12 August 2026
Pedrosa del Principe - Carrasquillo wind farm - Turbine 4 - Spain
Local circumstances
Location (on site GPS measurement) : North +42° 13' 59", West 4° 11 43"
Duration of totality : 1mn 46s
Conditions of observation
Equipement :
ZWO FF 80/600 refractor with 75 mm aperture stop (to avoid glare). Regular check of focus up to about 5 min before totality.
Nikon Z7II, 14-bit RAW mode (non destructive compression), electronic shutter mode. Pixel size = 4.35 micron. Image scale = 1.5 arsec/pixel.
Exposure time for partial phase : 1/8000 s to 1/1000 s with Astrosolar ND 3.8.
Exposure times for C2 : 1/4000 s to 1/60 s (7 EV bracketing at 1 EV interval) 7 frames taken in burts mode at 1 s interval.
Exposure times for totality : 1/250 s to 1 s (9 EV bracketing at 1 EV interval, 9 frames taken in burst mode. Remote control = 3 s push trigger + 1 s interval between two successive triggers.
Mount : iOptron HAE29EC. Latitude set with an inclinometer. Azimut set using the Sun shadow when the Sun crossed the meridian.
Sky conditions :
- no cloud, blue sky but with low transparency and heavy obscuration and diffusion due to the Sun low elevation (8°). The solar corona appeared brownish.
- no wind.
The ESF (Edge Spread Function) can be measured directly as the radial profile of intensity accross the lunar limb.
The PSF is the derivative of the ESF. It was calculated on the G channel. The derivative profile was then fitted with a gaussian curve with Fityk software.
The result is an FWHM of 2.6 pixels (or 3.9 arsec), measured on a single shot image taken 15 min before the eclipse, and 2.5 pixels (or 3.75 arsec) on a single shot image taken during totality.
1) Pre-processing
- flat/offset correction,
- hot pixels correction (home-made Julia code).
2) Registration of the image on solar corona
- step 1 : registration of lunar disk on 1 s exposures,
- step 2 : registration of all images on the basis of the Sun and Moon ephemerides.
3) HDR processing
- home made Julia software,
- calculation done in floating 64-bit, output fit file saved in 32-bit integer,
- image weight function = sinus function (except for 1 s, 1/125s and 1/250 s exposures),
- sensor response function is assumed to be linear (see measurement here : ../2024-Mexique/Mexique2024.html ).
4) Display of HDR image
Different algorithms have been evaluated.
Link to the Julia HDR code (old version)
Mid-exposure = mid-totality 18:29:35 UT
Stack of 162 frames : 18 x 9 exposures from 1/250 s to 1s (at 1 EV interval), total exposure = 35 s
Version 1
Link to full-resolution image
Stars are visible to only about mag 8 due to the 8° elevation of the Sun
The basic idea of the processing is to divide the HDR image by a "smooth" mask. The "smooth" mask is calculated using a gaussian convolution with sigma depending on the intensity of the pixel. The brighter the pixel, the smaller the sigma (because brighter details are close to the solar limb and smaller in size).
This is approximated by the following multi-scale approach:
image_visu = HDR_image / [gaussian (HDR_image, sigma (r) + constant]
- r = distance from solar disk center,
- a large value sigma increases the visibility of long streamers and large scale structures away from the solar center (and decreases apparent noise in the resulting image),
- a small value of sigma increase the contrats of small size details in the inner corona, three or four values of sigma are used (typically from 7 to 30 pixels)
- the constant increases/lowers the effect of the gaussian mask.
NB : partial convolution is use to avoid ring effect near the lunar limb (as described in Jonathan Hill video).
https://www.cloudynights.com/forums/topic/1005428-photo-collaboration-for-aug-12-tse/
Link to an older techical presentation about processing the eclipse images (in French)
Multi Gaussian Normalization (MGN) algorithm
M. Druckmuller. Multi-Scale Gaussian Normalization for Solar Image Processing (2014)
Python code from Sunpy/Sunkit lib : https://docs.sunpy.org/projects/sunkit-image/en/latest/generated/gallery/index.html
Julia script
Parameters:
scales = [7, 7 ,7 , 10, 20, 30, 40], k = 0.3, gamma = 4.2, h = 0.92
Wavelet-Optimized Whitening (WOW) algorithm
F. Auchère. Image enhancement with wavelet-optimized whitening. Astronomy & Astrophysics (2022)
Python code from Sunpy/Sunkit lib and F. Auchère's Github.
Julia script
Parameters:
scale number = 5, h = 0.97, gamma = 3.2, noise reduction, whitening
Normalizing Radial Grade Filter (NRGF) algorithm
Original Python code from Sunpy/Sunkit lib.
The code was modified in order to avoid numerical artefacts when the "rings" used for calculation of the radial reach the border of the image. In this part of the mask, the profile of the mask is extrapolated by an Hermite function (two anchors points are set to keep continuity of intensity and slope and avoid numerical artefacts).
Furthermore, the radial intensity mask is filtered with a gaussian (2.5 pixels) to remove high frequency numerical noise.
Julia code with standard NGRF (input image is divided by radial intensity mask and normalized by radial rms mask)
Julia code with simplified algo : input image is only divided by radial intensity mask
Use of radial intensity mask only.
(FNRGF) algorithm
No code was found on the Internet.
The Julia script was made with Claude using published algorithm.
Julia code
The image suffers from very strong circular numerical artefact due, in particular, to the lack of smoothing from one radius for the next.
C2 and C3 Contacts
