Merge branch 'develop' into cross-belts

shaketune/post_processing/__init__.py

@@ -0,0 +1,7 @@
+#!/usr/bin/env python3
+
+from .graph_creator import AxesMapFinder as AxesMapFinder
+from .graph_creator import BeltsGraphCreator as BeltsGraphCreator
+from .graph_creator import GraphCreator as GraphCreator
+from .graph_creator import ShaperGraphCreator as ShaperGraphCreator
+from .graph_creator import VibrationsGraphCreator as VibrationsGraphCreator

shaketune/post_processing/analyze_axesmap.py

@@ -0,0 +1,155 @@
+#!/usr/bin/env python3
+
+######################################
+###### AXE_MAP DETECTION SCRIPT ######
+######################################
+# Written by Frix_x#0161 #
+
+import optparse
+
+import numpy as np
+from scipy.signal import butter, filtfilt
+
+from ..helpers.console_output import ConsoleOutput
+
+NUM_POINTS = 500
+
+
+######################################################################
+# Computation
+######################################################################
+
+
+def accel_signal_filter(data, cutoff=2, fs=100, order=5):
+    nyq = 0.5 * fs
+    normal_cutoff = cutoff / nyq
+    b, a = butter(order, normal_cutoff, btype='low', analog=False)
+    filtered_data = filtfilt(b, a, data)
+    filtered_data -= np.mean(filtered_data)
+    return filtered_data
+
+
+def find_first_spike(data):
+    min_index, max_index = np.argmin(data), np.argmax(data)
+    return ('-', min_index) if min_index < max_index else ('', max_index)
+
+
+def get_movement_vector(data, start_idx, end_idx):
+    if start_idx < end_idx:
+        vector = []
+        for i in range(3):
+            vector.append(np.mean(data[i][start_idx:end_idx], axis=0))
+        return vector
+    else:
+        return np.zeros(3)
+
+
+def angle_between(v1, v2):
+    v1_u = v1 / np.linalg.norm(v1)
+    v2_u = v2 / np.linalg.norm(v2)
+    return np.arccos(np.clip(np.dot(v1_u, v2_u), -1.0, 1.0))
+
+
+def compute_errors(filtered_data, spikes_sorted, accel_value, num_points):
+    # Get the movement start points in the correct order from the sorted bag of spikes
+    movement_starts = [spike[0][1] for spike in spikes_sorted]
+
+    # Theoretical unit vectors for X, Y, Z printer axes
+    printer_axes = {'x': np.array([1, 0, 0]), 'y': np.array([0, 1, 0]), 'z': np.array([0, 0, 1])}
+
+    alignment_errors = {}
+    sensitivity_errors = {}
+    for i, axis in enumerate(['x', 'y', 'z']):
+        movement_start = movement_starts[i]
+        movement_end = movement_start + num_points
+        movement_vector = get_movement_vector(filtered_data, movement_start, movement_end)
+        alignment_errors[axis] = angle_between(movement_vector, printer_axes[axis])
+
+        measured_accel_magnitude = np.linalg.norm(movement_vector)
+        if accel_value != 0:
+            sensitivity_errors[axis] = abs(measured_accel_magnitude - accel_value) / accel_value * 100
+        else:
+            sensitivity_errors[axis] = None
+
+    return alignment_errors, sensitivity_errors
+
+
+######################################################################
+# Startup and main routines
+######################################################################
+
+
+def parse_log(logname):
+    with open(logname) as f:
+        for header in f:
+            if not header.startswith('#'):
+                break
+        if not header.startswith('freq,psd_x,psd_y,psd_z,psd_xyz'):
+            # Raw accelerometer data
+            return np.loadtxt(logname, comments='#', delimiter=',')
+    # Power spectral density data or shaper calibration data
+    raise ValueError(
+        'File %s does not contain raw accelerometer data and therefore '
+        'is not supported by this script. Please use the official Klipper '
+        'calibrate_shaper.py script to process it instead.' % (logname,)
+    )
+
+
+def axesmap_calibration(lognames, accel=None):
+    # Parse the raw data and get them ready for analysis
+    raw_datas = [parse_log(filename) for filename in lognames]
+    if len(raw_datas) > 1:
+        raise ValueError('Analysis of multiple CSV files at once is not possible with this script')
+
+    filtered_data = [accel_signal_filter(raw_datas[0][:, i + 1]) for i in range(3)]
+    spikes = [find_first_spike(filtered_data[i]) for i in range(3)]
+    spikes_sorted = sorted([(spikes[0], 'x'), (spikes[1], 'y'), (spikes[2], 'z')], key=lambda x: x[0][1])
+
+    # Using the previous variables to get the axes_map and errors
+    axes_map = ','.join([f'{spike[0][0]}{spike[1]}' for spike in spikes_sorted])
+    # alignment_error, sensitivity_error = compute_errors(filtered_data, spikes_sorted, accel, NUM_POINTS)
+
+    results = f'Be aware that this macro is experimental and has been known to sometimes produce incorrect results. Use it with caution and always check the results!\n'
+    results += f'Detected axes_map:\n  {axes_map}\n'
+
+    # TODO: work on this function that is currently not giving good results...
+    # results += "Accelerometer angle deviation:\n"
+    # for axis, angle in alignment_error.items():
+    #     angle_degrees = np.degrees(angle) # Convert radians to degrees
+    #     results += f"  {axis.upper()} axis: {angle_degrees:.2f} degrees\n"
+
+    # results += "Accelerometer sensitivity error:\n"
+    # for axis, error in sensitivity_error.items():
+    #     results += f"  {axis.upper()} axis: {error:.2f}%\n"
+
+    return results
+
+
+def main():
+    # Parse command-line arguments
+    usage = '%prog [options] <raw logs>'
+    opts = optparse.OptionParser(usage)
+    opts.add_option('-o', '--output', type='string', dest='output', default=None, help='filename of output graph')
+    opts.add_option(
+        '-a', '--accel', type='string', dest='accel', default=None, help='acceleration value used to do the movements'
+    )
+    options, args = opts.parse_args()
+    if len(args) < 1:
+        opts.error('No CSV file(s) to analyse')
+    if options.accel is None:
+        opts.error('You must specify the acceleration value used when generating the CSV file (option -a)')
+    try:
+        accel_value = float(options.accel)
+    except ValueError:
+        opts.error('Invalid acceleration value. It should be a numeric value.')
+
+    results = axesmap_calibration(args, accel_value)
+    ConsoleOutput.print(results)
+
+    if options.output is not None:
+        with open(options.output, 'w') as f:
+            f.write(results)
+
+
+if __name__ == '__main__':
+    main()

shaketune/post_processing/graph_belts.py

@@ -0,0 +1,562 @@
+#!/usr/bin/env python3
+
+#################################################
+######## CoreXY BELTS CALIBRATION SCRIPT ########
+#################################################
+# Written by Frix_x#0161 #
+
+import optparse
+import os
+from collections import namedtuple
+from datetime import datetime
+
+import matplotlib
+import matplotlib.colors
+import matplotlib.font_manager
+import matplotlib.pyplot as plt
+import matplotlib.ticker
+import numpy as np
+
+matplotlib.use('Agg')
+
+from ..helpers.common_func import detect_peaks, parse_log, setup_klipper_import
+from ..helpers.console_output import ConsoleOutput
+
+ALPHABET = 'ABCDEFGHIJKLMNOPQRSTUVWXYZ'  # For paired peaks names
+
+PEAKS_DETECTION_THRESHOLD = 0.1  # Threshold to detect peaks in the PSD signal (10% of max)
+DC_MAX_PEAKS = 2  # Maximum ideal number of peaks
+DC_MAX_UNPAIRED_PEAKS_ALLOWED = 0  # No unpaired peaks are tolerated
+
+# Define the SignalData namedtuple
+SignalData = namedtuple('CalibrationData', ['freqs', 'psd', 'peaks', 'paired_peaks', 'unpaired_peaks'])
+
+KLIPPAIN_COLORS = {
+    'purple': '#70088C',
+    'orange': '#FF8D32',
+    'dark_purple': '#150140',
+    'dark_orange': '#F24130',
+    'red_pink': '#F2055C',
+}
+
+
+######################################################################
+# Computation of the PSD graph
+######################################################################
+
+
+# This function create pairs of peaks that are close in frequency on two curves (that are known
+# to be resonances points and must be similar on both belts on a CoreXY kinematic)
+def pair_peaks(peaks1, freqs1, psd1, peaks2, freqs2, psd2):
+    # Compute a dynamic detection threshold to filter and pair peaks efficiently
+    # even if the signal is very noisy (this get clipped to a maximum of 10Hz diff)
+    distances = []
+    for p1 in peaks1:
+        for p2 in peaks2:
+            distances.append(abs(freqs1[p1] - freqs2[p2]))
+    distances = np.array(distances)
+
+    median_distance = np.median(distances)
+    iqr = np.percentile(distances, 75) - np.percentile(distances, 25)
+
+    threshold = median_distance + 1.5 * iqr
+    threshold = min(threshold, 10)
+
+    # Pair the peaks using the dynamic thresold
+    paired_peaks = []
+    unpaired_peaks1 = list(peaks1)
+    unpaired_peaks2 = list(peaks2)
+
+    while unpaired_peaks1 and unpaired_peaks2:
+        min_distance = threshold + 1
+        pair = None
+
+        for p1 in unpaired_peaks1:
+            for p2 in unpaired_peaks2:
+                distance = abs(freqs1[p1] - freqs2[p2])
+                if distance < min_distance:
+                    min_distance = distance
+                    pair = (p1, p2)
+
+        if pair is None:  # No more pairs below the threshold
+            break
+
+        p1, p2 = pair
+        paired_peaks.append(((p1, freqs1[p1], psd1[p1]), (p2, freqs2[p2], psd2[p2])))
+        unpaired_peaks1.remove(p1)
+        unpaired_peaks2.remove(p2)
+
+    return paired_peaks, unpaired_peaks1, unpaired_peaks2
+
+
+######################################################################
+# Computation of the differential spectrogram
+######################################################################
+
+
+def compute_mhi(similarity_factor, signal1, signal2):
+    num_unpaired_peaks = len(signal1.unpaired_peaks) + len(signal2.unpaired_peaks)
+    num_paired_peaks = len(signal1.paired_peaks)
+    # Combine unpaired peaks from both signals, tagging each peak with its respective signal
+    combined_unpaired_peaks = [(peak, signal1) for peak in signal1.unpaired_peaks] + [
+        (peak, signal2) for peak in signal2.unpaired_peaks
+    ]
+    psd_highest_max = max(signal1.psd.max(), signal2.psd.max())
+
+    # Start with the similarity factor directly scaled to a percentage
+    mhi = similarity_factor
+
+    # Bonus for ideal number of total peaks (1 or 2)
+    if num_paired_peaks >= DC_MAX_PEAKS:
+        mhi *= DC_MAX_PEAKS / num_paired_peaks  # Reduce MHI if more than ideal number of peaks
+
+    # Penalty from unpaired peaks weighted by their amplitude relative to the maximum PSD amplitude
+    unpaired_peak_penalty = 0
+    if num_unpaired_peaks > DC_MAX_UNPAIRED_PEAKS_ALLOWED:
+        for peak, signal in combined_unpaired_peaks:
+            unpaired_peak_penalty += (signal.psd[peak] / psd_highest_max) * 30
+        mhi -= unpaired_peak_penalty
+
+    # Ensure the result lies between 0 and 100 by clipping the computed value
+    mhi = np.clip(mhi, 0, 100)
+
+    return mhi_lut(mhi)
+
+
+# LUT to transform the MHI into a textual value easy to understand for the users of the script
+def mhi_lut(mhi):
+    ranges = [
+        (70, 100, 'Excellent mechanical health'),
+        (55, 70, 'Good mechanical health'),
+        (45, 55, 'Acceptable mechanical health'),
+        (30, 45, 'Potential signs of a mechanical issue'),
+        (15, 30, 'Likely a mechanical issue'),
+        (0, 15, 'Mechanical issue detected'),
+    ]
+    mhi = np.clip(mhi, 1, 100)
+    for lower, upper, message in ranges:
+        if lower < mhi <= upper:
+            return message
+
+    return 'Unknown mechanical health'  # Should never happen
