The location looks great but the fact that you rendered such a large image. But this is amazing. How long did it take?
The first pass took an hour and a half. Second and third were completed after around twenty minutes or so. From there out there wasn't much quality increase of the actual image. I (stupidly) upped the reference limit to 101 instead of 69.
I have 16 gigabytes installed RAM on my desktop, and Kalles Fraktaler was taking up around 10 gigabytes of memory while rendering the image. My Windows 7 PC was hovering at about 70-something percent memory utilized.
However, for some reason, the memory footprint spiked up every time KF started a new reference pixel, and after 4 or 5 references, stupid Windows 7 Task Manager started limiting the amount of RAM (Private Working Set) KF was allowed to use. As a result, Windows started caching the extra memory utilization to my hard drive, and during said times, the CPU usage dropped to 2-4%, and the constant hard drive chatter commenced. I routinely do .5 Gigapixel renders in Mandel Machine (530MP, typically I use 23040x23040 square aspect or 30720x17280 16:9), so was kind of peeved that KF was thrashing my primary 1.5Gb hard drive. I bought the drive when I upgraded my render machine in 2012, but I try to treat my hard drives nicely so they generally last a long time. I now have a 6Gb internal HD connected for backup as well as storage of KF Fractal Frame data, but my OS still remains on the older drive.
Anyway, the hard drive thrashing would last for about five minutes or so, then it would spend around one minute at 100% CPU utilization rendering the pass, before memory usage jumped up and it spent another 5 minutes thrashing again. All in all it spent around six hours thrashing through all of the reference pixels. I honestly don't know why memory usage ramps up dramatically whenever a reference pixel is computed, but it was a bit unnerving, and I was tempted to abort the stupid process on several occasions. What stopped me from doing so was the render was probably 99% computed.
I get the impression that with some difficult locations, the software picks bad reference pixels and wastes a lot of time rendering the same holes again and again. Being the target was such a light zoom depth (e23), it may have been faster to just use arbitrary precision to render the remainder of the holes within the image.
My first serious render of this area was at a native resolution of 7680x4320. At said resolution, the image was still fairly noisy even when downsampled to 1920x1080p for 4x4AA. This was not acceptable for me so I attempted to up the resolution to .5 gigapixel. I probably won't do any higher than 23040x12960 (~300MP, "4k" at 6x6 AA or "8k" at 3x3 AA) with KF in the future so memory utilization won't be so high that it thrashes my hard drive. I try really hard not to subject my disks to that kind of abuse.
Again, I am used to noise in the ABS fractals, but this area was so noisy that it necessitated huge resolution. My "8k" 7680x4320 source after downsampling with GIMP looked really good, but at 48Mb was too large to upload to DeviantArt. The 4k version is super duper smooth though.
Give me a few minutes to fire up my desktop so I can post the KF file, if anyone wants to explore this area. I plan on picking a zoom path through this location to a mini and rendering fractal frames at 7680x4320 with 1.25 custom zoom ratio. This should be sufficient to produce a clean 1080p zoom movie without rectangular borders, although youtube will ultimately barf on it upon upload.
@Kalles, you may want to increase the precision by a few bits when using extreme skew factors, ie in the millions. If I scroll out just a few zoom levels from this location and the image gets pixellated when it hits floating point precision. 5 million percent is 50 thousand or so times the magnification on the horizontal axis, which amounts to about 16 zoom levels or nearly five orders of magnitude higher than the vertical axis.
EDIT: KFR added to first post!