I'm not quite sure of what you are trying to achieve with your quiz, but if you set the marks for the essay questions for images to 0, then they should noo stop the quiz total from being calculated.

Thanks for the hint. Didn't know of google drawing files.

As to the grading problem with essays: My idea would be (not tested yet)
- set the essay question weight to zero. That, however, would make it hard to see which questions have been reviewed already.
- require all STACK questions to be 100% correct (with limitless attempts and no penalty in adaptive mode). This is easy to check. If the STACK questions sum up to 8, and the essays to another 2, this might result in a feasible workflow.

I introduced the STACK based simulation grading because the pdf reports handed in by students were full of trivial mistakes and the manual feedback loop taking a lot of work. My experience is now that 90% of the students don't ask questions at all and get the work done. Another 10 percent asks questions like "I did everything right but Moodle says wrong. What can I do?" for every input field. 75% of these questions are because of bad language skills, they just don't read or understand the question correctly. In these cases I add corresponding hints to the feedback. Another 25% are my fault, because the tolerances are too tight. If I loosen them too much, real errors won't be detected. Thus I have to start with a certain guess and adjust it based on experience.

The requirement to get every answer right for passing the course is sort of an experiment for me. It creates a lot of work for me and for some of the students, because if allowed to just ignore a certain percentage of wrong answers, the students are forced to fight for each individual answer.