Title: [Bug]: FigureCanvasTkAgg renders clipped/oversized when embedded in layout-managed container on Windows HiDPI · Issue #31126 · matplotlib/matplotlib · GitHub
Open Graph Title: [Bug]: FigureCanvasTkAgg renders clipped/oversized when embedded in layout-managed container on Windows HiDPI · Issue #31126 · matplotlib/matplotlib
X Title: [Bug]: FigureCanvasTkAgg renders clipped/oversized when embedded in layout-managed container on Windows HiDPI · Issue #31126 · matplotlib/matplotlib
Description: Bug summary When FigureCanvasTkAgg is embedded in a layout-managed container (e.g., a Frame with pack(fill=BOTH, expand=True)), the plot is rendered larger than the visible canvas area on Windows with HiDPI display scaling (>100%). The r...
Open Graph Description: Bug summary When FigureCanvasTkAgg is embedded in a layout-managed container (e.g., a Frame with pack(fill=BOTH, expand=True)), the plot is rendered larger than the visible canvas area on Windows w...
X Description: Bug summary When FigureCanvasTkAgg is embedded in a layout-managed container (e.g., a Frame with pack(fill=BOTH, expand=True)), the plot is rendered larger than the visible canvas area on Windows w...
Opengraph URL: https://github.com/matplotlib/matplotlib/issues/31126
X: @github
Domain: github.com
{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"[Bug]: FigureCanvasTkAgg renders clipped/oversized when embedded in layout-managed container on Windows HiDPI","articleBody":"### Bug summary\n\nWhen `FigureCanvasTkAgg` is embedded in a layout-managed container (e.g., a `Frame` with `pack(fill=BOTH, expand=True)`), the plot is rendered larger than the visible canvas area on Windows with HiDPI display scaling (\u003e100%). The right and bottom portions of the figure are clipped/cropped.\n\nThe root cause is in `FigureCanvasTk._update_device_pixel_ratio()`: it updates `figure.dpi` via `_set_device_pixel_ratio()` and then calls `self._tkcanvas.configure(width=physical_w, height=physical_h)`. When the canvas is constrained by a geometry manager (`pack`/`grid`), the actual displayed size does not change, so `\u003cConfigure\u003e` does not fire, and `resize()` is never called to recalculate `figure.size_inches` with the new DPI. This results in a render buffer that is `device_pixel_ratio` times larger than the visible area.\n\n### Code for reproduction\n\n```Python\n\"\"\"\nRun on Windows with display scaling \u003e 100%.\nExpected: Plot fills the visible canvas area correctly.\nActual: Plot is rendered larger than visible area; right/bottom portions are clipped.\n\"\"\"\nimport ctypes\nimport tkinter as tk\nimport numpy as np\nfrom matplotlib.figure import Figure\nfrom matplotlib.backends.backend_tkagg import FigureCanvasTkAgg\n\n# Standard Windows HiDPI setup\nctypes.windll.shcore.SetProcessDpiAwareness(1)\n\nroot = tk.Tk()\nroot.geometry(\"800x600\")\n\nframe = tk.Frame(root)\nframe.pack(fill=tk.BOTH, expand=True)\n\nfig = Figure(dpi=96)\nax = fig.add_subplot(111)\nx = np.linspace(0, 2 * np.pi, 100)\nax.plot(x, np.sin(x), label=\"sin(x)\")\nax.plot(x, np.cos(x), label=\"cos(x)\")\nax.set_xlabel(\"X Axis - may be clipped\")\nax.set_ylabel(\"Y Axis - may be clipped\")\nax.set_title(\"Embedded FigureCanvasTkAgg - HiDPI clipping bug\")\nax.legend()\nax.grid(True)\nfig.tight_layout()\n\ncanvas = FigureCanvasTkAgg(fig, master=frame)\ncanvas.get_tk_widget().pack(fill=tk.BOTH, expand=True)\n\n# Diagnostic output\ndef diagnostics(event=None):\n w = canvas.get_tk_widget()\n cfg_w, cfg_h = int(w[\"width\"]), int(w[\"height\"])\n act_w, act_h = w.winfo_width(), w.winfo_height()\n sz = fig.get_size_inches()\n render_w, render_h = int(sz[0] * fig.dpi), int(sz[1] * fig.dpi)\n