I think one of the issues with the FLIR One is that it's automatic gain control uses histogram equalization, which is not optimal for detecting objects in the night. Here's a example.
The following is a picture of a dog in a backyard (pulled from arstechnica FLIR One review).
While it's possible to see the dog, it's not very well defined - this is because of the way the FLIR One AGC works. If you look at the temperature bar on the right, you'll see that a large range of the temperature bar is white/light grey. This is because the histogram equalization - the AGC tries to make sure there's the same amount of each shade of grey visible on the screen at once. This means that when the dog (high temp) appears very small on the screen, a very large majority of the image is much lower temperature than the dog - therefore the AGC "boosts" the colour of these parts, such that the lower temperature parts are spread across a wider part of the grey spectrum. This significantly decreases visible contrast between the dog and it's background (even though there is a big temperature contrast), however it increases the details visible in the background. This makes this algorithm much more useful for thermography, and not very useful for detection in the night.
I have now changed the distribution mode to Linear from Histogram Equalization. Immediately you can see that the temperature bar on the right is completely linear and now the dog is very clearly visible (however you do lose some background details). This is because the high thermal contrast between the dog and the background is displayed as such.
Given a smaller target, this effect will be even more profound.
Unfortunately it's only possible to change the AGC mode FLIR Tools or similar after taking the picture - the FLIR One viewfinder doesn't allow for this (might be possible with SDK).
The Therm-App on the other hand allows for a "Night-vision" mode which afaik is similar to FLIR's DDE (as found in the NV monocular products) which allows for the high contrast to be seen while also preserving details in the background.
However, I understand that the Therm-App at $939 might be outside of your price range. Let's hope someone can apply some enhancements to the image via. the FLIR One SDK.
Redshift1340, if you could post the raw FLIR One images of your test, we will be able to check and probably find that with a linear AGC your sister will be much more visible on the image.