Better photos of the first casts
I am completely aware that above photos look underwhelming and seem to contradict what I’m writing in the text. This is because photographing shiny metal is pretty hard. I therefore decided to prime and quickly paint one of the good casts (the anatomically incorrect arm) to better show the actual level of detail.
Casting a big hat
The reason I did that now is that I wanted to convince myself as well that the level of detail is sufficient for the next steps. I decided to deviate from my plan to cast a the big hat (hat) next and do a proper big hat (dwarf) instead. This is somewhat risky as the sculpt has some undercuts and, of course, is the result of many hours of work. But I wanted a small motivational boost which is why I decided to risk it anyway.
With the same procedure describe above, I created a nice little mould. But I wanted to have more control over where the mould line goes to minimise the amount of tricky undercuts. So I decided insert the dwarf at an angle into the bottom half and formed small ‘hills’ and ‘valleys’ with a sculpting tool in the uncured silicone. Then I proceeded with the second half as before. Similarly to my first venture, the model submerged a bit too much into the bottom half. This is due to the pressure I exerted onto the figure when adding the top half causing it to sink in even more. Apparently, the art here is to insert the model just above where you want to have the actual mould line. I guess, this is where years of practise come in. Below you see the finished (and used) mould with the vents cut in.
Once again, the vulcanization worked as expected. The original sculpt was a mix of Procreate and Green Stuff. Both survived pretty well. The Green Stuff only stained the Procreate a bit but I cannot detect any loss of detail.
So, onto the casts. I am extremely happy with the results, especially with respect to the level of detail. While lacking behind some of the high-end commercial casts in the detail department, I would not say that they are significantly inferior. They are arguably better than many plastic miniatures I have seen. The details are a bit soft at difficult locations (generally: sharp edges) but the better casts very nicely capture even fine details such as the fine texture on curly hair at the back of the head, the moustache and the factial features. And this is as fine as my sculpting usually goes.
What surprised me was how well the mould dealt with undercuts (and survived). I could not notice any loss of detail near the mould lines. This mould also featured very thin silicone at some locations. Even there, after about 50-60 casts I could not notice any degradation or loss of detail. Overall, I got around a dozen good ones I will use, about as many ok ones (I will keep them for now in case the mould breaks unexpectedly and I am unable to use the green stuff master a second time but I would remelt them if I have enough copies / I am short of material) and the rest experimentation and miscasts. Mind that a lot of this session was dedicated to finding the optimal temperature / casting technique. Towards the end, I would say, around 50% were keepers.
After this, I will probably make moulds of the remaining chaos dwarfs and my hobgoblin sculpts in the coming weeks and months. This may take some time as both mould making and casting takes time. While mould making is mostly passive, the latter is only really worth it if you dedicate a few hours at once to it.
So, here are the pictures. As before, the shiny nature of the metal does not do them full justice, so I will have to paint them at some point. 
If you are interested in some technical findings I made, you may continue reading after the pictures.
Technical stuff:
I wanted to be a bit more methodical with the casting this time. Potential causes for my issues where (i) mould too cold / warm, (ii) temperature too high / low, (iii) too much / not enough talcum and (iv) tapping against the mould after the metal was cast or lack thereof. The last thing is something I have encountered repeatedly on-line (Prince August for instance suggests this in one of their videos) for helping with tricky casts but I have also found people strongly advising against it.
Last time, I used 350-360 degrees (mind that the optimal temperature depends on the alloy at hand!). I began to dial the temperature control to the same setting and started powdering my moulds more liberally. I quickly realised that this was not having the desired effect as the talcum started clogging up the details. The best results were achieved by using a make-up brush and brushing off any excess talcum such that only a thin coat remained on the mould halves (the pictures above show way too much talcum on the left, the right picture looks about right).
On to the temperature. I started increasing it to first 370 and eventually to 400 degrees C, worrying a bit that my mould would degrade (it’s rated for 350 degrees C) but I could not notice any problems. Contrary to what I was expecting, the level of detail decreased significantly around 400. Interestingly, I also did a few tests casts with the weapon arm mould and around 400 C the axe got cast completely and pretty reliably at that (I did not, however, repeat this experiment at a lower temperature). So the optimal casting temperature likely depends on what you want to cast. Thin, long objects requiring the metal to stay molten for longer may require a higher temperature than something bulky where you can optimise for detail. Next, I decreased the temperature to the range 300-330). I got the best results with the chaos dwarf sculpt at around 320.
Tapping against the mould while or after the metal is cast did not have a beneficial effect. I would even argue that it slightly degraded the results, perhaps due to the metal pushing out of the details again. What did have an effect was to increase the height at which I poured the metal. That way, the metal accelerated more and was thus able to displace air inside the mould more quickly resulting in finer details.
The temperature of the mould also seems to have an influence. The first few casts are usually inferior. With a model as large as a chaos dwarf, the mould became quickly warm (as opposed to the weapon arm mould). After the warming-up period, I could not notice any degradation due to the mould becoming too hot.