+
+
+######################################################################
+# Graphing
+######################################################################
+
+
+def plot_compare_frequency(ax, signal1, signal2, signal1_belt, signal2_belt, max_freq):
+    # Plot the two belts PSD signals
+    ax.plot(signal1.freqs, signal1.psd, label='Belt ' + signal1_belt, color=KLIPPAIN_COLORS['purple'])
+    ax.plot(signal2.freqs, signal2.psd, label='Belt ' + signal2_belt, color=KLIPPAIN_COLORS['orange'])
+
+    # Trace and annotate the peaks on the graph
+    paired_peak_count = 0
+    unpaired_peak_count = 0
+    offsets_table_data = []
+
+    for _, (peak1, peak2) in enumerate(signal1.paired_peaks):
+        label = ALPHABET[paired_peak_count]
+        amplitude_offset = abs(
+            ((signal2.psd[peak2[0]] - signal1.psd[peak1[0]]) / max(signal1.psd[peak1[0]], signal2.psd[peak2[0]])) * 100
+        )
+        frequency_offset = abs(signal2.freqs[peak2[0]] - signal1.freqs[peak1[0]])
+        offsets_table_data.append([f'Peaks {label}', f'{frequency_offset:.1f} Hz', f'{amplitude_offset:.1f} %'])
+
+        ax.plot(signal1.freqs[peak1[0]], signal1.psd[peak1[0]], 'x', color='black')
+        ax.plot(signal2.freqs[peak2[0]], signal2.psd[peak2[0]], 'x', color='black')
+        ax.plot(
+            [signal1.freqs[peak1[0]], signal2.freqs[peak2[0]]],
+            [signal1.psd[peak1[0]], signal2.psd[peak2[0]]],
+            ':',
+            color='gray',
+        )
+
+        ax.annotate(
+            label + '1',
+            (signal1.freqs[peak1[0]], signal1.psd[peak1[0]]),
+            textcoords='offset points',
+            xytext=(8, 5),
+            ha='left',
+            fontsize=13,
+            color='black',
+        )
+        ax.annotate(
+            label + '2',
+            (signal2.freqs[peak2[0]], signal2.psd[peak2[0]]),
+            textcoords='offset points',
+            xytext=(8, 5),
+            ha='left',
+            fontsize=13,
+            color='black',
+        )
+        paired_peak_count += 1
+
+    for peak in signal1.unpaired_peaks:
+        ax.plot(signal1.freqs[peak], signal1.psd[peak], 'x', color='black')
+        ax.annotate(
+            str(unpaired_peak_count + 1),
+            (signal1.freqs[peak], signal1.psd[peak]),
+            textcoords='offset points',
+            xytext=(8, 5),
+            ha='left',
+            fontsize=13,
+            color='red',
+            weight='bold',
+        )
+        unpaired_peak_count += 1
+
+    for peak in signal2.unpaired_peaks:
+        ax.plot(signal2.freqs[peak], signal2.psd[peak], 'x', color='black')
+        ax.annotate(
+            str(unpaired_peak_count + 1),
+            (signal2.freqs[peak], signal2.psd[peak]),
+            textcoords='offset points',
+            xytext=(8, 5),
+            ha='left',
+            fontsize=13,
+            color='red',
+            weight='bold',
+        )
+        unpaired_peak_count += 1
+
+    # Add estimated similarity to the graph
+    ax2 = ax.twinx()  # To split the legends in two box
+    ax2.yaxis.set_visible(False)
+    ax2.plot([], [], ' ', label=f'Number of unpaired peaks: {unpaired_peak_count}')
+
+    # Setting axis parameters, grid and graph title
+    ax.set_xlabel('Frequency (Hz)')
+    ax.set_xlim([0, max_freq])
+    ax.set_ylabel('Power spectral density')
+    psd_highest_max = max(signal1.psd.max(), signal2.psd.max())
+    ax.set_ylim([0, psd_highest_max * 1.1])
+
+    ax.xaxis.set_minor_locator(matplotlib.ticker.AutoMinorLocator())
+    ax.yaxis.set_minor_locator(matplotlib.ticker.AutoMinorLocator())
+    ax.ticklabel_format(axis='y', style='scientific', scilimits=(0, 0))
+    ax.grid(which='major', color='grey')
+    ax.grid(which='minor', color='lightgrey')
+    fontP = matplotlib.font_manager.FontProperties()
+    fontP.set_size('small')
+    ax.set_title(
+        'Belts frequency profiles',
+        fontsize=14,
+        color=KLIPPAIN_COLORS['dark_orange'],
+        weight='bold',
+    )
+
+    # Print the table of offsets ontop of the graph below the original legend (upper right)
+    if len(offsets_table_data) > 0:
+        columns = [
+            '',
+            'Frequency delta',
+            'Amplitude delta',
+        ]
+        offset_table = ax.table(
+            cellText=offsets_table_data,
+            colLabels=columns,
+            bbox=[0.66, 0.79, 0.33, 0.15],
+            loc='upper right',
+            cellLoc='center',
+        )
+        offset_table.auto_set_font_size(False)
+        offset_table.set_fontsize(8)
+        offset_table.auto_set_column_width([0, 1, 2])
+        offset_table.set_zorder(100)
+        cells = [key for key in offset_table.get_celld().keys()]
+        for cell in cells:
+            offset_table[cell].set_facecolor('white')
+            offset_table[cell].set_alpha(0.6)
+
+    ax.legend(loc='upper left', prop=fontP)
+    ax2.legend(loc='upper right', prop=fontP)
+
+    return
+
+
+# Compute quantile-quantile plot to compare the two belts
+def plot_versus_belts(ax, common_freqs, signal1, signal2, interp_psd1, interp_psd2, signal1_belt, signal2_belt):
+    ax.set_title('Cross-belts comparison plot', fontsize=14, color=KLIPPAIN_COLORS['dark_orange'], weight='bold')
+
+    max_psd = max(np.max(interp_psd1), np.max(interp_psd2))
+    ideal_line = np.linspace(0, max_psd * 1.1, 500)
+    green_boundary = ideal_line + (0.35 * max_psd * np.exp(-ideal_line / (0.6 * max_psd)))
+    ax.fill_betweenx(ideal_line, ideal_line, green_boundary, color='green', alpha=0.15)
+    ax.fill_between(ideal_line, ideal_line, green_boundary, color='green', alpha=0.15, label='Good zone')
+    ax.plot(
+        ideal_line,
+        ideal_line,
+        '--',
+        label='Ideal line',
+        color='red',
+        linewidth=2,
+    )
+
+    ax.plot(interp_psd1, interp_psd2, color='dimgrey', marker='o', markersize=1.5)
+    ax.fill_betweenx(interp_psd2, interp_psd1, color=KLIPPAIN_COLORS['red_pink'], alpha=0.1)
+
+    paired_peak_count = 0
+    unpaired_peak_count = 0
+
+    for _, (peak1, peak2) in enumerate(signal1.paired_peaks):
+        label = ALPHABET[paired_peak_count]
+        freq1 = signal1.freqs[peak1[0]]
+        freq2 = signal2.freqs[peak2[0]]
+        nearest_idx1 = np.argmin(np.abs(common_freqs - freq1))
+        nearest_idx2 = np.argmin(np.abs(common_freqs - freq2))
+
+        if nearest_idx1 == nearest_idx2:
+            psd1_peak_value = interp_psd1[nearest_idx1]
+            psd2_peak_value = interp_psd2[nearest_idx1]
+            ax.plot(psd1_peak_value, psd2_peak_value, marker='o', color='black', markersize=7)
+            ax.annotate(
+                f'{label}1/{label}2',
+                (psd1_peak_value, psd2_peak_value),
+                textcoords='offset points',
+                xytext=(-7, 7),
+                fontsize=13,
+                color='black',
+            )
+        else:
+            psd1_peak_value = interp_psd1[nearest_idx1]
+            psd1_on_peak = interp_psd1[nearest_idx2]
+            psd2_peak_value = interp_psd2[nearest_idx2]
+            psd2_on_peak = interp_psd2[nearest_idx1]
+            ax.plot(psd1_on_peak, psd2_peak_value, marker='o', color=KLIPPAIN_COLORS['orange'], markersize=7)
+            ax.plot(psd1_peak_value, psd2_on_peak, marker='o', color=KLIPPAIN_COLORS['purple'], markersize=7)
+            ax.annotate(
+                f'{label}1',
+                (psd1_peak_value, psd2_on_peak),
+                textcoords='offset points',
+                xytext=(0, 7),
+                fontsize=13,
+                color='black',
+            )
+            ax.annotate(
+                f'{label}2',
+                (psd1_on_peak, psd2_peak_value),
+                textcoords='offset points',
+                xytext=(0, 7),
+                fontsize=13,
+                color='black',
+            )
+        paired_peak_count += 1
+
+    for _, peak_index in enumerate(signal1.unpaired_peaks):
+        freq1 = signal1.freqs[peak_index]
+        freq2 = signal2.freqs[peak_index]
+        nearest_idx1 = np.argmin(np.abs(common_freqs - freq1))
+        nearest_idx2 = np.argmin(np.abs(common_freqs - freq2))
+        psd1_peak_value = interp_psd1[nearest_idx1]
+        psd2_peak_value = interp_psd2[nearest_idx1]
+        ax.plot(psd1_peak_value, psd2_peak_value, marker='o', color=KLIPPAIN_COLORS['purple'], markersize=7)
+        ax.annotate(
+            str(unpaired_peak_count + 1),
+            (psd1_peak_value, psd2_peak_value),
+            textcoords='offset points',
+            fontsize=13,
+            weight='bold',
+            color=KLIPPAIN_COLORS['red_pink'],
+            xytext=(0, 7),
+        )
+        unpaired_peak_count += 1
+
+    for _, peak_index in enumerate(signal2.unpaired_peaks):
+        freq1 = signal1.freqs[peak_index]
+        freq2 = signal2.freqs[peak_index]
+        nearest_idx1 = np.argmin(np.abs(common_freqs - freq1))
+        nearest_idx2 = np.argmin(np.abs(common_freqs - freq2))
+        psd1_peak_value = interp_psd1[nearest_idx1]
+        psd2_peak_value = interp_psd2[nearest_idx1]
+        ax.plot(psd1_peak_value, psd2_peak_value, marker='o', color=KLIPPAIN_COLORS['orange'], markersize=7)
+        ax.annotate(
+            str(unpaired_peak_count + 1),
+            (psd1_peak_value, psd2_peak_value),
+            textcoords='offset points',
+            fontsize=13,
+            weight='bold',
+            color=KLIPPAIN_COLORS['red_pink'],
+            xytext=(0, 7),
+        )
+        unpaired_peak_count += 1
+
+    ax.set_xlabel(f'Belt {signal1_belt}')
+    ax.set_ylabel(f'Belt {signal2_belt}')
+    ax.set_xlim([0, max_psd * 1.1])
+    ax.set_ylim([0, max_psd * 1.1])
+
+    ax.xaxis.set_minor_locator(matplotlib.ticker.AutoMinorLocator())
+    ax.yaxis.set_minor_locator(matplotlib.ticker.AutoMinorLocator())
+    ax.ticklabel_format(axis='y', style='scientific', scilimits=(0, 0))
+    ax.grid(which='major', color='grey')
+    ax.grid(which='minor', color='lightgrey')
+
+    fontP = matplotlib.font_manager.FontProperties()
+    fontP.set_size('medium')
+    ax.legend(loc='upper left', prop=fontP)
+
+    return
+
+
+######################################################################
+# Custom tools
+######################################################################
+
+
+# Original Klipper function to get the PSD data of a raw accelerometer signal
+def compute_signal_data(data, max_freq):
+    helper = shaper_calibrate.ShaperCalibrate(printer=None)
+    calibration_data = helper.process_accelerometer_data(data)
+
+    freqs = calibration_data.freq_bins[calibration_data.freq_bins <= max_freq]
+    psd = calibration_data.get_psd('all')[calibration_data.freq_bins <= max_freq]
+
+    _, peaks, _ = detect_peaks(psd, freqs, PEAKS_DETECTION_THRESHOLD * psd.max())
+
+    return SignalData(freqs=freqs, psd=psd, peaks=peaks, paired_peaks=None, unpaired_peaks=None)
+
+
+######################################################################
+# Startup and main routines
+######################################################################
+
+
+def belts_calibration(
+    lognames, kinematics, klipperdir='~/klipper', max_freq=200.0, accel_per_hz=None, st_version='unknown'
+):
+    global shaper_calibrate
+    shaper_calibrate = setup_klipper_import(klipperdir)
+
+    # Parse data from the log files while ignoring CSV in the wrong format
+    datas = [data for data in (parse_log(fn) for fn in lognames) if data is not None]
+    if len(datas) > 2:
+        raise ValueError('Incorrect number of .csv files used (this function needs exactly two files to compare them)!')
