Anders Holck
Well-known member
Well, I would agree that the most visible artifact is the aliasing.
It's pretty obvious that going from 4.9k to 1k, demand a lot from the algorithm in use, and any aliasing that maybe is invisible in a still image will be very visible when it's in a motion sequence.
As these shots were done without an optical low pass filter some amount of color aliasing will even be present in the original. I'll guess that it will even be pretty visible at 4k in the fine hair (left side) when there is no motion.
Also the movie file is still in Adobe RGB (1998) color space, but without a profile attached. I will be so bold to say that it would be better to do a colorspace conversion to sRGB for these motion display purposes, as that, while causing some colors to be out of gamut and clip, would show a more exact replication on the computer screen. Quicktime doesn't seem to do the correct color space conversion, at least on my system.
Here is a sample:
The upper image is the 4k jpg in Adobe RGB (1998) colorspace in a color managed application.
In the lower part, the image on the left is the Bubblegum movie as it is from RED, the one on the right has been color space converted to sRGB while exporting to a new Quicktime file. This can be done in the quicktime export options dialog>effects>Colorsync.
All images are rendered thru my colorsync screen profile, so the overall look might be off. What I'm trying to show is that when converted to sRGB it (pretty much) matches the color managed image.
Regarding the noise, Jim has stated many times that there is no CMOS noise reduction in place on these early images, and the bad scaling and MJPEG compression will make it more pronounced.
It's pretty obvious that going from 4.9k to 1k, demand a lot from the algorithm in use, and any aliasing that maybe is invisible in a still image will be very visible when it's in a motion sequence.
As these shots were done without an optical low pass filter some amount of color aliasing will even be present in the original. I'll guess that it will even be pretty visible at 4k in the fine hair (left side) when there is no motion.
Also the movie file is still in Adobe RGB (1998) color space, but without a profile attached. I will be so bold to say that it would be better to do a colorspace conversion to sRGB for these motion display purposes, as that, while causing some colors to be out of gamut and clip, would show a more exact replication on the computer screen. Quicktime doesn't seem to do the correct color space conversion, at least on my system.
Here is a sample:
The upper image is the 4k jpg in Adobe RGB (1998) colorspace in a color managed application.
In the lower part, the image on the left is the Bubblegum movie as it is from RED, the one on the right has been color space converted to sRGB while exporting to a new Quicktime file. This can be done in the quicktime export options dialog>effects>Colorsync.
All images are rendered thru my colorsync screen profile, so the overall look might be off. What I'm trying to show is that when converted to sRGB it (pretty much) matches the color managed image.
Regarding the noise, Jim has stated many times that there is no CMOS noise reduction in place on these early images, and the bad scaling and MJPEG compression will make it more pronounced.
Last edited: