2024 eclipse

April 08, 2024

Lions suspect what began as flexor digitorium longus tendinitis is going to end up as plantar fasciitis. The symptoms tend to migrate around the paw as different areas are stressed to avoid other areas. The symptoms migrated over 3 weeks to the bottom, near the planta fascia. It might actually have just spread to a larger area with the original tendon still painful but no longer swollen.

--------------------------------------------------------------------------------------------------

Being the last total eclipse pass of the lion kingdom's lifetime & manely clouded over by a perfectly aligned cold front, it was a good time to ponder the last one after 7 years.

Lions manely can't believe they once braved the chaos of traveling to a total eclipse. It was something only crazy animals did & only documented in pictures of the 1970's. Today, the animals racing around the country look crazy. 1 reason might be a lot more money to be made from social media than even in 2017. The twitter & tube stars definitely milked it for all it's worth, this time.

The army of cameras only yielded 1 unique & decent shot. Hard to believe only 1 good shot came out of that. The smoke is clearly visible near the horizon but the sun was high enough for the brightest parts to get through. The smoke completely hid the stars. 1 star was visible in real life but not seen on any cameras. It might have been venus.

Lions still remember how much longer the corona tails were than in photos. The tails stretched many sun diameters away & changed shape over time. They started thin & got fatter over time. They were a solid, velvet white.

Then the sun instantly became blinding again. It wasn't a gradual return of light but a switch.

The lion kingdom's timelapse was roughly what everyone else got. It may not seem very valuable, but there's always the knowledge of it being the one lions got. As soon as totality was over, everyone else packed up & tried to leave. Only lions continued shooting until the very last contact.

For the totality closeups, it was an APS-C sensor + 200mm ISO 200, F5.6, 1/200s. Totality tracked across 1/4 of the frame. With a 2x teleconverter, it would track across 1/2 the frame. There was no exposure bracketing in the closeup. A faster camera might have been able to bracket exposures but the lion wanted to make a video.

After it ended, no-one cared anymore. It was on to the next news cycle.

The scene today.

https://nationaleclipse.com/maps/map_08122045.html

The lion kingdom will be retired & probably living in the path of the next total eclipse. Clear weather is unlikely though. Given the convective type of weather in the south, it might actually clear up because of the reduced sunlight.

It will be dark, no matter what. Who knows what amount of vision lions will have left.

2017 was the last time lions would have that much gear for an eclipse. It might be worth renting a van for another one & spending a week driving. Flying doesn't allow packing that much gear.

------------------------------------------------------------------------------------

Upon closer inspection, the 50mm did manage to detect several stars in just the right exposure brackets.

Noticed how the brightest area in Cinelerra is clipping more than gqview/geeqie, making the stars disappear. This is a bug. The clipping is visible in the bayer pattern.

Fixing the histogram to show bins above the red lines revealed a lot of data being clamped to the 100% bin. A lot is being clipped somewhere before the bayer pattern.

Dcraw has a raw_image intermediate which goes up to 15464 or not quite the maximum of 14 bits.

LibRaw generates the same amount of clipping as the original dcraw.

Running dcraw -4 on the command line generates either the same image as Cinelerra or a blown out image if auto brightness is enabled, so the magic is in geeqie.

https://www.dechifro.org/dcraw/

Reviewing the dcraw source code, lions noted the author named his license plate after it.

He must have been pretty proud of it & gained a lot of fame before moving on after 2018.

Then there were at least 2 problems, the color curve & the bayer interpolation no longer works in OpenGL. Given the lack of use of the bayer interpolator over the last 20 years of DSLR's & the continual manetenance required, the decision was made to abandon the openGL support.

It has been noted interpolating & white balancing in libraw takes out information because it clips everything at 65535. On the other paw, most of the information you're trying to recover with raw mode is going to be on the bottom end.

But it was also noted Cinelerra's basic interpolation has checkerboard artifacts while drcraw's interpolation has sharper edges. It really needs a complete port from dcraw to get floating point support & because that uses a FIR filter, it would have to be CPU only.

The aggregation support became a nightmare. Suspect none of the aggregated pathways still work. Everything that uses it needs to specify the title of the plugins. There have been too many optimizations to the opengl support.

The number of cases where aggregating pixel computations is going to meaningfully improve performance & is going to be affordable to implement is so rare these days, it's probably getting dropped. That was only used for raw camera photos which are so slow to load, the improvement would never be noticed.

That was implemented for colorbalance, gamma, interpolate bayer, color3way, framefield, histogram, rgb601. It's quite entrenched.

Experimenting with Gimp, it seems Cinelerra needs a color curve plugin in addition to the histogram. It probably needs to reuse a lot of histogram code. That might be how geeqie is filling in the clipped region if it's not applying curves from the camera.