+
+    # Get the belts name for the legend to avoid putting the full file name
+    belt_info = {'A': ' (axis 1,-1)', 'B': ' (axis 1, 1)'}
+    signal1_belt = (lognames[0].split('/')[-1]).split('_')[-1][0]
+    signal2_belt = (lognames[1].split('/')[-1]).split('_')[-1][0]
+    signal1_belt += belt_info.get(signal1_belt, '')
+    signal2_belt += belt_info.get(signal2_belt, '')
+
+    # Compute calibration data for the two datasets with automatic peaks detection
+    signal1 = compute_signal_data(datas[0], max_freq)
+    signal2 = compute_signal_data(datas[1], max_freq)
+    del datas
+
+    # Pair the peaks across the two datasets
+    paired_peaks, unpaired_peaks1, unpaired_peaks2 = pair_peaks(
+        signal1.peaks, signal1.freqs, signal1.psd, signal2.peaks, signal2.freqs, signal2.psd
+    )
+    signal1 = signal1._replace(paired_peaks=paired_peaks, unpaired_peaks=unpaired_peaks1)
+    signal2 = signal2._replace(paired_peaks=paired_peaks, unpaired_peaks=unpaired_peaks2)
+
+    # Re-interpolate the PSD signals to a common frequency range to be able to plot them one against the other point by point
+    common_freqs = np.linspace(0, max_freq, 500)
+    interp_psd1 = np.interp(common_freqs, signal1.freqs, signal1.psd)
+    interp_psd2 = np.interp(common_freqs, signal2.freqs, signal2.psd)
+
+    # Calculating R^2 to y=x line to compute the similarity between the two belts
+    ss_res = np.sum((interp_psd2 - interp_psd1) ** 2)
+    ss_tot = np.sum((interp_psd2 - np.mean(interp_psd2)) ** 2)
+    similarity_factor = (1 - (ss_res / ss_tot)) * 100
+    ConsoleOutput.print(f'Belts estimated similarity: {similarity_factor:.1f}%')
+
+    # mhi = compute_mhi(similarity_factor, num_peaks, num_unpaired_peaks)
+    mhi = compute_mhi(similarity_factor, signal1, signal2)
+    ConsoleOutput.print(f'[experimental] Mechanical health: {mhi}')
+
+    fig, ((ax1, ax3)) = plt.subplots(
+        1,
+        2,
+        gridspec_kw={
+            'width_ratios': [5, 3],
+            'bottom': 0.080,
+            'top': 0.840,
+            'left': 0.050,
+            'right': 0.985,
+            'hspace': 0.166,
+            'wspace': 0.138,
+        },
+    )
+    fig.set_size_inches(15, 7)
+
+    # Add title
+    title_line1 = 'RELATIVE BELTS CALIBRATION TOOL'
+    fig.text(
+        0.060, 0.947, title_line1, ha='left', va='bottom', fontsize=20, color=KLIPPAIN_COLORS['purple'], weight='bold'
+    )
+    try:
+        filename = lognames[0].split('/')[-1]
+        dt = datetime.strptime(f"{filename.split('_')[1]} {filename.split('_')[2]}", '%Y%m%d %H%M%S')
+        title_line2 = dt.strftime('%x %X')
+        if kinematics is not None:
+            title_line2 += ' -- ' + kinematics.upper() + ' kinematics'
+    except Exception:
+        ConsoleOutput.print(
+            'Warning: CSV filenames look to be different than expected (%s , %s)' % (lognames[0], lognames[1])
+        )
+        title_line2 = lognames[0].split('/')[-1] + ' / ' + lognames[1].split('/')[-1]
+    fig.text(0.060, 0.939, title_line2, ha='left', va='top', fontsize=16, color=KLIPPAIN_COLORS['dark_purple'])
+
+    # We add the estimated similarity and the MHI value to the title only if the kinematics is CoreXY
+    # as it make no sense to compute these values for other kinematics that doesn't have paired belts
+    if kinematics == 'corexy':
+        title_line3 = f'| Estimated similarity: {similarity_factor:.1f}%'
+        title_line4 = f'| {mhi} (experimental)'
+        fig.text(0.55, 0.985, title_line3, ha='left', va='top', fontsize=14, color=KLIPPAIN_COLORS['dark_purple'])
+        fig.text(0.55, 0.950, title_line4, ha='left', va='top', fontsize=14, color=KLIPPAIN_COLORS['dark_purple'])
+
+    # Add the accel_per_hz value to the title
+    title_line5 = f'| Accel per Hz used: {accel_per_hz} mm/s²/Hz'
+    fig.text(0.55, 0.915, title_line5, ha='left', va='top', fontsize=14, color=KLIPPAIN_COLORS['dark_purple'])
+
+    # Plot the graphs
+    plot_compare_frequency(ax1, signal1, signal2, signal1_belt, signal2_belt, max_freq)
+    plot_versus_belts(ax3, common_freqs, signal1, signal2, interp_psd1, interp_psd2, signal1_belt, signal2_belt)
+
+    # Adding a small Klippain logo to the top left corner of the figure
+    ax_logo = fig.add_axes([0.001, 0.894, 0.105, 0.105], anchor='NW')
+    ax_logo.imshow(plt.imread(os.path.join(os.path.dirname(os.path.abspath(__file__)), 'klippain.png')))
+    ax_logo.axis('off')
+
+    # Adding Shake&Tune version in the top right corner
+    if st_version != 'unknown':
+        fig.text(0.995, 0.985, st_version, ha='right', va='bottom', fontsize=8, color=KLIPPAIN_COLORS['purple'])
+
+    return fig
+
+
+def main():
+    # Parse command-line arguments
+    usage = '%prog [options] <raw logs>'
+    opts = optparse.OptionParser(usage)
+    opts.add_option('-o', '--output', type='string', dest='output', default=None, help='filename of output graph')
+    opts.add_option('-f', '--max_freq', type='float', default=200.0, help='maximum frequency to graph')
+    opts.add_option('--accel_per_hz', type='float', default=None, help='accel_per_hz used during the measurement')
+    opts.add_option(
+        '-k', '--klipper_dir', type='string', dest='klipperdir', default='~/klipper', help='main klipper directory'
+    )
+    opts.add_option(
+        '-m',
+        '--kinematics',
+        type='string',
+        dest='kinematics',
+        help='machine kinematics configuration',
+    )
+    options, args = opts.parse_args()
+    if len(args) < 1:
+        opts.error('Incorrect number of arguments')
+    if options.output is None:
+        opts.error('You must specify an output file.png to use the script (option -o)')
+
+    fig = belts_calibration(
+        args, options.kinematics, options.klipperdir, options.max_freq, options.accel_per_hz, 'unknown'
+    )
+    fig.savefig(options.output, dpi=150)
+
+
+if __name__ == '__main__':
+    main()

shaketune/post_processing/graph_creator.py

@@ -0,0 +1,279 @@
+#!/usr/bin/env python3
+
+import abc
+import re
+import shutil
+import tarfile
+from datetime import datetime
+from pathlib import Path
+from typing import Callable, Optional
+
+from matplotlib.figure import Figure
+
+from ..helpers import filemanager as fm
+from ..helpers.console_output import ConsoleOutput
+from ..measurement.motorsconfigparser import MotorsConfigParser
+from ..shaketune_config import ShakeTuneConfig
+from .analyze_axesmap import axesmap_calibration
+from .graph_belts import belts_calibration
+from .graph_shaper import shaper_calibration
+from .graph_vibrations import vibrations_profile
+
+
+class GraphCreator(abc.ABC):
+    def __init__(self, config: ShakeTuneConfig):
+        self._config = config
+
+        self._graph_date = datetime.now().strftime('%Y%m%d_%H%M%S')
+        self._version = ShakeTuneConfig.get_git_version()
+
+        self._type = None
+        self._folder = None
+
+    def _setup_folder(self, graph_type: str) -> None:
+        self._type = graph_type
+        self._folder = self._config.get_results_folder(graph_type)
+
+    def _move_and_prepare_files(
+        self,
+        glob_pattern: str,
+        min_files_required: Optional[int] = None,
+        custom_name_func: Optional[Callable[[Path], str]] = None,
+    ) -> list[Path]:
+        tmp_path = Path('/tmp')
+        globbed_files = list(tmp_path.glob(glob_pattern))
+
+        # If min_files_required is not set, use the number of globbed files as the minimum
+        min_files_required = min_files_required or len(globbed_files)
+
+        if not globbed_files:
+            raise FileNotFoundError(f'no CSV files found in the /tmp folder to create the {self._type} graphs!')
+        if len(globbed_files) < min_files_required:
+            raise FileNotFoundError(f'{min_files_required} CSV files are needed to create the {self._type} graphs!')
+
+        lognames = []
+        for filename in sorted(globbed_files, key=lambda f: f.stat().st_mtime, reverse=True)[:min_files_required]:
+            fm.wait_file_ready(filename)
+            custom_name = custom_name_func(filename) if custom_name_func else filename.name
+            new_file = self._folder / f'{self._type}_{self._graph_date}_{custom_name}.csv'
+            # shutil.move() is needed to move the file across filesystems (mainly for BTT CB1 Pi default OS image)
+            shutil.move(filename, new_file)
+            lognames.append(new_file)
+        return lognames
+
+    def _save_figure_and_cleanup(self, fig: Figure, lognames: list[Path], axis_label: Optional[str] = None) -> None:
+        axis_suffix = f'_{axis_label}' if axis_label else ''
+        png_filename = self._folder / f'{self._type}_{self._graph_date}{axis_suffix}.png'
+        fig.savefig(png_filename, dpi=self._config.dpi)
+
+        if self._config.keep_csv:
+            self._archive_files(lognames)
+        else:
+            self._remove_files(lognames)
+
+    def _archive_files(self, _: list[Path]) -> None:
+        return
+
+    def _remove_files(self, lognames: list[Path]) -> None:
+        for csv in lognames:
+            csv.unlink(missing_ok=True)
+
+    def get_type(self) -> str:
+        return self._type
+
+    @abc.abstractmethod
+    def create_graph(self) -> None:
+        pass
+
+    @abc.abstractmethod
+    def clean_old_files(self, keep_results: int) -> None:
+        pass
+
+
+class BeltsGraphCreator(GraphCreator):
+    def __init__(self, config: ShakeTuneConfig):
+        super().__init__(config)
+
+        self._kinematics = None
+        self._accel_per_hz = None
+
+        self._setup_folder('belts')
+
+    def configure(self, kinematics: str = None, accel_per_hz: float = None) -> None:
+        self._kinematics = kinematics
+        self._accel_per_hz = accel_per_hz
+
+    def create_graph(self) -> None:
+        lognames = self._move_and_prepare_files(
+            glob_pattern='shaketune-belt_*.csv',
+            min_files_required=2,
+            custom_name_func=lambda f: f.stem.split('_')[1].upper(),
+        )
+        fig = belts_calibration(
+            lognames=[str(path) for path in lognames],
+            klipperdir=str(self._config.klipper_folder),
+            accel_per_hz=self._accel_per_hz,
+            st_version=self._version,
+        )
+        self._save_figure_and_cleanup(fig, lognames)
+
+    def clean_old_files(self, keep_results: int = 3) -> None:
+        # Get all PNG files in the directory as a list of Path objects
+        files = sorted(self._folder.glob('*.png'), key=lambda f: f.stat().st_mtime, reverse=True)
+
+        if len(files) <= keep_results:
+            return  # No need to delete any files
+
+        # Delete the older files
+        for old_file in files[keep_results:]:
+            file_date = '_'.join(old_file.stem.split('_')[1:3])
+            for suffix in ['A', 'B']:
+                csv_file = self._folder / f'belts_{file_date}_{suffix}.csv'
+                csv_file.unlink(missing_ok=True)
+            old_file.unlink()
+
+
+class ShaperGraphCreator(GraphCreator):
+    def __init__(self, config: ShakeTuneConfig):
+        super().__init__(config)
+
+        self._max_smoothing = None
+        self._scv = None
+
+        self._setup_folder('shaper')
+
+    def configure(self, scv: float, max_smoothing: float = None, accel_per_hz: float = None) -> None:
+        self._scv = scv
+        self._max_smoothing = max_smoothing
+        self._accel_per_hz = accel_per_hz
+
+    def create_graph(self) -> None:
+        if not self._scv:
+            raise ValueError('scv must be set to create the input shaper graph!')