print(f\"device_pixel_ratio: {canvas.device_pixel_ratio}\")\n print(f\"figure.dpi: {fig.dpi} (original: {fig._original_dpi})\")\n print(f\"figure.get_size_inches(): [{sz[0]:.2f}, {sz[1]:.2f}]\")\n print(f\"Render size: {render_w}x{render_h}\")\n print(f\"Canvas configured: {cfg_w}x{cfg_h}\")\n print(f\"Canvas actual: {act_w}x{act_h}\")\n if render_w != act_w or render_h != act_h:\n print(f\"BUG: render size ({render_w}x{render_h}) != \"\n f\"actual size ({act_w}x{act_h})\")\n\nroot.after(500, diagnostics)\ncanvas.draw()\nroot.mainloop()\n```\n\n### Actual outcome\n\nOn Windows 11 with 150% display scaling:\n\n```\ndevice_pixel_ratio: 1.5\nfigure.dpi: 144.0 (original: 96)\nfigure.get_size_inches(): [8.33, 6.25]\nRender size: 1200x900\nCanvas configured: 1200x900\nCanvas actual: 800x600\nBUG: render size (1200x900) != actual size (800x600)\n```\n\nThe plot is rendered at 1200×900 pixels but only 800×600 pixels are visible. The bottom and right portions (axis labels, legend, etc.) are cropped.\n\nNote: this also reproduces **without** manually setting `tk scaling` — just `SetProcessDpiAwareness(1)` is sufficient, because Tk automatically adjusts its scaling factor on HiDPI displays, and `_update_device_pixel_ratio` reads it.\n\n### Expected outcome\n\nThe plot should be fully visible within the canvas area with no clipping, regardless of the display scaling factor. The render size should match the actual displayed size.\n\n### Additional information\n\nThe chain of events:\n\n1. Canvas is packed → `\u003cConfigure\u003e` fires with actual size (800×600) → `resize()` sets `figure.size_inches = (800/96, 600/96) = (8.33, 6.25)`\n2. Canvas becomes visible → `\u003cMap\u003e` fires → `_update_device_pixel_ratio()` runs\n3. `_set_device_pixel_ratio(1.5)` changes `figure.dpi` from 96 to 144\n4. `configure(width=1200, height=900)` sets the canvas **requested** size\n5. `pack(fill=BOTH, expand=True)` constrains the **actual** size to 800×600\n6. Since actual size didn't change, `\u003cConfigure\u003e` does NOT fire\n7. `figure.size_inches` is still `(8.33, 6.25)` (calculated with the old dpi=96)\n8. Render size = `size_inches × new_dpi = (8.33×144, 6.25×144) = (1200, 900)`\n9. But only 800×600 is visible → **right and bottom portions are clipped**\n\nThis works correctly in `FigureManagerTk` (matplotlib's own window) because the canvas is packed directly in the top-level window, and `configure()` causes the window to resize, which triggers `\u003cConfigure\u003e` → `resize()`. But when the canvas is embedded in a user-managed layout, the geometry manager constrains the size and `\u003cConfigure\u003e` may not fire.\n\n### Suggested fix \n\nAfter `_set_device_pixel_ratio` changes the DPI and `configure` sets the new requested size, `_update_device_pixel_ratio` should ensure `resize()` is called even if `\u003cConfigure\u003e` doesn't fire. For example:\n\n```python\ndef _update_device_pixel_ratio(self, event=None):\n ratio = None\n if sys.platform == 'win32':\n ratio = round(self._tkcanvas.tk.call('tk', 'scaling') / (96 / 72), 2)\n elif sys.platform == \"linux\":\n ratio = self._tkcanvas.winfo_fpixels('1i') / 96\n if ratio is not None and self._set_device_pixel_ratio(ratio):\n w, h = self.get_width_height(physical=True)\n self._tkcanvas.configure(width=w, height=h)\n # If the actual displayed size is constrained by a layout manager\n # and didn't change, \u003cConfigure\u003e won't fire and resize() won't run.\n # Force a resize to recalculate figure.size_inches with the new DPI.