[Warner306]

@ashlar
Nice explanation of the target refresh rate! Please correct the typo "Target refresh rate = 23.97602 * 0.9999758 = 23.79544" in order to not confuse the people even more
Would be cool if you could explain in such a short compact practical oriented manner the second part of the story concerning the custom refresh rates (I´m not very good in explaining things SHORT

Unfortunately, as far as coming up with the correct (deviation compensating) refresh is concerned... there's A LOT of trial and error in my experience. Luckily CRU now includes the restart driver option which previously lacked (if I had to reboot for every check I made... I would have fallen asleep halfway through). I'm pretty sure that the trial and error stems from my lack of complete understanding of all the variables involved. I might study in the future, but I don't feel qualified to write an how to about it.

ALL THAT FOLLOWS IS WORTH A TRY AND IT SURELY GIVES A CLEARER IDEA OF WHAT GOES ON "BEHIND SCENES". I'VE HAD MIXED RESULTS WITH IT, SO FAR.

In any case, in my explorations of the web, I think that a wise combination of these three tools could help in coming up with resolution timings to try with CRU. Combining the three tools it's probably possible to limit the number of trials and errors, as DTD Calculator provides refresh rate results with many more decimal points, up to a precision that satisfies our needs:

Pixel Clock Calculator
Custom Resolution Utility (CRU)
DTD Calculator (this is meant for Intel GPUs but timings are the same for all GPUs, it's the implementation that changes)

With Pixel Clock Calculator you can come up with a bunch of settings to try in CRU.
In CRU you proceed to add a resolution, you set the refresh rate up to the third decimal (the maximum precision CRU offers) and then you try inputting the different totals that Pixel Clock Calculator offers. You are going to act on the Total and Refresh rate entries.
Some of them will not work (CRU will show the numbers you input in red, it has to do with the minimum values accepted, by playing with the Front porch and Sync width values you can solve it). Some of them will be accepted by CRU but will end up producing undesirable results (portions of the imag distorted or shifted with last pixels repeated till screen end, you'll see what I mean when and if you see them).
When you find one that works without producing a bad image and that is close enough to the refresh rate you desire, you can switch to DTD Calculator to have it calculate the refresh rate with more precision (to save the trials and errors).

But how to translate what you see in CRU to what you see in DTD Calculator?
Start off by inputting the Active entries (horizontal and vertical, 1920x1080 tipically) from CRU under the two entries "Active Pixels" (horizontal and vertical) in DTD.
You then sum the Front Porch value (from CRU) to active pixels to get the "Start of Sync Pulse" value for DTD.
Then you sum Sync Width (from CRU) to "Start of Synch Pulse" in DTD and you get the value for "End of Sync Pulse".
Then you sum Back Porch (from CRU) to "End of Sync Pulse" in DTD and you get the value for "End of Sync Pulse".
Then you input the Total value (from CRU) to "End of Blanking Interval" in DTD.
All of the above must be repeated for both horizontal and vertical values, clearly.
Lastly, take the Pixel clock value from CRU and input it in the same entry in DTD.

You will see that DTD in the bottom right corner provides the resulting refresh rate with much more precision than CRU (and if you hover on the value, with your mouse, it shows even more decimals).

You iterate the process from CRU to DTD until you come up with the closest approximation of the refresh rate you need. Then you use that in CRU and test with madVR. Hopefully it will provide you with what you need. As mentioned previously in the thread, the calculations are exact but the different graphic cards might require some offsets, which brings us back to more trial and error. With all these instruments... let's say we could be flying a little less blindly.

With all of the above I've managed to improve my 23.976 to the point where it oscillates between 6.x days and no frame-drops/repeats expected. I have no frame-drops/repeats expected for 50Hz as well and one frame drop between 3.5 and 4.2 hours at 59.94Hz. Color me impressed hannes69!

I linked to this post in the DSPlayer guide. Clean it up, add to it, or edit it. It doesn't matter. Just keep it in that post for others to read.