+
+        lognames = self._move_and_prepare_files(
+            glob_pattern='shaketune-axis_*.csv',
+            min_files_required=1,
+            custom_name_func=lambda f: f.stem.split('_')[1].upper(),
+        )
+        fig = shaper_calibration(
+            lognames=[str(path) for path in lognames],
+            klipperdir=str(self._config.klipper_folder),
+            max_smoothing=self._max_smoothing,
+            scv=self._scv,
+            accel_per_hz=self._accel_per_hz,
+            st_version=self._version,
+        )
+        self._save_figure_and_cleanup(fig, lognames, lognames[0].stem.split('_')[-1])
+
+    def clean_old_files(self, keep_results: int = 3) -> None:
+        # Get all PNG files in the directory as a list of Path objects
+        files = sorted(self._folder.glob('*.png'), key=lambda f: f.stat().st_mtime, reverse=True)
+
+        if len(files) <= 2 * keep_results:
+            return  # No need to delete any files
+
+        # Delete the older files
+        for old_file in files[2 * keep_results :]:
+            csv_file = old_file.with_suffix('.csv')
+            csv_file.unlink(missing_ok=True)
+            old_file.unlink()
+
+
+class VibrationsGraphCreator(GraphCreator):
+    def __init__(self, config: ShakeTuneConfig):
+        super().__init__(config)
+
+        self._kinematics = None
+        self._accel = None
+        self._motors = None
+
+        self._setup_folder('vibrations')
+
+    def configure(self, kinematics: str, accel: float, motor_config_parser: MotorsConfigParser) -> None:
+        self._kinematics = kinematics
+        self._accel = accel
+        self._motors = motor_config_parser.get_motors()
+
+    def _archive_files(self, lognames: list[Path]) -> None:
+        tar_path = self._folder / f'{self._type}_{self._graph_date}.tar.gz'
+        with tarfile.open(tar_path, 'w:gz') as tar:
+            for csv_file in lognames:
+                tar.add(csv_file, arcname=csv_file.name, recursive=False)
+                csv_file.unlink()
+
+    def create_graph(self) -> None:
+        if not self._accel or not self._kinematics:
+            raise ValueError('accel, chip_name and kinematics must be set to create the vibrations profile graph!')
+
+        lognames = self._move_and_prepare_files(
+            glob_pattern='shaketune-vib_*.csv',
+            min_files_required=None,
+            custom_name_func=lambda f: re.search(r'shaketune-vib_(.*?)_\d{8}_\d{6}', f.name).group(1),
+        )
+        fig = vibrations_profile(
+            lognames=[str(path) for path in lognames],
+            klipperdir=str(self._config.klipper_folder),
+            kinematics=self._kinematics,
+            accel=self._accel,
+            st_version=self._version,
+            motors=self._motors,
+        )
+        self._save_figure_and_cleanup(fig, lognames)
+
+    def clean_old_files(self, keep_results: int = 3) -> None:
+        # Get all PNG files in the directory as a list of Path objects
+        files = sorted(self._folder.glob('*.png'), key=lambda f: f.stat().st_mtime, reverse=True)
+
+        if len(files) <= keep_results:
+            return  # No need to delete any files
+
+        # Delete the older files
+        for old_file in files[keep_results:]:
+            old_file.unlink()
+            tar_file = old_file.with_suffix('.tar.gz')
+            tar_file.unlink(missing_ok=True)
+
+
+class AxesMapFinder(GraphCreator):
+    def __init__(self, config: ShakeTuneConfig):
+        super().__init__(config)
+
+        self._graph_date = datetime.now().strftime('%Y%m%d_%H%M%S')
+        self._type = 'axesmap'
+        self._folder = config.get_results_folder()
+
+        self._accel = None
+
+    def configure(self, accel: int) -> None:
+        self._accel = accel
+
+    def find_axesmap(self) -> None:
+        tmp_folder = Path('/tmp')
+        globbed_files = list(tmp_folder.glob('shaketune-axemap_*.csv'))
+
+        if not globbed_files:
+            raise FileNotFoundError('no CSV files found in the /tmp folder to find the axes map!')
+
+        # Find the CSV files with the latest timestamp and process it
+        logname = sorted(globbed_files, key=lambda f: f.stat().st_mtime, reverse=True)[0]
+        results = axesmap_calibration(
+            lognames=[str(logname)],
+            accel=self._accel,
+        )
+        ConsoleOutput.print(results)
+
+        result_filename = self._folder / f'{self._type}_{self._graph_date}.txt'
+        with result_filename.open('w') as f:
+            f.write(results)
+
+    # While the AxesMapFinder doesn't directly create a graph, we need to implement this
+    # method to allow using it seemlessly like all the other GraphCreator objects
+    def create_graph(self) -> None:
+        self.find_axesmap()
+
+    def clean_old_files(self, keep_results: int) -> None:
+        tmp_folder = Path('/tmp')
+        globbed_files = list(tmp_folder.glob('shaketune-axemap_*.csv'))
+        for csv_file in globbed_files:
+            csv_file.unlink()

shaketune/post_processing/graph_shaper.py

@@ -0,0 +1,435 @@
+#!/usr/bin/env python3
+
+#################################################
+######## INPUT SHAPER CALIBRATION SCRIPT ########
+#################################################
+# Derived from the calibrate_shaper.py official Klipper script
+# Copyright (C) 2020  Dmitry Butyugin <dmbutyugin@google.com>
+# Copyright (C) 2020  Kevin O'Connor <kevin@koconnor.net>
+# Highly modified and improved by Frix_x#0161 #
+
+import optparse
+import os
+from datetime import datetime
+
+import matplotlib
+import matplotlib.font_manager
+import matplotlib.pyplot as plt
+import matplotlib.ticker
+import numpy as np
+
+matplotlib.use('Agg')
+
+from ..helpers.common_func import (
+    compute_mechanical_parameters,
+    compute_spectrogram,
+    detect_peaks,
+    parse_log,
+    setup_klipper_import,
+)
+from ..helpers.console_output import ConsoleOutput
+
+PEAKS_DETECTION_THRESHOLD = 0.05
+PEAKS_EFFECT_THRESHOLD = 0.12
+SPECTROGRAM_LOW_PERCENTILE_FILTER = 5
+MAX_SMOOTHING = 0.1
+
+KLIPPAIN_COLORS = {
+    'purple': '#70088C',
+    'orange': '#FF8D32',
+    'dark_purple': '#150140',
+    'dark_orange': '#F24130',
+    'red_pink': '#F2055C',
+}
+
+
+######################################################################
+# Computation
+######################################################################
+
+
+# Find the best shaper parameters using Klipper's official algorithm selection with
+# a proper precomputed damping ratio (zeta) and using the configured printer SQV value
+def calibrate_shaper(datas, max_smoothing, scv, max_freq):
+    helper = shaper_calibrate.ShaperCalibrate(printer=None)
+    calibration_data = helper.process_accelerometer_data(datas)
+    calibration_data.normalize_to_frequencies()
+
+    fr, zeta, _, _ = compute_mechanical_parameters(calibration_data.psd_sum, calibration_data.freq_bins)
+
+    # If the damping ratio computation fail, we use Klipper default value instead
+    if zeta is None:
+        zeta = 0.1
+
+    compat = False
+    try:
+        shaper, all_shapers = helper.find_best_shaper(
+            calibration_data,
+            shapers=None,
+            damping_ratio=zeta,
+            scv=scv,
+            shaper_freqs=None,
+            max_smoothing=max_smoothing,
+            test_damping_ratios=None,
+            max_freq=max_freq,
+            logger=ConsoleOutput.print,
+        )
+    except TypeError:
+        ConsoleOutput.print(
+            '[WARNING] You seem to be using an older version of Klipper that is not compatible with all the latest Shake&Tune features!'
+        )
+        ConsoleOutput.print(
+            'Shake&Tune now runs in compatibility mode: be aware that the results may be slightly off, since the real damping ratio cannot be used to create the filter recommendations'
+        )
+        compat = True
+        shaper, all_shapers = helper.find_best_shaper(calibration_data, max_smoothing, ConsoleOutput.print)
+
+    ConsoleOutput.print(
+        '\n-> Recommended shaper is %s @ %.1f Hz (when using a square corner velocity of %.1f and a damping ratio of %.3f)'
+        % (shaper.name.upper(), shaper.freq, scv, zeta)
+    )
+
+    return shaper.name, all_shapers, calibration_data, fr, zeta, compat
+
+
+######################################################################
+# Graphing
+######################################################################
+
+
+def plot_freq_response(
+    ax, calibration_data, shapers, performance_shaper, peaks, peaks_freqs, peaks_threshold, fr, zeta, max_freq
+):
+    freqs = calibration_data.freqs
+    psd = calibration_data.psd_sum
+    px = calibration_data.psd_x
+    py = calibration_data.psd_y
+    pz = calibration_data.psd_z
+
+    fontP = matplotlib.font_manager.FontProperties()
+    fontP.set_size('x-small')
+
+    ax.set_xlabel('Frequency (Hz)')
+    ax.set_xlim([0, max_freq])
+    ax.set_ylabel('Power spectral density')
+    ax.set_ylim([0, psd.max() + psd.max() * 0.05])
+
+    ax.plot(freqs, psd, label='X+Y+Z', color='purple', zorder=5)
+    ax.plot(freqs, px, label='X', color='red')
+    ax.plot(freqs, py, label='Y', color='green')
+    ax.plot(freqs, pz, label='Z', color='blue')
+
+    ax.xaxis.set_minor_locator(matplotlib.ticker.MultipleLocator(5))
+    ax.yaxis.set_minor_locator(matplotlib.ticker.AutoMinorLocator())
+    ax.ticklabel_format(axis='y', style='scientific', scilimits=(0, 0))
+    ax.grid(which='major', color='grey')
+    ax.grid(which='minor', color='lightgrey')
+
+    ax2 = ax.twinx()
+    ax2.yaxis.set_visible(False)
+
+    lowvib_shaper_vibrs = float('inf')
+    lowvib_shaper = None
+    lowvib_shaper_freq = None
+    lowvib_shaper_accel = 0
+
+    # Draw the shappers curves and add their specific parameters in the legend
+    # This adds also a way to find the best shaper with a low level of vibrations (with a resonable level of smoothing)
+    for shaper in shapers:
+        shaper_max_accel = round(shaper.max_accel / 100.0) * 100.0
+        label = '%s (%.1f Hz, vibr=%.1f%%, sm~=%.2f, accel<=%.f)' % (
+            shaper.name.upper(),
+            shaper.freq,
+            shaper.vibrs * 100.0,
+            shaper.smoothing,
+            shaper_max_accel,
+        )
+        ax2.plot(freqs, shaper.vals, label=label, linestyle='dotted')
+
+        # Get the performance shaper
+        if shaper.name == performance_shaper:
+            performance_shaper_freq = shaper.freq
+            performance_shaper_vibr = shaper.vibrs * 100.0
+            performance_shaper_vals = shaper.vals
+
+        # Get the low vibration shaper
+        if (
+            shaper.vibrs * 100 < lowvib_shaper_vibrs
+            or (shaper.vibrs * 100 == lowvib_shaper_vibrs and shaper_max_accel > lowvib_shaper_accel)
+        ) and shaper.smoothing < MAX_SMOOTHING:
+            lowvib_shaper_accel = shaper_max_accel
+            lowvib_shaper = shaper.name
+            lowvib_shaper_freq = shaper.freq
+            lowvib_shaper_vibrs = shaper.vibrs * 100
+            lowvib_shaper_vals = shaper.vals
+
+    # User recommendations are added to the legend: one is Klipper's original suggestion that is usually good for performances
+    # and the other one is the custom "low vibration" recommendation that looks for a suitable shaper that doesn't have excessive
+    # smoothing and that have a lower vibration level. If both recommendation are the same shaper, or if no suitable "low
+    # vibration" shaper is found, then only a single line as the "best shaper" recommendation is added to the legend
+    if (
+        lowvib_shaper is not None
+        and lowvib_shaper != performance_shaper
+        and lowvib_shaper_vibrs <= performance_shaper_vibr
+    ):
+        ax2.plot(
+            [],
+            [],
+            ' ',
+            label='Recommended performance shaper: %s @ %.1f Hz'
+            % (performance_shaper.upper(), performance_shaper_freq),
+        )
+        ax.plot(
+            freqs, psd * performance_shaper_vals, label='With %s applied' % (performance_shaper.upper()), color='cyan'
+        )
+        ax2.plot(
+            [],
+            [],
+            ' ',
+            label='Recommended low vibrations shaper: %s @ %.1f Hz' % (lowvib_shaper.upper(), lowvib_shaper_freq),
+        )
+        ax.plot(freqs, psd * lowvib_shaper_vals, label='With %s applied' % (lowvib_shaper.upper()), color='lime')
+    else:
+        ax2.plot(
+            [],
+            [],
+            ' ',
+            label='Recommended best shaper: %s @ %.1f Hz' % (performance_shaper.upper(), performance_shaper_freq),
+        )
+        ax.plot(
+            freqs, psd * performance_shaper_vals, label='With %s applied' % (performance_shaper.upper()), color='cyan'
+        )
+
+    # And the estimated damping ratio is finally added at the end of the legend
+    ax2.plot([], [], ' ', label='Estimated damping ratio (ζ): %.3f' % (zeta))
+
+    # Draw the detected peaks and name them
+    # This also draw the detection threshold and warning threshold (aka "effect zone")
+    ax.plot(peaks_freqs, psd[peaks], 'x', color='black', markersize=8)
+    for idx, peak in enumerate(peaks):
+        if psd[peak] > peaks_threshold[1]:
+            fontcolor = 'red'
+            fontweight = 'bold'
+        else:
+            fontcolor = 'black'
+            fontweight = 'normal'
+        ax.annotate(
+            f'{idx+1}',
+            (freqs[peak], psd[peak]),
+            textcoords='offset points',
+            xytext=(8, 5),
+            ha='left',
+            fontsize=13,
+            color=fontcolor,
+            weight=fontweight,
+        )
+    ax.axhline(y=peaks_threshold[0], color='black', linestyle='--', linewidth=0.5)
+    ax.axhline(y=peaks_threshold[1], color='black', linestyle='--', linewidth=0.5)
+    ax.fill_between(freqs, 0, peaks_threshold[0], color='green', alpha=0.15, label='Relax Region')
+    ax.fill_between(freqs, peaks_threshold[0], peaks_threshold[1], color='orange', alpha=0.2, label='Warning Region')
+
+    # Add the main resonant frequency and damping ratio of the axis to the graph title
+    ax.set_title(
+        'Axis Frequency Profile (ω0=%.1fHz, ζ=%.3f)' % (fr, zeta),
+        fontsize=14,
+        color=KLIPPAIN_COLORS['dark_orange'],
+        weight='bold',
+    )
+    ax.legend(loc='upper left', prop=fontP)
+    ax2.legend(loc='upper right', prop=fontP)
+
+    return
+
+
+# Plot a time-frequency spectrogram to see how the system respond over time during the
+# resonnance test. This can highlight hidden spots from the standard PSD graph from other harmonics
+def plot_spectrogram(ax, t, bins, pdata, peaks, max_freq):
+    ax.set_title('Time-Frequency Spectrogram', fontsize=14, color=KLIPPAIN_COLORS['dark_orange'], weight='bold')
+
+    # We need to normalize the data to get a proper signal on the spectrogram
+    # However, while using "LogNorm" provide too much background noise, using
+    # "Normalize" make only the resonnance appearing and hide interesting elements
+    # So we need to filter out the lower part of the data (ie. find the proper vmin for LogNorm)
+    vmin_value = np.percentile(pdata, SPECTROGRAM_LOW_PERCENTILE_FILTER)
+
+    # Draw the spectrogram using imgshow that is better suited here than pcolormesh since its result is already rasterized and
+    # we doesn't need to keep vector graphics when saving to a final .png file. Using it also allow to
+    # save ~150-200MB of RAM during the "fig.savefig" operation.