\n self._tkcanvas.update_idletasks()\n actual_w = self._tkcanvas.winfo_width()\n actual_h = self._tkcanvas.winfo_height()\n if actual_w \u003e 0 and actual_h \u003e 0 and (actual_w != w or actual_h != h):\n self.resize(type('Event', (), {'width': actual_w, 'height': actual_h})())\n```\n\n### Operating system\n\nWindows 11 (10.0.26200)\n\n### Matplotlib Version\n\n3.10.7 \n\n### Matplotlib Backend\n\nTkAgg\n\n### Python version\n\n3.12.12\n\n### Jupyter version\n\n_No response_\n\n### Installation\n\nconda","author":{"url":"https://github.com/lzz8246","@type":"Person","name":"lzz8246"},"datePublished":"2026-02-10T06:53:11.000Z","interactionStatistic":{"@type":"InteractionCounter","interactionType":"https://schema.org/CommentAction","userInteractionCount":5},"url":"https://github.com/31126/matplotlib/issues/31126"}
| route-pattern | /_view_fragments/issues/show/:user_id/:repository/:id/issue_layout(.:format) |
| route-controller | voltron_issues_fragments |
| route-action | issue_layout |
| fetch-nonce | v2:66d04240-e1c7-7ac2-b432-40cbcbce1ee6 |
| current-catalog-service-hash | 81bb79d38c15960b92d99bca9288a9108c7a47b18f2423d0f6438c5b7bcd2114 |
| request-id | ED06:238007:110D43A:16CE582:6A51C5E8 |
| html-safe-nonce | 5583724c7769e4050e8b7d4ab83103d7c3287406ac193e2c886bb737d5ff7b8c |
| visitor-payload | eyJyZWZlcnJlciI6IiIsInJlcXVlc3RfaWQiOiJFRDA2OjIzODAwNzoxMTBENDNBOjE2Q0U1ODI6NkE1MUM1RTgiLCJ2aXNpdG9yX2lkIjoiNzMyNTYwOTIwNTI1MTAzMjU1MiIsInJlZ2lvbl9lZGdlIjoiaWFkIiwicmVnaW9uX3JlbmRlciI6ImlhZCJ9 |
| visitor-hmac | 9871d1c6818368e8e16f8e858628bec62579d516a940e62603d76e43136ef004 |
| hovercard-subject-tag | issue:3919775506 |
| github-keyboard-shortcuts | repository,issues,copilot |
| google-site-verification | Apib7-x98H0j5cPqHWwSMm6dNU4GmODRoqxLiDzdx9I |
| octolytics-url | https://collector.github.com/github/collect |
| analytics-location | / |
| fb:app_id | 1401488693436528 |
| apple-itunes-app | app-id=1477376905, app-argument=https://github.com/_view_fragments/issues/show/matplotlib/matplotlib/31126/issue_layout |
| twitter:image | https://opengraph.githubassets.com/e09c9297f76597c1db2ed200fdd9e25f7cb9a1113b08da91a353c6ca9af3e62f/matplotlib/matplotlib/issues/31126 |
| twitter:card | summary_large_image |
| og:image | https://opengraph.githubassets.com/e09c9297f76597c1db2ed200fdd9e25f7cb9a1113b08da91a353c6ca9af3e62f/matplotlib/matplotlib/issues/31126 |
| og:image:alt | Bug summary When FigureCanvasTkAgg is embedded in a layout-managed container (e.g., a Frame with pack(fill=BOTH, expand=True)), the plot is rendered larger than the visible canvas area on Windows w... |
| og:image:width | 1200 |
| og:image:height | 600 |
| og:site_name | GitHub |
| og:type | object |
| og:author:username | lzz8246 |
| hostname | github.com |
| expected-hostname | github.com |
| None | b9a586c06a05a7a86fc7e3f4dbd03e42f6869085879aa184aa6369456dbd50fb |
| turbo-cache-control | no-preview |
| go-import | github.com/matplotlib/matplotlib git https://github.com/matplotlib/matplotlib.git |
| octolytics-dimension-user_id | 215947 |
| octolytics-dimension-user_login | matplotlib |
| octolytics-dimension-repository_id | 1385122 |
| octolytics-dimension-repository_nwo | matplotlib/matplotlib |
| octolytics-dimension-repository_public | true |
| octolytics-dimension-repository_is_fork | false |
| octolytics-dimension-repository_network_root_id | 1385122 |
| octolytics-dimension-repository_network_root_nwo | matplotlib/matplotlib |
| turbo-body-classes | logged-out env-production page-responsive |
| disable-turbo | false |
| browser-stats-url | https://api.github.com/_private/browser/stats |
| browser-errors-url | https://api.github.com/_private/browser/errors |
| release | 7aed05249554b889eb33d002851a973eebcc7e91 |
| ui-target | full |
| theme-color | #1e2327 |
| color-scheme | light dark |
Links:
Viewport: width=device-width