I decided to test this, because all I could find was all talk and 'do your own testing' with no examples given. But it is clear some level of preflash is absolutely useless, and then a small change can all of a sudden have a massive impact. So really, you do need to do your own test, but at least it's a ball park figure of what to aim for, and of course, what to look for when it is working.
Essentially.. what you're going to need is a sacrificial test roll. Just 1 will do.
Figure your desired EI speed that you need to shoot at. 6400 is probably a good choice for Portra 400. Meter through your diffusion material, and shoot scenes, no preflash.. -5 stops, -4.5 etc etc. Though it will probably also be -3 to -2 from 6400.
You also need the developer time too to make the most of it.. 6 minutes in C-41 at normal process temp (~38 celsius) is still probably about right.
That's your starting point.
This will also work for your B&W film too btw, I've seen it primarily referenced for B&W in fact, so I decided to try it on colour.
In regards to pre-flashing with no push.. I do not know. But I suspect the preflash should be less -2 stops from middle grey reading, as there is still plenty of straight line scale left.. you probably want to pre-flash it at the bottom of the linear scale. Which is -4 stops from Box speed for Portra 400 (and most others too)
But the zone 3/-2 stops @ 12800 on Superia 800.. it's -2 stops below the published straight line curve according to Fuji.
So, you will need 1 test roll. Pre-flash different amounts and shoot a test scene at the desired underexposure/EI+push/etc, and take notes. The one to use will be when you see a sudden jump in detail like the above example.
If you shoot Portra 400 normally.. you can just save 6 or so exposures at the end of a roll for that purpose etc on your next roll. I imagine it wont be as good as combined with a push.
For example. Let us say your preflash has a value of x.
If one part of the image also has an exposure of x, and another 2x, instead of being 1 stop of separation between these two points if normally recorded.. with a preflash, it's essentially 2x and 3x, or the equivalent of 0.58 stops of separation on the neg.
So the push helps stretch this contrast back out, which can be very important. Although compressed values can also be desirable.
If you DIY C-41, I'd recommend just using 6 minutes @ 38c or so.. no additives needed.