It’s normal that the results of these three MTF analysis are slightly different. They have different assumptions and the calculations are executed differently.

In this article, you can find these information.

Methods for analyzing MTF in OpticStudio – Knowledgebase (zemax.com)

Normally, the Huygens PSF is the one to trust but here I want to discuss about the cases that Huygens MTF may not be accurate if we use the default Huygens PSF settings.

Like the one mentioned above, the system is not well focused in the image plane.

We can tell from the spot size that this system is highly aberrated and not limited by the diffraction.

The Huygens MTF shall not have such a high value.

So let’s check the Huygens PSF to see the source data used by Huygens MTF.

You can see the Huygens calculated by default considers a small area. It’s calculated like below:

Let’s try to set the delta image so that the Huygens PSF’s image  is roughly the same size as spot size.

We can see the Huygens PSF is still unlike the result obtained in the Spot diagram.

This is not what we expect, so we try to increase the pupil and the image sampling.

The PSF size increases. In this case, we run the Huygens MTF again:

You can see the MTF results become less different. I would like to perform a higher sampling Huygens MTF result but the analysis takes too much time..

There is a forum post about the cutoff frequency of Huygens MTF. Sorry it’s in Chinese but I think this information is quite useful so I want to share.

 Huygens MTF的截止頻率 | Zemax Community

In summary:

1. Huygens MTF is the most accurate MTF for most cases. But it’s not always the case.
2. For models that is not well focused, please check the Huygens PSF and compare with the spot diagram.
3. In the case that Huygens PSF and the spot diagram are quite different, pay attention to the PSF sampling and the image delta. 

Lastly I would like to comment:

Huygens PSF and MTF requires lots of computation resources. If the system is not close to a diffraction limited situation, using this analysis feature might not be too effective.

Assuming you are near focus, the Huygens PSF is the gold standard. The FFT PSF will give an approximation, and is faster, but the Huygens is the more accurate. The geometric spot will be plain wrong if the spot size is less than the Airy disk.

If you are not near focus, the geometric spot is the correct one to use. The FFT and Huygens will be horribly undersampled if the spot size is not near to the Airy disk.

Thank you Mark, your summary is better! Do you mind me localizing your comment in Chinese so that more engineers can see this?

I’ve just said it more times than you, that’s all  But sure, please feel free to use it however you like. - Mark

You can find related information about Huygens MTF settings in Huygens PSF section: 

The Analyze Tab (sequential ui mode) > Image Quality Group > PSF > Huygens PSF