+    cm = 'inferno'
+    norm = matplotlib.colors.LogNorm(vmin=vmin_value)
+    ax.imshow(
+        pdata.T,
+        norm=norm,
+        cmap=cm,
+        aspect='auto',
+        extent=[t[0], t[-1], bins[0], bins[-1]],
+        origin='lower',
+        interpolation='antialiased',
+    )
+
+    ax.set_xlim([0.0, max_freq])
+    ax.set_ylabel('Time (s)')
+    ax.set_xlabel('Frequency (Hz)')
+
+    # Add peaks lines in the spectrogram to get hint from peaks found in the first graph
+    if peaks is not None:
+        for idx, peak in enumerate(peaks):
+            ax.axvline(peak, color='cyan', linestyle='dotted', linewidth=1)
+            ax.annotate(
+                f'Peak {idx+1}',
+                (peak, bins[-1] * 0.9),
+                textcoords='data',
+                color='cyan',
+                rotation=90,
+                fontsize=10,
+                verticalalignment='top',
+                horizontalalignment='right',
+            )
+
+    return
+
+
+######################################################################
+# Startup and main routines
+######################################################################
+
+
+def shaper_calibration(
+    lognames,
+    klipperdir='~/klipper',
+    max_smoothing=None,
+    scv=5.0,
+    max_freq=200.0,
+    accel_per_hz=None,
+    st_version='unknown',
+):
+    global shaper_calibrate
+    shaper_calibrate = setup_klipper_import(klipperdir)
+
+    # Parse data from the log files while ignoring CSV in the wrong format
+    datas = [data for data in (parse_log(fn) for fn in lognames) if data is not None]
+    if len(datas) > 1:
+        ConsoleOutput.print('Warning: incorrect number of .csv files detected. Only the first one will be used!')
+
+    # Compute shapers, PSD outputs and spectrogram
+    performance_shaper, shapers, calibration_data, fr, zeta, compat = calibrate_shaper(
+        datas[0], max_smoothing, scv, max_freq
+    )
+    pdata, bins, t = compute_spectrogram(datas[0])
+    del datas
+
+    # Select only the relevant part of the PSD data
+    freqs = calibration_data.freq_bins
+    calibration_data.psd_sum = calibration_data.psd_sum[freqs <= max_freq]
+    calibration_data.psd_x = calibration_data.psd_x[freqs <= max_freq]
+    calibration_data.psd_y = calibration_data.psd_y[freqs <= max_freq]
+    calibration_data.psd_z = calibration_data.psd_z[freqs <= max_freq]
+    calibration_data.freqs = freqs[freqs <= max_freq]
+
+    # Peak detection algorithm
+    peaks_threshold = [
+        PEAKS_DETECTION_THRESHOLD * calibration_data.psd_sum.max(),
+        PEAKS_EFFECT_THRESHOLD * calibration_data.psd_sum.max(),
+    ]
+    num_peaks, peaks, peaks_freqs = detect_peaks(calibration_data.psd_sum, calibration_data.freqs, peaks_threshold[0])
+
+    # Print the peaks info in the console
+    peak_freqs_formated = ['{:.1f}'.format(f) for f in peaks_freqs]
+    num_peaks_above_effect_threshold = np.sum(calibration_data.psd_sum[peaks] > peaks_threshold[1])
+    ConsoleOutput.print(
+        '\nPeaks detected on the graph: %d @ %s Hz (%d above effect threshold)'
+        % (num_peaks, ', '.join(map(str, peak_freqs_formated)), num_peaks_above_effect_threshold)
+    )
+
+    # Create graph layout
+    fig, (ax1, ax2) = plt.subplots(
+        2,
+        1,
+        gridspec_kw={
+            'height_ratios': [4, 3],
+            'bottom': 0.050,
+            'top': 0.890,
+            'left': 0.085,
+            'right': 0.966,
+            'hspace': 0.169,
+            'wspace': 0.200,
+        },
+    )
+    fig.set_size_inches(8.3, 11.6)
+
+    # Add a title with some test info
+    title_line1 = 'INPUT SHAPER CALIBRATION TOOL'
+    fig.text(
+        0.12, 0.965, title_line1, ha='left', va='bottom', fontsize=20, color=KLIPPAIN_COLORS['purple'], weight='bold'
+    )
+    try:
+        filename_parts = (lognames[0].split('/')[-1]).split('_')
+        dt = datetime.strptime(f'{filename_parts[1]} {filename_parts[2]}', '%Y%m%d %H%M%S')
+        title_line2 = dt.strftime('%x %X') + ' -- ' + filename_parts[3].upper().split('.')[0] + ' axis'
+        if compat:
+            title_line3 = '| Older Klipper version detected, damping ratio'
+            title_line4 = '| and SCV are not used for filter recommendations!'
+            title_line5 = f'| Accel per Hz used: {accel_per_hz} mm/s²/Hz' if accel_per_hz is not None else ''
+        else:
+            title_line3 = f'| Square corner velocity: {scv} mm/s'
+            title_line4 = f'| Max allowed smoothing: {max_smoothing}'
+            title_line5 = f'| Accel per Hz used: {accel_per_hz} mm/s²/Hz' if accel_per_hz is not None else ''
+    except Exception:
+        ConsoleOutput.print('Warning: CSV filename look to be different than expected (%s)' % (lognames[0]))
+        title_line2 = lognames[0].split('/')[-1]
+        title_line3 = ''
+        title_line4 = ''
+        title_line5 = ''
+    fig.text(0.12, 0.957, title_line2, ha='left', va='top', fontsize=16, color=KLIPPAIN_COLORS['dark_purple'])
+    fig.text(0.58, 0.965, title_line3, ha='left', va='top', fontsize=10, color=KLIPPAIN_COLORS['dark_purple'])
+    fig.text(0.58, 0.951, title_line4, ha='left', va='top', fontsize=10, color=KLIPPAIN_COLORS['dark_purple'])
+    fig.text(0.58, 0.919, title_line5, ha='left', va='top', fontsize=10, color=KLIPPAIN_COLORS['dark_purple'])
+
+    # Plot the graphs
+    plot_freq_response(
+        ax1, calibration_data, shapers, performance_shaper, peaks, peaks_freqs, peaks_threshold, fr, zeta, max_freq
+    )
+    plot_spectrogram(ax2, t, bins, pdata, peaks_freqs, max_freq)
+
+    # Adding a small Klippain logo to the top left corner of the figure
+    ax_logo = fig.add_axes([0.001, 0.8995, 0.1, 0.1], anchor='NW')
+    ax_logo.imshow(plt.imread(os.path.join(os.path.dirname(os.path.abspath(__file__)), 'klippain.png')))
+    ax_logo.axis('off')
+
+    # Adding Shake&Tune version in the top right corner
+    if st_version != 'unknown':
+        fig.text(0.995, 0.985, st_version, ha='right', va='bottom', fontsize=8, color=KLIPPAIN_COLORS['purple'])
+
+    return fig
+
+
+def main():
+    # Parse command-line arguments
+    usage = '%prog [options] <logs>'
+    opts = optparse.OptionParser(usage)
+    opts.add_option('-o', '--output', type='string', dest='output', default=None, help='filename of output graph')
+    opts.add_option('-f', '--max_freq', type='float', default=200.0, help='maximum frequency to graph')
+    opts.add_option('-s', '--max_smoothing', type='float', default=None, help='maximum shaper smoothing to allow')
+    opts.add_option(
+        '--scv', '--square_corner_velocity', type='float', dest='scv', default=5.0, help='square corner velocity'
+    )
+    opts.add_option('--accel_per_hz', type='float', default=None, help='accel_per_hz used during the measurement')
+    opts.add_option(
+        '-k', '--klipper_dir', type='string', dest='klipperdir', default='~/klipper', help='main klipper directory'
+    )
+    options, args = opts.parse_args()
+    if len(args) < 1:
+        opts.error('Incorrect number of arguments')
+    if options.output is None:
+        opts.error('You must specify an output file.png to use the script (option -o)')
+    if options.max_smoothing is not None and options.max_smoothing < 0.05:
+        opts.error('Too small max_smoothing specified (must be at least 0.05)')
+
+    fig = shaper_calibration(
+        args, options.klipperdir, options.max_smoothing, options.scv, options.max_freq, options.accel_per_hz, 'unknown'
+    )
+    fig.savefig(options.output, dpi=150)
+
+
+if __name__ == '__main__':
+    main()

shaketune/post_processing/graph_vibrations.py

@@ -0,0 +1,841 @@
+#!/usr/bin/env python3
+
+##################################################
+#### DIRECTIONAL VIBRATIONS PLOTTING SCRIPT ######
+##################################################
+# Written by Frix_x#0161 #
+
+import math
+import optparse
+import os
+import re
+from collections import defaultdict
+from datetime import datetime
+
+import matplotlib
+import matplotlib.font_manager
+import matplotlib.gridspec
+import matplotlib.pyplot as plt
+import matplotlib.ticker
+import numpy as np
+
+matplotlib.use('Agg')
+
+from ..helpers.common_func import (
+    compute_mechanical_parameters,
+    detect_peaks,
+    identify_low_energy_zones,
+    parse_log,
+    setup_klipper_import,
+)
+from ..helpers.console_output import ConsoleOutput
+
+PEAKS_DETECTION_THRESHOLD = 0.05
+PEAKS_RELATIVE_HEIGHT_THRESHOLD = 0.04
+CURVE_SIMILARITY_SIGMOID_K = 0.5
+SPEEDS_VALLEY_DETECTION_THRESHOLD = 0.7  # Lower is more sensitive
+SPEEDS_AROUND_PEAK_DELETION = 3  # to delete +-3mm/s around a peak
+ANGLES_VALLEY_DETECTION_THRESHOLD = 1.1  # Lower is more sensitive
+
+KLIPPAIN_COLORS = {
+    'purple': '#70088C',
+    'orange': '#FF8D32',
+    'dark_purple': '#150140',
+    'dark_orange': '#F24130',
+    'red_pink': '#F2055C',
+}
+
+
+######################################################################
+# Computation
+######################################################################
+
+
+# Call to the official Klipper input shaper object to do the PSD computation
+def calc_freq_response(data):
+    helper = shaper_calibrate.ShaperCalibrate(printer=None)
+    return helper.process_accelerometer_data(data)
+
+
+# Calculate motor frequency profiles based on the measured Power Spectral Density (PSD) measurements for the machine kinematics
+# main angles and then create a global motor profile as a weighted average (from their own vibrations) of all calculated profiles
+def compute_motor_profiles(freqs, psds, all_angles_energy, measured_angles=None, energy_amplification_factor=2):
+    if measured_angles is None:
+        measured_angles = [0, 90]
+
+    motor_profiles = {}
+    weighted_sum_profiles = np.zeros_like(freqs)
+    total_weight = 0
+    conv_filter = np.ones(20) / 20
+
+    # Creating the PSD motor profiles for each angles
+    for angle in measured_angles:
+        # Calculate the sum of PSDs for the current angle and then convolve
+        sum_curve = np.sum(np.array([psds[angle][speed] for speed in psds[angle]]), axis=0)
+        motor_profiles[angle] = np.convolve(sum_curve / len(psds[angle]), conv_filter, mode='same')
+
+        # Calculate weights
+        angle_energy = (
+            all_angles_energy[angle] ** energy_amplification_factor
+        )  # First weighting factor is based on the total vibrations of the machine at the specified angle
+        curve_area = (
+            np.trapz(motor_profiles[angle], freqs) ** energy_amplification_factor
+        )  # Additional weighting factor is based on the area under the current motor profile at this specified angle
+        total_angle_weight = angle_energy * curve_area
+
+        # Update weighted sum profiles to get the global motor profile
+        weighted_sum_profiles += motor_profiles[angle] * total_angle_weight
+        total_weight += total_angle_weight
+
+    # Creating a global average motor profile that is the weighted average of all the PSD motor profiles
+    global_motor_profile = weighted_sum_profiles / total_weight if total_weight != 0 else weighted_sum_profiles
+
+    return motor_profiles, global_motor_profile
+
+
+# Since it was discovered that there is no non-linear mixing in the stepper "steps" vibrations, instead of measuring
+# the effects of each speeds at each angles, this function simplify it by using only the main motors axes (X/Y for Cartesian
+# printers and A/B for CoreXY) measurements and project each points on the [0,360] degrees range using trigonometry
+# to "sum" the vibration impact of each axis at every points of the generated spectrogram. The result is very similar at the end.
