Patch.viz.waterfall() renders different plots across Matplotlib versions

[ahmadtourei]

Description

Patch.viz.waterfall() produces noticeably different plots with apparent saturation/contrast of traces when using matplotlib==3.9.2 versus the newer Matplotlib that gets installed alongside DASCore.

Example

matplotlib==3.9.2

pa.viz.waterfall(scale=0.0001)

matplotlib==3.10.5

pa.viz.waterfall(scale=0.0001)

pa.viz.waterfall(scale=0.01)

Expected behavior

Versions

• OS [e.g. Ubuntu 20.04]: MacOS 15.6
• DASCore Version [e.g. 0.0.5]: 0.1.9
• Python Version [e.g. 3.10]: 3.12

[ahmadtourei]

data can be accessed with:

import gdown, zipfile
import numpy as np
from pathlib import Path
import dascore as dc

url = "https://drive.google.com/file/d/15xMONKL4E00JANze20OW65xBYmPZGmDV/view?usp=sharing"
out = Path("ctemps_das_walking.zip")

# If the zip already exists, skip downloading
if not out.exists():
    # Ensure we save to the expected filename (avoid gdown renaming surprises)
    gdown.download(url=url, output=str(out), quiet=False, fuzzy=True)

das_dir = Path("ctemps_das_walking")

# If the das_dir already exists, skip unzipping
if not das_dir.exists():
    das_dir.mkdir(exist_ok=True)
    with zipfile.ZipFile(out, "r") as zf:
        zf.extractall(das_dir)
    print(f"Unzipped the downloaded data to: {das_dir.resolve()}")

spool = dc.spool(das_dir).update()
patch = spool[0]

[d-chambers]

That's annoying...

[d-chambers]

Hey @ahmadtourei, have you had any thoughts on how we should go about addressing this?

[ahmadtourei]

Hey @d-chambers, I haven’t had a chance yet to investigate what changed in matplotlib.pyplot or imshow() that caused this. Also, I'm curious to see if downsampling would help with visualization (in the 3.10.5 version), as the data is in 40k Hz, and I have seen some weird visualization effects for data with very high sampling rates in the past.

I'm busy with IMAGE stuff till the end of next week so feel free to work on it if you have some time. Thanks!