[X3D-Ecosystem] Gaussian Splat data (from Aaron) negative scale data

[John Carlson]

It could be that my reader is bad.  I just assumed that stuff like:

I just assumed stuff like:

const f = view.getFloat32(offset, true); offset += 4;

And this would work:

  const header = new DataView(buffer, 0);
  const header_text = decoder.decode(new DataView(header.buffer, 0,
binaryStart));
  const view = new DataView(buffer, binaryStart);

On Tue, Feb 17, 2026 at 8:03 PM Bergstrom, Aaron <aaron.bergstrom at und.edu>
wrote:

> That could be, I didn’t look at the raw files. I saved what SuperSplat
> generated.
>
>
>
> However, I provided the same files to the AOUSD emerging geometries group,
> and the feedback was positive.
>
>
>
> Aaron
>
>
>
> *From:* John Carlson <yottzumm at gmail.com>
> *Sent:* Tuesday, February 17, 2026 7:59 PM
> *To:* X3D Ecosystem public discussion <x3d-ecosystem at web3d.org>;
> Interaction of AI with Web3D Technologies <ai at web3d.org>; x3d-ai at web3d.org;
> Anita Havele <anita.havele at web3d.org>; Bergstrom, Aaron <
> aaron.bergstrom at und.edu>
> *Subject:* Gaussian Splat data (from Aaron) negative scale data
>
>
>
> I looked again at the gaussian splat data from Aaron, and I discovered
> that there were negative scale values!' Ouch.
>
>
>
> That might mean that the file is corrupted, my reader is bad, or negative
> scales are normal,
>
>
>
> If anyone is looking at the data from the PLY files, can they confirm or
> deny?
>
> Thanks,
>
>
>
> John
>
>
>
> Don had a cool way of sorting Points into PointSets by scale, but if we
> can't get the right scale, that's going to be problematic.
>
>
>
> John
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://web3d.org/pipermail/x3d-ecosystem_web3d.org/attachments/20260217/29f8de2a/attachment-0001.html>