+def compute_dir_speed_spectrogram(measured_speeds, data, kinematics='cartesian', measured_angles=None):
+    if measured_angles is None:
+        measured_angles = [0, 90]
+
+    # We want to project the motor vibrations measured on their own axes on the [0, 360] range
+    spectrum_angles = np.linspace(0, 360, 720)  # One point every 0.5 degrees
+    spectrum_speeds = np.linspace(min(measured_speeds), max(measured_speeds), len(measured_speeds) * 6)
+    spectrum_vibrations = np.zeros((len(spectrum_angles), len(spectrum_speeds)))
+
+    def get_interpolated_vibrations(data, speed, speeds):
+        idx = np.clip(np.searchsorted(speeds, speed, side='left'), 1, len(speeds) - 1)
+        lower_speed = speeds[idx - 1]
+        upper_speed = speeds[idx]
+        lower_vibrations = data.get(lower_speed, 0)
+        upper_vibrations = data.get(upper_speed, 0)
+        return lower_vibrations + (speed - lower_speed) * (upper_vibrations - lower_vibrations) / (
+            upper_speed - lower_speed
+        )
+
+    # Precompute trigonometric values and constant before the loop
+    angle_radians = np.deg2rad(spectrum_angles)
+    cos_vals = np.cos(angle_radians)
+    sin_vals = np.sin(angle_radians)
+    sqrt_2_inv = 1 / math.sqrt(2)
+
+    # Compute the spectrum vibrations for each angle and speed combination
+    for target_angle_idx, (cos_val, sin_val) in enumerate(zip(cos_vals, sin_vals)):
+        for target_speed_idx, target_speed in enumerate(spectrum_speeds):
+            if kinematics == 'cartesian':
+                speed_1 = np.abs(target_speed * cos_val)
+                speed_2 = np.abs(target_speed * sin_val)
+            elif kinematics == 'corexy':
+                speed_1 = np.abs(target_speed * (cos_val + sin_val) * sqrt_2_inv)
+                speed_2 = np.abs(target_speed * (cos_val - sin_val) * sqrt_2_inv)
+
+            vibrations_1 = get_interpolated_vibrations(data[measured_angles[0]], speed_1, measured_speeds)
+            vibrations_2 = get_interpolated_vibrations(data[measured_angles[1]], speed_2, measured_speeds)
+            spectrum_vibrations[target_angle_idx, target_speed_idx] = vibrations_1 + vibrations_2
+
+    return spectrum_angles, spectrum_speeds, spectrum_vibrations
+
+
+def compute_angle_powers(spectrogram_data):
+    angles_powers = np.trapz(spectrogram_data, axis=1)
+
+    # Since we want to plot it on a continuous polar plot later on, we need to append parts of
+    # the array to start and end of it to smooth transitions when doing the convolution
+    # and get the same value at modulo 360. Then we return the array without the extras
+    extended_angles_powers = np.concatenate([angles_powers[-9:], angles_powers, angles_powers[:9]])
+    convolved_extended = np.convolve(extended_angles_powers, np.ones(15) / 15, mode='same')
+
+    return convolved_extended[9:-9]
+
+
+def compute_speed_powers(spectrogram_data, smoothing_window=15):
+    min_values = np.amin(spectrogram_data, axis=0)
+    max_values = np.amax(spectrogram_data, axis=0)
+    var_values = np.var(spectrogram_data, axis=0)
+
+    # rescale the variance to the same range as max_values to plot it on the same graph
+    var_values = var_values / var_values.max() * max_values.max()
+
+    # Create a vibration metric that is the product of the max values and the variance to quantify the best
+    # speeds that have at the same time a low global energy level that is also consistent at every angles
+    vibration_metric = max_values * var_values
+
+    # utility function to pad and smooth the data avoiding edge effects
+    conv_filter = np.ones(smoothing_window) / smoothing_window
+    window = int(smoothing_window / 2)
+
+    def pad_and_smooth(data):
+        data_padded = np.pad(data, (window,), mode='edge')
+        smoothed_data = np.convolve(data_padded, conv_filter, mode='valid')
+        return smoothed_data
+
+    # Stack the arrays and apply padding and smoothing in batch
+    data_arrays = np.stack([min_values, max_values, var_values, vibration_metric])
+    smoothed_arrays = np.array([pad_and_smooth(data) for data in data_arrays])
+
+    return smoothed_arrays
+
+
+# Function that filter and split the good_speed ranges. The goal is to remove some zones around
+# additional detected small peaks in order to suppress them if there is a peak, even if it's low,
+# that's probably due to a crossing in the motor resonance pattern that still need to be removed
+def filter_and_split_ranges(all_speeds, good_speeds, peak_speed_indices, deletion_range):
+    # Process each range to filter out and split based on peak indices
+    filtered_good_speeds = []
+    for start, end, energy in good_speeds:
+        start_speed, end_speed = all_speeds[start], all_speeds[end]
+        # Identify peaks that intersect with the current speed range
+        intersecting_peaks_indices = [
+            idx for speed, idx in peak_speed_indices.items() if start_speed <= speed <= end_speed
+        ]
+
+        if not intersecting_peaks_indices:
+            filtered_good_speeds.append((start, end, energy))
+        else:
+            intersecting_peaks_indices.sort()
+            current_start = start
+
+            for peak_index in intersecting_peaks_indices:
+                before_peak_end = max(current_start, peak_index - deletion_range)
+                if current_start < before_peak_end:
+                    filtered_good_speeds.append((current_start, before_peak_end, energy))
+                current_start = peak_index + deletion_range + 1
+
+            if current_start < end:
+                filtered_good_speeds.append((current_start, end, energy))
+
+    # Sorting by start point once and then merge overlapping ranges
+    sorted_ranges = sorted(filtered_good_speeds, key=lambda x: x[0])
+    merged_ranges = [sorted_ranges[0]]
+
+    for current in sorted_ranges[1:]:
+        last_merged_start, last_merged_end, last_merged_energy = merged_ranges[-1]
+        if current[0] <= last_merged_end:
+            new_end = max(last_merged_end, current[1])
+            new_energy = min(last_merged_energy, current[2])
+            merged_ranges[-1] = (last_merged_start, new_end, new_energy)
+        else:
+            merged_ranges.append(current)
+
+    return merged_ranges
+
+
+# This function allow the computation of a symmetry score that reflect the spectrogram apparent symmetry between
+# measured axes on both the shape of the signal and the energy level consistency across both side of the signal
+def compute_symmetry_analysis(all_angles, spectrogram_data, measured_angles=None):
+    if measured_angles is None:
+        measured_angles = [0, 90]
+
+    total_spectrogram_angles = len(all_angles)
+    half_spectrogram_angles = total_spectrogram_angles // 2
+
+    # Extend the spectrogram by adding half to the beginning (in order to not get an out of bounds error later)
+    extended_spectrogram = np.concatenate((spectrogram_data[-half_spectrogram_angles:], spectrogram_data), axis=0)
+
+    # Calculate the split index directly within the slicing
+    midpoint_angle = np.mean(measured_angles)
+    split_index = int(midpoint_angle * (total_spectrogram_angles / 360) + half_spectrogram_angles)
+    half_segment_length = half_spectrogram_angles // 2
+
+    # Slice out the two segments of the spectrogram and flatten them for comparison
+    segment_1_flattened = extended_spectrogram[split_index - half_segment_length : split_index].flatten()
+    segment_2_flattened = extended_spectrogram[split_index : split_index + half_segment_length].flatten()
+
+    # Compute the correlation coefficient between the two segments of spectrogram
+    correlation = np.corrcoef(segment_1_flattened, segment_2_flattened)[0, 1]
+    percentage_correlation_biased = (100 * np.power(correlation, 0.75)) + 10
+
+    return np.clip(0, 100, percentage_correlation_biased)
+
+
+######################################################################
+# Graphing
+######################################################################
+
+
+def plot_angle_profile_polar(ax, angles, angles_powers, low_energy_zones, symmetry_factor):
+    angles_radians = np.deg2rad(angles)
+
+    ax.set_title('Polar angle energy profile', fontsize=14, color=KLIPPAIN_COLORS['dark_orange'], weight='bold')
+    ax.set_theta_zero_location('E')
+    ax.set_theta_direction(1)
+
+    ax.plot(angles_radians, angles_powers, color=KLIPPAIN_COLORS['purple'], zorder=5)
+    ax.fill(angles_radians, angles_powers, color=KLIPPAIN_COLORS['purple'], alpha=0.3)
+    ax.set_xlim([0, np.deg2rad(360)])
+    ymax = angles_powers.max() * 1.05
+    ax.set_ylim([0, ymax])
+    ax.set_thetagrids([theta * 15 for theta in range(360 // 15)])
+
+    ax.text(
+        0,
+        0,
+        f'Symmetry: {symmetry_factor:.1f}%',
+        ha='center',
+        va='center',
+        color=KLIPPAIN_COLORS['red_pink'],
+        fontsize=12,
+        fontweight='bold',
+        zorder=6,
+    )
+
+    for _, (start, end, _) in enumerate(low_energy_zones):
+        ax.axvline(
+            angles_radians[start],
+            angles_powers[start] / ymax,
+            color=KLIPPAIN_COLORS['red_pink'],
+            linestyle='dotted',
+            linewidth=1.5,
+        )
+        ax.axvline(
+            angles_radians[end],
+            angles_powers[end] / ymax,
+            color=KLIPPAIN_COLORS['red_pink'],
+            linestyle='dotted',
+            linewidth=1.5,
+        )
+        ax.fill_between(
+            angles_radians[start:end], angles_powers[start:end], angles_powers.max() * 1.05, color='green', alpha=0.2
+        )
+
+    ax.xaxis.set_minor_locator(matplotlib.ticker.AutoMinorLocator())
+    ax.yaxis.set_minor_locator(matplotlib.ticker.AutoMinorLocator())
+    ax.grid(which='major', color='grey')
+    ax.grid(which='minor', color='lightgrey')
+
+    # Polar plot doesn't follow the gridspec margin, so we adjust it manually here
+    pos = ax.get_position()
+    new_pos = [pos.x0 - 0.01, pos.y0 - 0.01, pos.width, pos.height]
+    ax.set_position(new_pos)
+
+    return
+
+
+def plot_global_speed_profile(
+    ax,
+    all_speeds,
+    sp_min_energy,
+    sp_max_energy,
+    sp_variance_energy,
+    vibration_metric,
+    num_peaks,
+    peaks,
+    low_energy_zones,
+):
+    ax.set_title('Global speed energy profile', fontsize=14, color=KLIPPAIN_COLORS['dark_orange'], weight='bold')
+    ax.set_xlabel('Speed (mm/s)')
+    ax.set_ylabel('Energy')
+    ax2 = ax.twinx()
+    ax2.yaxis.set_visible(False)
+
+    ax.plot(all_speeds, sp_min_energy, label='Minimum', color=KLIPPAIN_COLORS['dark_purple'], zorder=5)
+    ax.plot(all_speeds, sp_max_energy, label='Maximum', color=KLIPPAIN_COLORS['purple'], zorder=5)
+    ax.plot(all_speeds, sp_variance_energy, label='Variance', color=KLIPPAIN_COLORS['orange'], zorder=5, linestyle='--')
+    ax2.plot(
+        all_speeds,
+        vibration_metric,
+        label=f'Vibration metric ({num_peaks} bad peaks)',
+        color=KLIPPAIN_COLORS['red_pink'],
+        zorder=5,
+    )
+
+    ax.set_xlim([all_speeds.min(), all_speeds.max()])
+    ax.set_ylim([0, sp_max_energy.max() * 1.15])
+
+    y2min = -(vibration_metric.max() * 0.025)
+    y2max = vibration_metric.max() * 1.07
