No, you didn't miss anything, this version has the whole interface changed to use keyword arguments instead of positional as before, so Siril does not know how to run it properly. I will try to make it backwards-compatible, but not sure if I can...
Hi Nikita,
I use Linux and Siril (so I am mainly interested in your CLI code, not the PI extension) and I would like to use a PC that has a non-AVX compatible CPU. I know that my only hopes would be to A) recompile Tensorflow with SSE4.2 compatibility rather than AVX, or B) use Tensorflow coupled with CUDA (but I am currently having a hard time getting CUDA to work).
However, I noticed you created an "experimental" version of Starnet for Apple M1 (i.e. also natively not AVX-compatible, unless using Rosetta) using other libraries rather than Tensorflow.
So my question is: could there be a similar solution for you, to recompile a Linux CLI experimental version of Starnet using other libraries that do not depend on AVX-compatibility? If you could, I think it would be amazing!
Thanks in advance for your reply.
PS: it is worthwhile reiterating how important your tool has been for this community, and I thank you a lot for it!
Regards,
Renzo