-camera flat-fielding or from simple unevenness of the etalon. Mess with that some.

 

06/08/2024. A second try at full-disk imaging began with very carefully centering the Sun prior to placing a double-layer of diffusing plastic film loosely over the objective. That produced better flats for the four required component frames. The first attempt to Photomerge them failed when I cropped too tightly and didn't leave enough extra data for the algorithm to seize. A second attempt using just the raw frames, no pseudo-flatting, cropping, sharpening -- no prep work at all -- worked much better. I do need tp experiment more to find out how to control the Z-order in which images are layered. More show and tell when I am just a little more certain of these processes. (Yes, this seems to work: stick prefixes in front of file names to determine order of assembly; put the more detailed frames on top of the stack, or at least make sure that a blurry top image is not the sole contributor to any region. If you keep the original stack prior to further manipulation, you can pretty straightforwardly layer in whatever better data might be available. The pseudo-flat action will correct a world of uneven lighting.)

The third thing, toward a digital aesthetic. This needs its own page. Or more. Anyway, let's start something fresh and see how far it goes.

 

:: top ::