GMaps, OverlayView, GTiff - Image is shifted/squished the further N latitude you go

I have a 3857 GTiff (no difference if I use a 4326 version) that correctly produces tiles (we use TiTiler), but if I try cropping that image and serve a single image that I try overlaying with OverlayView (like the USGS example) it seems to shift North and/or "squishes" the image to fit within the div.

Here's a fiddle I made, https://jsfiddle.net/m6pvLe0d/1/, using the exact USGS source from the Google doc's and if you toggle the image, you can see where the Great Lakes should be, and where they are being put due to the shift/crop issue.

/**
 * @license
 * Copyright 2019 Google LLC. All Rights Reserved.
 * SPDX-License-Identifier: Apache-2.0
 */
// This example adds hide() and show() methods to a custom overlay's prototype.
// These methods toggle the visibility of the container <div>.
// overlay to or from the map.
function initMap() {
  const map = new google.maps.Map(document.getElementById("map"), {
    zoom: 5,
    center: { lat: 41.3139, lng: -75.77767 },
    mapTypeId: "satellite",
  });
  const bounds = new google.maps.LatLngBounds(
    new google.maps.LatLng(25.76400806673997, -117.59163484375),
    new google.maps.LatLng(53.88085455391184, -33.96370515625001)
  );
  // The photograph is courtesy of the U.S. Geological Survey.
  let image = "https://i.imgur.com/la39KvI.png";
  /**
   * The custom USGSOverlay object contains the USGS image,
   * the bounds of the image, and a reference to the map.
   */
  class USGSOverlay extends google.maps.OverlayView {
    bounds;
    image;
    div;
    constructor(bounds, image) {
      super();
      this.bounds = bounds;
      this.image = image;
    }
    /**
     * onAdd is called when the map's panes are ready and the overlay has been
     * added to the map.
     */
    onAdd() {
      this.div = document.createElement("div");
      this.div.style.borderStyle = "none";
      this.div.style.borderWidth = "0px";
      this.div.style.position = "absolute";

      // Create the img element and attach it to the div.
      const img = document.createElement("img");

      img.src = this.image;
      img.style.width = "100%";
      img.style.height = "100%";
      img.style.position = "absolute";
      this.div.appendChild(img);

      // Add the element to the "overlayLayer" pane.
      const panes = this.getPanes();

      panes.overlayLayer.appendChild(this.div);
    }
    draw() {
      // We use the south-west and north-east
      // coordinates of the overlay to peg it to the correct position and size.
      // To do this, we need to retrieve the projection from the overlay.
      const overlayProjection = this.getProjection();
      // Retrieve the south-west and north-east coordinates of this overlay
      // in LatLngs and convert them to pixel coordinates.
      // We'll use these coordinates to resize the div.
      const sw = overlayProjection.fromLatLngToDivPixel(
        this.bounds.getSouthWest()
      );
      const ne = overlayProjection.fromLatLngToDivPixel(
        this.bounds.getNorthEast()
      );

      // Resize the image's div to fit the indicated dimensions.
      if (this.div) {
        this.div.style.left = sw.x + "px";
        this.div.style.top = ne.y + "px";
        this.div.style.width = ne.x - sw.x + "px";
        this.div.style.height = sw.y - ne.y + "px";
      }
    }
    /**
     * The onRemove() method will be called automatically from the API if
     * we ever set the overlay's map property to 'null'.
     */
    onRemove() {
      if (this.div) {
        this.div.parentNode.removeChild(this.div);
        delete this.div;
      }
    }
    /**
     *  Set the visibility to 'hidden' or 'visible'.
     */
    hide() {
      if (this.div) {
        this.div.style.visibility = "hidden";
      }
    }
    show() {
      if (this.div) {
        this.div.style.visibility = "visible";
      }
    }
    toggle() {
      if (this.div) {
        if (this.div.style.visibility === "hidden") {
          this.show();
        } else {
          this.hide();
        }
      }
    }
    toggleDOM(map) {
      if (this.getMap()) {
        this.setMap(null);
      } else {
        this.setMap(map);
      }
    }
  }

  const overlay = new USGSOverlay(bounds, image);

  overlay.setMap(map);

  const toggleButton = document.createElement("button");

  toggleButton.textContent = "Toggle";
  toggleButton.classList.add("custom-map-control-button");

  const toggleDOMButton = document.createElement("button");

  toggleDOMButton.textContent = "Toggle DOM Attachment";
  toggleDOMButton.classList.add("custom-map-control-button");
  toggleButton.addEventListener("click", () => {
    overlay.toggle();
  });
  toggleDOMButton.addEventListener("click", () => {
    overlay.toggleDOM(map);
  });
  map.controls[google.maps.ControlPosition.TOP_RIGHT].push(toggleDOMButton);
  map.controls[google.maps.ControlPosition.TOP_RIGHT].push(toggleButton);
}

window.initMap = initMap;

I'm really trying to understand "where" in the process this would be happening, as the same underlying GTiff that produced that large image correctly makes tiles. The cut-off on the Western side of the United States is expected as that was outside of the bounding box (lat/lon) I used to pass to TiTiler.

I know there's presumably "a lot" that can go into exactly where in the stack it's causing this to happen, but being very fresh with mapping, I don't even really know where to start looking and I guess the real head-scratcher for me is how it correctly produces tiles.

from GMaps, OverlayView, GTiff - Image is shifted/squished the further N latitude you go