I’m creating a program, which places circles on a fixed graph based on user input, and after that the user has the option to generate a PDF from that figure. However, if I try to save the figure as PDF, the figure never keeps it’s exact physical size, but it is very important. For example, if my math is right, the Y axis should be always 100mm long, as you can see in the adjust_graph_X() function. The user has the option to relength the X axis, thats why we have this adjust function. After generating the PDF, the Y axis is never 100mm on the PDF, it is bigger, or smaller, depending on the X axis length, and the PDF autoscale or I do not know.

import matplotlib.pyplot as plt from matplotlib.backends.backend_pdf import PdfPages import tkinter as tk from matplotlib.backends.backend_tkagg import FigureCanvasTkAgg from matplotlib.figure import Figure

fig, ax = plt.subplots(figsize=(210/25.4, 297/25.4), dpi=96) # Define PDF size in mm (A4) 210x297mm fig.suptitle(f"Enter name", fontsize=14)

#Create canvas to display the figure canvas = FigureCanvasTkAgg(fig, master=window) canvas.get_tk_widget().pack(side="right", expand=False, fill="none")

`def adjust_graph_x():

   global pixels_per_1mm
   global graph_diameter
   global x_axis_length_pixels
   global y_axis_length_pixels
   pixels_per_1mm = window.winfo_fpixels('1m') # Return the number of pixels for 1mm distance
   graph_diameter = float(entry_graft_diameter.get()) #user input based, must adjust the x axis length
   x_axis_length_mm = graph_diameter * math.pi #d*pi circle area
   x_axis_length_pixels = pixels_per_1mm * x_axis_length_mm
   #x_axis_length_pixels = (x_axis_length_mm * 96) / 25.4 #96 dpi
   dpi = pixels_per_1mm * 25.4
   print(dpi)
   global pixel_distance_x
   pixel_distance_x = x_axis_length_pixels / 720  #720 values between 6h AM and 6h PM, evenly divided 
   y_axis_length_mm = 100
   y_axis_length_pixels = pixels_per_1mm * y_axis_length_mm

   ax.clear()
   # Set X-axis limits and aspect ratio
   ax.set_xlim(0, x_axis_length_pixels)
   ax.set_ylim(y_axis_length_pixels, 0)

   # Create custom X-axis ticks and labels at specific hour intervals
   hours = ['6h', '7h', '8h', '9h', '10h', '11h', '12h', '1h', '2h', '3h', '4h', '5h', '6h']
   tick_positions = [i * (pixel_distance_x * 60) for i in range(len(hours))]  
   ax.set_xticks(tick_positions)  # Set tick positions
   ax.set_xticklabels(hours)  # Set labels 

   # Create custom Y-axis ticks and labels at specific intervals (10mm)
   distances = ['0mm', '10mm', '20mm', '30mm', '40mm', '50mm', '60mm', '70mm', '80mm', '90mm', '100mm']
   y_ticks = [i * (pixels_per_1mm*10) for i in range(11)]  # placing distance labels with 10mm step
   ax.set_yticks(y_ticks)  # Set y-axis tick positions
   ax.set_yticklabels(distances)  # Set labels 
   ax.grid(True, which='both', linestyle='--', linewidth=0.5)
   # Draw a thicker vertical line at 12h
   twelve_hour_position = tick_positions[6]  # Assuming '12h' is at index 6
   ax.axvline(twelve_hour_position, color='black', linewidth=2, linestyle='--')  

   ax.axis('on')
   canvas.draw()
   draw_circles_and_show_preview(circle_data)`

For example, the graft_diameter_entry() is 25mm, with that we get X axis length 78,54mm , and let’s say the user draws 1 circle like this:

After that we generate the PDF, and as you can see with Adobe PDF measuring tool, the Y axis is nearly 120mm instead of fixed 100:

I have tried a lot of PDF generation methods, including:

 `pp = PdfPages(filename)
  pp.savefig(fig, dpi=96)
  pp.close()`

or

 `canvas.print_figure(filename, dpi=96)`

or

`width=fig.get_figwidth()
 height=fig.get_figheight()
 fig.set_size_inches(width, height)
 fig.set_dpi(96)
 fig.savefig(filename, dpi=96)`

I also tried saving the figure to image, then reload it and adjust it accordingly:

`canvas2 = Canvas(filename, pagesize=A4)
 canvas2.drawImage("my_figure.png",0,0,width=x_axis_length_pixels, height=y_axis_length_pixels,      preserveAspectRatio=True)
 canvas2.showPage()
 canvas2.save()`

My new, and last Idea is that i’ll create a new canvas with

`from reportlab.pdfgen.canvas import Canvas`

and with that I will redraw the whole thing with lines and circles like this:

`canvas2.circle(1 * mm, 9 * mm, 0.5 * mm, stroke=1, fill=0) etc....`

because i’m getting done with the simple PDF saver functions. What am I doing wrong? All of the generation methods optimize my figure someway, so I cannot get the figure size which I draw on the monitor.