How to accurately scrollto with <ScrollControls>?

[czrest]

I have a navigation menu that scrolls to specific sections of my page

It does scroll down/up but the offset is not accurate because of the props from my scroll control

Here's what I've done so far

<ScrollControls pages={6} distance={2} damping={0.5} maxSpeed={0.8}>
  <SheetProvider sheet={sheet}>
    <SectionHtml />
  </SheetProvider>
</ScrollControls>

const SectionHtml = () => {
  const scroll = useScroll();

  const test1 = useRef(null);
  const test2 = useRef(null);

  const scrollToSection = (elementRef) => {
    scroll.el.scrollTo({
      top: elementRef.current.offsetTop,
      behavior: "smooth",
    });
  };
  return (
    <>
      <Header>
        <NavigationMenu onClick={()=>scrollToSection(test1)}>test1</NavigationMenu>
        <NavigationMenu onClick={()=>scrollToSection(test2)}>test2</NavigationMenu>
      </Header>

      <Scroll html>
        <section ref={test1}> test1 </section>
        <section ref={test2}> test2 </section>
      </Scroll>
    </>
  );
}

The elementRef.current.offsetTop lacks a formula I think it needs the offset for the scroll controls but I don't know how to get it

[czrest]

I tried this method

elementRef.current.scrollIntoView({
      behavior: "smooth",
      block: "start"
    })

screen-recording.webm

But the problem is that the scroll bar/scrollcontrol doesn't move, only the element from < Scroll html > I think. That is why when I click the button I cannot go up and also the animation for my gltf doesn't activate

Any suggestions, advice, or solution is highly appreciated. Thanks.

[czrest]

For now I here's my fix

scroll.el.scrollTo({
      top: elementRef.current.offsetTop * 2.4,
      behavior: "smooth",
    });

I went back to my initial attempt and simply multiplied it by 2.4.

It works quite well, though I think there is a much better and accurate solution for it since It only works with the current settings of my scroll control. If you make any adjustments to the scroll control, the scrolling becomes inaccurate again.