Do 72 individual pictures and convert those into an avi outside

For[i = 1, i <= 72, i++,
 Export[NotebookDirectory[] <> "1080-Matthen-ChessRotate" <> ToString[i] <> ".jpg",
   photograph[trueImg, -0.3 + 0.025*i]]
 ]

which comes in at about two gigabytes of memory

or do 4 short avi and fuse those into one avi outside

For[i = -0.3, i < 1.5, i = i + 0.45,
 Export[NotebookDirectory[] <> "1080-Matthen-ChessRotate" <> ToString[i] <> ".avi",
   Table[photograph[trueImg, t], {t, i, i + 0.45 - 0.025, 0.025}]]
 ]

Which comes in at just over 1.5 gigabytes of memory

Watch out for little "off by a bit" when adding a sequence of floating point values and comparing against other floating point values. You might consider turning all those into exact rational values which should avoid that problem.

Why I use 'Parallel' is easy: that is the first I try every anim, cause the Pi is extreme slow here - some animations took hours or days to evaluate and export in higher resolutions - and in view of that, I don't like it, when the cpu-usage only shows 25% all the time, hoping that more cpu-power speeds things up a bit.

You're right in view of saving memory, but in this case it even seems to be too large on one kernel: General::nomem: The current computation was aborted because there was insufficient memory available to complete the computation.

Maybe I'll try it with exporting as single pictures tomorrow, I already have had such a .jpg-avi-tool somewhere...hopefully the memory is enough for pictures.....

If I run your code with the only change being replacing ParallelTable with Table and saving the file on my Win desktop then I find (when I carefully restart the kernel each time)

With table step size .025

In[8]:= MaxMemoryUsed[]

Out[8]= 5544518752

With table step size .05

In[8]:= MaxMemoryUsed[]

Out[8]= 2923990656

With table step size.1

In[8]:= MaxMemoryUsed[]

Out[8]= 1614277072

With table step size .2

In[8]:= MaxMemoryUsed[]

Out[8]= 959420360

Those numbers will likely be somewhat different on the Pi. And you will need to look at the documentation for how memory use is reported when using parallel kernels. But the results will probably be close enough to this that the answer will be the same.

So it appears that you need at least 6 gigabytes free to do this. I suppose what is really needed is for someone to build a Raspberry Pi with 64 gigabytes and supported by Mathematica. I'd sign up for a handful of those.