Web Analytics
Re: Noise reduction in Red’s IPP2

Re: Noise reduction in Red’s IPP2


Art Adams <art.cml.only@...>
 

Yeah, there's something not right with that Helium JPEG. Knowing that the test was shot at 7K is an important clue. Following the 80% rule, that would drop effective resolution to 5.6K. Also, if these were pulled from UHD, as Geoff says:

"These are sections of the full UHD image from each camera grabbed at 1:1 so you can clearly compare colour and resolution..."

Then:

If the Helium was recorded at 7K—higher resolution than UHD, which is what the other cameras can record at natively—then it had to be scaled somehow.

Following the 80% rule, the Helium resolution should have landed around 5.6K, which is beyond the range of a 4K chart to show (unless you frame it to fill 1/4 of the frame).

If the scaling from 7K to UHD went wrong, and resulted in a resolution loss of 50%, then we should expect to see the chart show problems in the range between 2.6K and 3K.

That's exactly what we see in the chart.

It's pretty clear there was some sort of scaling issue in bringing that frame from 7K down to UHD. The camera should have shown higher resolution than any of the others, and instead it shows exactly half of what we should expect, based on the captured resolution. That's why it's important to detail every step of these tests. (It's also why I don't do many of them anymore. It's too freakin' complicated, and a small mistake has large consequences for a product or a manufacturer.) Apparently the raw frame shows the proper resolution.

Also, focus on the Alexa SXT appears soft.

--
Art Adams
Director of Photography
San Francisco Bay Area

Join cml-raw-log-hdr@cml.news to automatically receive all group messages.