by Marko Riikonen and Petri Martikainen
One of us, Martikainen, was passing on May 25 though Joensuu where the other one, Riikonen, was staying, and we had decided to do a stacking comparison with our cameras weather allowing. It went swimmingly. There was high cloud from morning to the sun set and when we rendezvoused in the evening on a grassy field next to lake Pyhäselkä to set up tripods, cameras and blockers for the test, the display reached its peak.
Martikainen was photographing with Nikon D7000, Riikonen with Canon 80D. We both had Samyang 8 mm fisheye, though not the same model. Our blockers differed a bit. Martikainen had the blocker on a separate tripod, whereas Riikonen's was attached to the camera hot shoe.
We set identical iso, shutter speed and aperture. Then we took 136 photos every 5 seconds, at around 1/3 of the of way there was a little longer simultaneous break.
In postprocessing we harmonized the color temp and tint of raw's before turning them into 16 bit tiffs for stacking with Halostack in full size. As the 80D images are larger than D7000, the former's final stack was downsized to match the latter before we gave them identical application of the usual methods of usm, br and bgr.
Downsizing the 80D stack before the enhancements could make it lose some of its higher resolution advantage, so we made a test. Usm with a radius higher relative to the two cameras image size was applied to the 80D stack before downsizing and this was compared to the version where the 80D stack was first downsized and then usmed (with the smaller radius). There was zippo difference. So it seems ok to downsize the 80D images to D7000 scale before enhancements
The test showed D7000 as the better halo camera, something that Riikonen had suspected all along. We do not know how to characterize in proper terms the differences seen in the photo comparisons, but it seems like 80D has more problems with color noise.
Terminology:
usm = unsharp mask
Was this at the base ISO? I'd like to see such a comparison at, say, ISO 320.
ReplyDelete