Reduce number of points if 2D is too big

[lpatiny]

We should consider the raw FID that we are processing in NMRium and also the processed spectrum.

If the dimension > 1024 we will reduce to 1024

https://github.com/mljs/spectra-processing/blob/main/src/xy/xyEquallySpaced.ts

Future improvements will include recalculation of curves when zooming.

[jobo322]

I will reduce/limit the number of zeroFilling in auto processing

[lpatiny]

If we cut the FID we will have more artefact no ? Also we don't always have the processing. So to be consistent I think we need to process after FT. WDYT ?

[jobo322]

yes, I am limiting the SI value, never truncate the FID to avoid problems.

I did check the processing with noesy, hsqc and hmbc, limiting the SI to 1024, the processing is also faster and the result is correct.

[lpatiny]

Ok this is a first step but after we still need to reduce the number of points if we are loading processed data that would not go through this pipeline.

[jobo322]

@lpatiny, the point reduction should be handled at the nmrium level. Would it make sense to apply it immediately after the load-filter step?
This way, spectra acquired with nbPoints > 1024 would be properly reduced right after the filters are applied.

[hamed-musallam]

I think we should reduce the points only just before rendering the contours and do not touch the original data

[jobo322]

Yes, I agree as well. Processed data and some experiments (especially in metabonomics) are acquired with a large number of indirect points.