This type of test has questionable relevance to most real-world TC use, as the target is twice as far from the camera in the 1000mm version as the 500mm version. These tests were definitely not set up to aid in assessing equipment for focal-length-limited photography. In the real world, you usually put a TC on because you can't get close enough, and would otherwise have to crop; you usually don't step twice as far away because someone forced the TC into your system. One should expect a slightly softer render at 1000mm, but not so much softer that any item in the center of the target wouldn't be recorded better overall if both focal lengths were shot at the same distance from the target, which is more representative of most real-world TC use. BTW, CA can be corrected better than what we see in these tests, and better-sampled versions of the same CA can be better corrected. Some of what looks like luminance blur in the repeating lines is actually CA that is mostly cancelling its colors out by adding fringes from both sides, re-combining the colors. IOW, correcting the CA in the 1000mm version would not only remove the color, but also improve the contrast in the repeating lines.
The proper test for what really matters for much TC use would be to set the distance to that which frames the target at 1000mm, and then shoot 700mm and 500mm from that same distance, and crop the target and upsample them to 45MP, and also do CA correction on them all. If we want to extend the test to consider noise, using low light, then the same shutter speed could be used for all, too.
The owls have very fine detail in their feathers, and the IQ is a bit "mushy" in my opinion. The lab test for the 1.4x is better:https://www.the-digital-picture.com …omp=0&FLIComp=5&APIComp=0
In my experience so far, the 2x might have a future on my R6 but probably not on the R5. More shooting to follow over the summer.
The R5 with a 1.4x is similar to the R6 with a 2x, in pixels-on-subject, from the same distance (a slight advantage to the R5 of 1.125x pixels or 1.06x in each dimension). Same for no TC on the R5, and a 1.4x on the R6. So, the R5 helps with TC avoidance, by getting the pixels-on-subject in an optically better way.