rerun VS msdfgen

Maybe the folks at Rerun [1] know something about it? I imagine at least some of their customers are Rust robotics shops.

[1] https://github.com/rerun-io/rerun

Full release notes: https://github.com/rerun-io/rerun/blob/main/CHANGELOG.md

No specific support for audio embeddings yet, though depending on you're specific use-case, you might be able to map your data to more generic primitives such a TimeSeriesScalar, LineStrips2D, BarChart, or Tensor.

We have an open issue for adding more support for audio-based projects -- if you have specific needs feel free to add a comment: https://github.com/rerun-io/rerun/issues/2852

Read the full changelog at https://github.com/rerun-io/rerun/releases/tag/0.6.0

https://github.com/rerun-io/rerun is such a great tool for creating interactive dashboards to look at real-time data. Curious if there are others looking for something like this and what the other use cases would be interesting. I have heard ideas like:

egui's lead author is using it for his computer vision startup.

To test the power draw, I conducted an unscientific experiment by circling the cursor over both a GTK4 app (GNOME Console) and an egui app (Rerun Viewer), while monitoring the CPU/GPU utilization and stabilized power draw with PowerTOP. The results indicated that when circling the cursor over the GTK4 app, the CPU/GPU usage was nearly negligible, and the stabilized power draw was around 13 W. In contrast, when circling the cursor over the egui app, the CPU/GPU usage was considerably higher, resulting in a stabilized power draw of around 21 W. This amounts to an increase of approximately 60%. Taking into account a power draw of around 10 W when doing nothing at all, the increase becomes roughly 260%. I also tried the same experiment with Iced, and while the increase was not as extreme, it still utilized resources when circling the cursor over a blank surface.

This very well explained here https://github.com/Chlumsky/msdfgen and with more details in link d pdf.

Basically, signed distance fields allow high resolution renders from low resolution rasters which represent character shape.

No idea how to help you but I will just drop this since it improved the quality for me by 1000 https://github.com/Chlumsky/msdfgen

And that's the complicated state of the art version for 3D perspective. Other versions are even simpler.

This is known as a “multi-channel signed distance field”, or “msdf”.

https://github.com/Chlumsky/msdfgen

I use this, free and has been very good for me https://github.com/Chlumsky/msdfgen

Distance fields are not slow to render. They don't need a powerful gpu. Valve was already using SDF for textures in 2007 and released a paper about it. MSDF (multi channel signed distance fields) is a popular text libraries for game engine devs that uses distance fields. Distance fields are fast to render in 2D and even 3D. The problem is with everything around it. Lighting, shadows, shading will all require specialized tooling and likely a specialized engine for very little benefit (imo).

Signed distance fields only work well for relatively simple characters.

If you have highly detailed characters like Chinese or emojis, you need larger resolution to faithfully represent every detail. One way to get around excessive memory requirements is to store the characters in their default vector forms and only render the required characters on demand, but then you might as well render them at the required pixel resolution and do away with the additional complexity of SDF rendering.

SDFs are still useful though if you have to render text at many different resolutions, for example on signs in computer games, as seen in the original paper https://steamcdn-a.akamaihd.net/apps/valve/2007/SIGGRAPH2007...

In the past, SDFs also had problems with sharp corners, which has been solved in https://github.com/Chlumsky/msdfgen

MSDFGen looks pretty sweet. https://github.com/Chlumsky/msdfgen