+    ax2.set_ylim([y2min, y2max])
+
+    if peaks is not None and len(peaks) > 0:
+        ax2.plot(all_speeds[peaks], vibration_metric[peaks], 'x', color='black', markersize=8, zorder=10)
+        for idx, peak in enumerate(peaks):
+            ax2.annotate(
+                f'{idx+1}',
+                (all_speeds[peak], vibration_metric[peak]),
+                textcoords='offset points',
+                xytext=(5, 5),
+                fontweight='bold',
+                ha='left',
+                fontsize=13,
+                color=KLIPPAIN_COLORS['red_pink'],
+                zorder=10,
+            )
+
+    for idx, (start, end, _) in enumerate(low_energy_zones):
+        # ax2.axvline(all_speeds[start], color=KLIPPAIN_COLORS['red_pink'], linestyle='dotted', linewidth=1.5, zorder=8)
+        # ax2.axvline(all_speeds[end], color=KLIPPAIN_COLORS['red_pink'], linestyle='dotted', linewidth=1.5, zorder=8)
+        ax2.fill_between(
+            all_speeds[start:end],
+            y2min,
+            vibration_metric[start:end],
+            color='green',
+            alpha=0.2,
+            label=f'Zone {idx+1}: {all_speeds[start]:.1f} to {all_speeds[end]:.1f} mm/s',
+        )
+
+    ax.xaxis.set_minor_locator(matplotlib.ticker.AutoMinorLocator())
+    ax.yaxis.set_minor_locator(matplotlib.ticker.AutoMinorLocator())
+    ax.grid(which='major', color='grey')
+    ax.grid(which='minor', color='lightgrey')
+
+    fontP = matplotlib.font_manager.FontProperties()
+    fontP.set_size('small')
+    ax.legend(loc='upper left', prop=fontP)
+    ax2.legend(loc='upper right', prop=fontP)
+
+    return
+
+
+def plot_angular_speed_profiles(ax, speeds, angles, spectrogram_data, kinematics='cartesian'):
+    ax.set_title('Angular speed energy profiles', fontsize=14, color=KLIPPAIN_COLORS['dark_orange'], weight='bold')
+    ax.set_xlabel('Speed (mm/s)')
+    ax.set_ylabel('Energy')
+
+    # Define mappings for labels and colors to simplify plotting commands
+    angle_settings = {
+        0: ('X (0 deg)', 'purple', 10),
+        90: ('Y (90 deg)', 'dark_purple', 5),
+        45: ('A (45 deg)' if kinematics == 'corexy' else '45 deg', 'orange', 10),
+        135: ('B (135 deg)' if kinematics == 'corexy' else '135 deg', 'dark_orange', 5),
+    }
+
+    # Plot each angle using settings from the dictionary
+    for angle, (label, color, zorder) in angle_settings.items():
+        idx = np.searchsorted(angles, angle, side='left')
+        ax.plot(speeds, spectrogram_data[idx], label=label, color=KLIPPAIN_COLORS[color], zorder=zorder)
+
+    ax.set_xlim([speeds.min(), speeds.max()])
+    max_value = max(spectrogram_data[angle].max() for angle in [0, 45, 90, 135])
+    ax.set_ylim([0, max_value * 1.1])
+
+    ax.xaxis.set_minor_locator(matplotlib.ticker.AutoMinorLocator())
+    ax.yaxis.set_minor_locator(matplotlib.ticker.AutoMinorLocator())
+    ax.grid(which='major', color='grey')
+    ax.grid(which='minor', color='lightgrey')
+
+    fontP = matplotlib.font_manager.FontProperties()
+    fontP.set_size('small')
+    ax.legend(loc='upper right', prop=fontP)
+
+    return
+
+
+def plot_motor_profiles(ax, freqs, main_angles, motor_profiles, global_motor_profile, max_freq):
+    ax.set_title('Motor frequency profile', fontsize=14, color=KLIPPAIN_COLORS['dark_orange'], weight='bold')
+    ax.set_ylabel('Energy')
+    ax.set_xlabel('Frequency (Hz)')
+
+    ax2 = ax.twinx()
+    ax2.yaxis.set_visible(False)
+
+    # Global weighted average motor profile
+    ax.plot(freqs, global_motor_profile, label='Combined', color=KLIPPAIN_COLORS['purple'], zorder=5)
+    max_value = global_motor_profile.max()
+
+    # Mapping of angles to axis names
+    angle_settings = {0: 'X', 90: 'Y', 45: 'A', 135: 'B'}
+
+    # And then plot the motor profiles at each measured angles
+    for angle in main_angles:
+        profile_max = motor_profiles[angle].max()
+        if profile_max > max_value:
+            max_value = profile_max
+        label = f'{angle_settings[angle]} ({angle} deg)' if angle in angle_settings else f'{angle} deg'
+        ax.plot(freqs, motor_profiles[angle], linestyle='--', label=label, zorder=2)
+
+    ax.set_xlim([0, max_freq])
+    ax.set_ylim([0, max_value * 1.1])
+    ax.ticklabel_format(axis='y', style='scientific', scilimits=(0, 0))
+
+    # Then add the motor resonance peak to the graph and print some infos about it
+    motor_fr, motor_zeta, motor_res_idx, lowfreq_max = compute_mechanical_parameters(global_motor_profile, freqs, 30)
+    if lowfreq_max:
+        ConsoleOutput.print(
+            '[WARNING] There are a lot of low frequency vibrations that can alter the readings. This is probably due to the test being performed at too high an acceleration!'
+        )
+        ConsoleOutput.print(
+            'Try lowering the ACCEL value and/or increasing the SIZE value before restarting the macro to ensure that only constant speeds are being recorded and that the dynamic behavior of the machine is not affecting the measurements'
+        )
+    if motor_zeta is not None:
+        ConsoleOutput.print(
+            'Motors have a main resonant frequency at %.1fHz with an estimated damping ratio of %.3f'
+            % (motor_fr, motor_zeta)
+        )
+    else:
+        ConsoleOutput.print(
+            'Motors have a main resonant frequency at %.1fHz but it was impossible to estimate a damping ratio.'
+            % (motor_fr)
+        )
+
+    ax.plot(freqs[motor_res_idx], global_motor_profile[motor_res_idx], 'x', color='black', markersize=10)
+    ax.annotate(
+        'R',
+        (freqs[motor_res_idx], global_motor_profile[motor_res_idx]),
+        textcoords='offset points',
+        xytext=(15, 5),
+        ha='right',
+        fontsize=14,
+        color=KLIPPAIN_COLORS['red_pink'],
+        weight='bold',
+    )
+
+    ax2.plot([], [], ' ', label='Motor resonant frequency (ω0): %.1fHz' % (motor_fr))
+    if motor_zeta is not None:
+        ax2.plot([], [], ' ', label='Motor damping ratio (ζ): %.3f' % (motor_zeta))
+    else:
+        ax2.plot([], [], ' ', label='No damping ratio computed')
+
+    ax.xaxis.set_minor_locator(matplotlib.ticker.AutoMinorLocator())
+    ax.yaxis.set_minor_locator(matplotlib.ticker.AutoMinorLocator())
+    ax.grid(which='major', color='grey')
+    ax.grid(which='minor', color='lightgrey')
+
+    fontP = matplotlib.font_manager.FontProperties()
+    fontP.set_size('small')
+    ax.legend(loc='upper left', prop=fontP)
+    ax2.legend(loc='upper right', prop=fontP)
+
+    return
+
+
+def plot_vibration_spectrogram_polar(ax, angles, speeds, spectrogram_data):
+    angles_radians = np.radians(angles)
+
+    # Assuming speeds defines the radial distance from the center, we need to create a meshgrid
+    # for both angles and speeds to map the spectrogram data onto a polar plot correctly
+    radius, theta = np.meshgrid(speeds, angles_radians)
+
+    ax.set_title(
+        'Polar vibrations heatmap', fontsize=14, color=KLIPPAIN_COLORS['dark_orange'], weight='bold', va='bottom'
+    )
+    ax.set_theta_zero_location('E')
+    ax.set_theta_direction(1)
+
+    ax.pcolormesh(theta, radius, spectrogram_data, norm=matplotlib.colors.LogNorm(), cmap='inferno', shading='auto')
+    ax.set_thetagrids([theta * 15 for theta in range(360 // 15)])
+    ax.tick_params(axis='y', which='both', colors='white', labelsize='medium')
+    ax.set_ylim([0, max(speeds)])
+
+    # Polar plot doesn't follow the gridspec margin, so we adjust it manually here
+    pos = ax.get_position()
+    new_pos = [pos.x0 - 0.01, pos.y0 - 0.01, pos.width, pos.height]
+    ax.set_position(new_pos)
+
+    return
+
+
+def plot_vibration_spectrogram(ax, angles, speeds, spectrogram_data, peaks):
+    ax.set_title('Vibrations heatmap', fontsize=14, color=KLIPPAIN_COLORS['dark_orange'], weight='bold')
+    ax.set_xlabel('Speed (mm/s)')
+    ax.set_ylabel('Angle (deg)')
+
+    ax.imshow(
+        spectrogram_data,
+        norm=matplotlib.colors.LogNorm(),
+        cmap='inferno',
+        aspect='auto',
+        extent=[speeds[0], speeds[-1], angles[0], angles[-1]],
+        origin='lower',
+        interpolation='antialiased',
+    )
+
+    # Add peaks lines in the spectrogram to get hint from peaks found in the first graph
+    if peaks is not None and len(peaks) > 0:
+        for idx, peak in enumerate(peaks):
+            ax.axvline(speeds[peak], color='cyan', linewidth=0.75)
+            ax.annotate(
+                f'Peak {idx+1}',
+                (speeds[peak], angles[-1] * 0.9),
+                textcoords='data',
+                color='cyan',
+                rotation=90,
+                fontsize=10,
+                verticalalignment='top',
+                horizontalalignment='right',
+            )
+
+    return
+
+
+def plot_motor_config_txt(fig, motors, differences):
+    motor_details = [(motors[0], 'X motor'), (motors[1], 'Y motor')]
+
+    distance = 0.12
+    if motors[0].get_config('autotune_enabled'):
+        distance = 0.27
+        config_blocks = [
+            f"| {lbl}: {mot.get_config('motor').upper()} on {mot.get_config('tmc').upper()} @ {mot.get_config('voltage'):0.1f}V {mot.get_config('run_current'):0.2f}A - {mot.get_config('microsteps')}usteps"
+            for mot, lbl in motor_details
+        ]
+        config_blocks.append('| TMC Autotune enabled')
+    else:
+        config_blocks = [
+            f"| {lbl}: {mot.get_config('tmc').upper()} @ {mot.get_config('run_current'):0.2f}A - {mot.get_config('microsteps')}usteps"
+            for mot, lbl in motor_details
+        ]
+        config_blocks.append('| TMC Autotune not detected')
+
+    for idx, block in enumerate(config_blocks):
+        fig.text(
+            0.41, 0.990 - 0.015 * idx, block, ha='left', va='top', fontsize=10, color=KLIPPAIN_COLORS['dark_purple']
+        )
+
+    tmc_registers = motors[0].get_registers()
+    idx = -1
+    for idx, (register, settings) in enumerate(tmc_registers.items()):
+        settings_str = ' '.join(f'{k}={v}' for k, v in settings.items())
+        tmc_block = f'| {register.upper()}: {settings_str}'
+        fig.text(
+            0.41 + distance,
+            0.990 - 0.015 * idx,
+            tmc_block,
+            ha='left',
+            va='top',
+            fontsize=10,
+            color=KLIPPAIN_COLORS['dark_purple'],
+        )
+
+    if differences is not None:
+        differences_text = f'| Y motor diff: {differences}'
+        fig.text(
+            0.41 + distance,
+            0.990 - 0.015 * (idx + 1),
+            differences_text,
+            ha='left',
+            va='top',
+            fontsize=10,
+            color=KLIPPAIN_COLORS['dark_purple'],
+        )
+
+
+######################################################################
+# Startup and main routines
+######################################################################
+
+
+def extract_angle_and_speed(logname):
+    try:
+        match = re.search(r'an(\d+)_\d+sp(\d+)_\d+', os.path.basename(logname))
+        if match:
+            angle = match.group(1)
+            speed = match.group(2)
+        else:
+            raise ValueError(f'File {logname} does not match expected format. Clean your /tmp folder and start again!')
+    except AttributeError as err:
+        raise ValueError(
+            f'File {logname} does not match expected format. Clean your /tmp folder and start again!'
+        ) from err
+    return float(angle), float(speed)
+
+
+def vibrations_profile(
+    lognames, klipperdir='~/klipper', kinematics='cartesian', accel=None, max_freq=1000.0, st_version=None, motors=None
+):
+    global shaper_calibrate
+    shaper_calibrate = setup_klipper_import(klipperdir)
+
+    if kinematics == 'cartesian':
+        main_angles = [0, 90]
+    elif kinematics == 'corexy':
+        main_angles = [45, 135]
+    else:
+        raise ValueError('Only Cartesian and CoreXY kinematics are supported by this tool at the moment!')
+
+    psds = defaultdict(lambda: defaultdict(list))
+    psds_sum = defaultdict(lambda: defaultdict(list))
+    target_freqs_initialized = False
+
+    for logname in lognames:
+        data = parse_log(logname)
+        if data is None:
+            continue  # File is not in the expected format, skip it
+        angle, speed = extract_angle_and_speed(logname)
+        freq_response = calc_freq_response(data)
+        first_freqs = freq_response.freq_bins
+        psd_sum = freq_response.psd_sum
+
+        if not target_freqs_initialized:
+            target_freqs = first_freqs[first_freqs <= max_freq]
+            target_freqs_initialized = True
+
+        psd_sum = psd_sum[first_freqs <= max_freq]
+        first_freqs = first_freqs[first_freqs <= max_freq]
+
+        # Store the interpolated PSD and integral values
+        psds[angle][speed] = np.interp(target_freqs, first_freqs, psd_sum)
+        psds_sum[angle][speed] = np.trapz(psd_sum, first_freqs)
+
+    measured_angles = sorted(psds_sum.keys())
+    measured_speeds = sorted({speed for angle_speeds in psds_sum.values() for speed in angle_speeds.keys()})
+
+    for main_angle in main_angles:
+        if main_angle not in measured_angles:
+            raise ValueError('Measurements not taken at the correct angles for the specified kinematics!')
+
+    # Precompute the variables used in plot functions
+    all_angles, all_speeds, spectrogram_data = compute_dir_speed_spectrogram(
+        measured_speeds, psds_sum, kinematics, main_angles
+    )
+    all_angles_energy = compute_angle_powers(spectrogram_data)
+    sp_min_energy, sp_max_energy, sp_variance_energy, vibration_metric = compute_speed_powers(spectrogram_data)
+    motor_profiles, global_motor_profile = compute_motor_profiles(target_freqs, psds, all_angles_energy, main_angles)
+
+    # symmetry_factor = compute_symmetry_analysis(all_angles, all_angles_energy)
+    symmetry_factor = compute_symmetry_analysis(all_angles, spectrogram_data, main_angles)
+    ConsoleOutput.print(f'Machine estimated vibration symmetry: {symmetry_factor:.1f}%')
+
+    # Analyze low variance ranges of vibration energy across all angles for each speed to identify clean speeds
+    # and highlight them. Also find the peaks to identify speeds to avoid due to high resonances
+    num_peaks, vibration_peaks, peaks_speeds = detect_peaks(
+        vibration_metric,
+        all_speeds,
+        PEAKS_DETECTION_THRESHOLD * vibration_metric.max(),
+        PEAKS_RELATIVE_HEIGHT_THRESHOLD,
+        10,
+        10,
+    )
+    formated_peaks_speeds = ['{:.1f}'.format(pspeed) for pspeed in peaks_speeds]
+    ConsoleOutput.print(
+        'Vibrations peaks detected: %d @ %s mm/s (avoid setting a speed near these values in your slicer print profile)'
+        % (num_peaks, ', '.join(map(str, formated_peaks_speeds)))
+    )
+
+    good_speeds = identify_low_energy_zones(vibration_metric, SPEEDS_VALLEY_DETECTION_THRESHOLD)
+    if good_speeds is not None:
+        deletion_range = int(SPEEDS_AROUND_PEAK_DELETION / (all_speeds[1] - all_speeds[0]))
+        peak_speed_indices = {pspeed: np.where(all_speeds == pspeed)[0][0] for pspeed in set(peaks_speeds)}
+
+        # Filter and split ranges based on peak indices, avoiding overlaps
+        good_speeds = filter_and_split_ranges(all_speeds, good_speeds, peak_speed_indices, deletion_range)
+
+        # Add some logging about the good speeds found
+        ConsoleOutput.print(f'Lowest vibrations speeds ({len(good_speeds)} ranges sorted from best to worse):')
+        for idx, (start, end, _) in enumerate(good_speeds):
+            ConsoleOutput.print(f'{idx+1}: {all_speeds[start]:.1f} to {all_speeds[end]:.1f} mm/s')
+
+    # Angle low energy valleys identification (good angles ranges) and print them to the console
+    good_angles = identify_low_energy_zones(all_angles_energy, ANGLES_VALLEY_DETECTION_THRESHOLD)
+    if good_angles is not None:
+        ConsoleOutput.print(f'Lowest vibrations angles ({len(good_angles)} ranges sorted from best to worse):')
+        for idx, (start, end, energy) in enumerate(good_angles):
+            ConsoleOutput.print(
+                f'{idx+1}: {all_angles[start]:.1f}° to {all_angles[end]:.1f}° (mean vibrations energy: {energy:.2f}% of max)'
+            )
+
+    # Create graph layout
+    fig, ((ax1, ax2, ax3), (ax4, ax5, ax6)) = plt.subplots(
+        2,
+        3,
+        gridspec_kw={
+            'height_ratios': [1, 1],
+            'width_ratios': [4, 8, 6],
+            'bottom': 0.050,
+            'top': 0.890,
+            'left': 0.040,
+            'right': 0.985,
+            'hspace': 0.166,
+            'wspace': 0.138,
+        },
+    )
+
+    # Transform ax3 and ax4 to polar plots
+    ax1.remove()
+    ax1 = fig.add_subplot(2, 3, 1, projection='polar')
+    ax4.remove()
+    ax4 = fig.add_subplot(2, 3, 4, projection='polar')
+
+    # Set the global .png figure size
+    fig.set_size_inches(20, 11.5)
+
+    # Add title
+    title_line1 = 'MACHINE VIBRATIONS ANALYSIS TOOL'
+    fig.text(
+        0.060, 0.965, title_line1, ha='left', va='bottom', fontsize=20, color=KLIPPAIN_COLORS['purple'], weight='bold'
+    )
+    try:
+        filename_parts = (lognames[0].split('/')[-1]).split('_')
+        dt = datetime.strptime(f"{filename_parts[1]} {filename_parts[2].split('-')[0]}", '%Y%m%d %H%M%S')
+        title_line2 = dt.strftime('%x %X')
+        if accel is not None:
+            title_line2 += ' at ' + str(accel) + ' mm/s² -- ' + kinematics.upper() + ' kinematics'
+    except Exception:
+        ConsoleOutput.print('Warning: CSV filenames appear to be different than expected (%s)' % (lognames[0]))
+        title_line2 = lognames[0].split('/')[-1]
+    fig.text(0.060, 0.957, title_line2, ha='left', va='top', fontsize=16, color=KLIPPAIN_COLORS['dark_purple'])
+
+    # Add the motors infos to the top of the graph
+    if motors is not None and len(motors) == 2:
+        differences = motors[0].compare_to(motors[1])
+        plot_motor_config_txt(fig, motors, differences)
+        if differences is not None and kinematics == 'corexy':
+            ConsoleOutput.print(f'Warning: motors have different TMC configurations!\n{differences}')
+
+    # Plot the graphs
+    plot_angle_profile_polar(ax1, all_angles, all_angles_energy, good_angles, symmetry_factor)
+    plot_vibration_spectrogram_polar(ax4, all_angles, all_speeds, spectrogram_data)
+
+    plot_global_speed_profile(
+        ax2,
+        all_speeds,
+        sp_min_energy,
+        sp_max_energy,
+        sp_variance_energy,
+        vibration_metric,
+        num_peaks,
+        vibration_peaks,
+        good_speeds,
+    )
+    plot_angular_speed_profiles(ax3, all_speeds, all_angles, spectrogram_data, kinematics)
+    plot_vibration_spectrogram(ax5, all_angles, all_speeds, spectrogram_data, vibration_peaks)
+
+    plot_motor_profiles(ax6, target_freqs, main_angles, motor_profiles, global_motor_profile, max_freq)
+
+    # Adding a small Klippain logo to the top left corner of the figure
+    ax_logo = fig.add_axes([0.001, 0.924, 0.075, 0.075], anchor='NW')
+    ax_logo.imshow(plt.imread(os.path.join(os.path.dirname(os.path.abspath(__file__)), 'klippain.png')))
+    ax_logo.axis('off')
+
+    # Adding Shake&Tune version in the top right corner
+    if st_version != 'unknown':
+        fig.text(0.995, 0.985, st_version, ha='right', va='bottom', fontsize=8, color=KLIPPAIN_COLORS['purple'])
+
+    return fig
+
+
+def main():
+    # Parse command-line arguments
+    usage = '%prog [options] <raw logs>'
+    opts = optparse.OptionParser(usage)
+    opts.add_option('-o', '--output', type='string', dest='output', default=None, help='filename of output graph')
+    opts.add_option(
+        '-c', '--accel', type='int', dest='accel', default=None, help='accel value to be printed on the graph'
+    )
+    opts.add_option('-f', '--max_freq', type='float', default=1000.0, help='maximum frequency to graph')
+    opts.add_option(
+        '-k', '--klipper_dir', type='string', dest='klipperdir', default='~/klipper', help='main klipper directory'
+    )
+    opts.add_option(
+        '-m',
+        '--kinematics',
+        type='string',
+        dest='kinematics',
+        default='cartesian',
+        help='machine kinematics configuration',
+    )
+    options, args = opts.parse_args()
+    if len(args) < 1:
+        opts.error('No CSV file(s) to analyse')
+    if options.output is None:
+        opts.error('You must specify an output file.png to use the script (option -o)')
+    if options.kinematics not in ['cartesian', 'corexy']:
+        opts.error('Only cartesian and corexy kinematics are supported by this tool at the moment!')
+
+    fig = vibrations_profile(args, options.klipperdir, options.kinematics, options.accel, options.max_freq)
+    fig.savefig(options.output, dpi=150)
+
+
+if __name__ == '__main__':
+    main()