Improving Quality of SVG Images

Hi Everyone,

Finally had time to test out my LaserCube 7.5 W Ultra in the venue that we’ll be using. Happy to report that it should be sufficient for our intended use case; creating a virtual backdrop for a ballet performance.

However, as I consider entering the design phase for this project, I’m running into some issues creating suitable SVG images that translate well for our purposes. Those running from LaserOS seem quite a bit more suited naturally to the display, whereas the test images that we’ve created tend to flicker and are generally less stable. What are some principles for efficient design that would translate an image well to a laser format? It doesn’t seem to be a problem with complexity of the image, because I feel as though I’m encountering far more complex images than we’ve attempted that nonetheless look quite a bit better for all that.

Any advice?

Thanks!

For the record, images have been created using Inkscape, Illustrator, etc.

When I search for pre-rendered SVG images for illustrative purposes, I am only able to find images using for engraving and laser cutting. Is there a repository for an of these types of images on the Web?

If using Inkscape, dont auto trace images. Very often, the auto traces the outside snd inside edges of line so it effectively doubles your point count. Manually tracing using the Pen tool is how you cam minimize point count snd make better graphics. All the images in LaserOS have been optimized for laser.

Food for thought.

Dustin L. Derry
CEO, LSO
(312) 498 - 2317 Mobile
dustin@stonewolf.com

As a general rule, for any laser device, don’t create graphics with more than 500 points. It doesn’t matter what laser software you use.

Hi @David_Eames,

Forgive my delayed response - it’s a very long and complicated subject and realistically I probably need to make a whole video about it!

In the meantime, the advice above is good. Here’s a quick fly-past of some top-level guidance.

1. Be realistic about what a single laser can achieve

Take a look at the Liberation logo clip for an example. Notice that it’s actually made from very few lines, and most of them are curved.

Now compare that with writing “Liberation” using a normal text node. It will likely become a much longer frame and start to flicker. That’s because each letter isn’t a single line any more - the whole outline of every letter is traced all the way around.

This is essentially what @StonewolfActual was referring to about tracing.

2. Be very disciplined about the lines in your SVG

If you trace bitmaps you will usually get outlines around every shape. If two shapes touch each other, you will end up with two laser lines along the join - one from each shape.

It’s important to make sure that text, circles, rounded rectangles etc are converted to shapes before exporting. Manual line drawing is often the best approach.

You might also find this section of the manual helpful:

3. Laser tuning matters more with graphics

When displaying graphics, scanner tuning becomes more critical.

If you see tails, you will need to adjust the scanner sync. If you can’t remove them completely, you can either adjust the render settings (be careful though - incorrect settings can damage scanners) - in particular adding more blank points at the start or end can reduce the tails, or increase the point rate in the laser’s Advanced settings to give you more resolution to finely tune out the tails. See Laser output settings panel | Liberation User Manual

In recent versions of Liberation this increases the temporal resolution of the scanners.

Note that since 1.0.3 the point rate setting does not change the actual speed of the scanners. I realise this can be confusing at first, but remember that Liberation generates the point stream in real time, so it can compensate for the actual scanner motion.

4. If you’d rather avoid deep tuning

If you’d prefer not to adjust advanced settings, just keep graphics simple and experiment with the scanner presets (for example Fast, Graphics, etc) or slightly adjust the scan speed.

5. More lasers is often the better solution

It’s usually better to add more lasers than to push scanners harder.

For example, four 30kpps lasers will often look brighter and cleaner than two 60kpps lasers, and may even cost less overall. You can use the canvas system in Liberation to spread content across multiple lasers.

Finally, it’s worth noting that traditional guidelines like “500 points per frame” don’t really apply any more, because Liberation doesn’t work with fixed point frames in the traditional sense.

I hope this helps give a rough overview. I still plan to record a video about this soon - it’s a complex topic and much easier to demonstrate than describe, and it requires some practice and experimentation to get it right.

All the best,

Seb

I meant to post about this.

I tried to convert some simple clip arty images into SVGs to load into Liberation and they would not load. I actually sourced them as SVG files but they did not work, I opened them in Inkscape, exported them again, still no dice.

What is the step by step process for converting images to SVGs for liberation?

What settings should be used as there are many options?

Can you send me the svgs so I can take a look?

The trick is making sure your SVG shapes have a Stroke of at least 1 set. Its helpful if the stroke is not Black.(but Liberation has a tool to compensate for this).

Thanks everyone; very helpful tips.

I feel as though I’m trying to create fairly simple geometric shapes. I’ve included an image below as an example. I do understand that there are a lot of lines involved, which is probably what is skewing the design. What, if anything, could I do to simplify the image while retaining its integrity?

Hi David,

That looks like it’s come from an AI-generated bitmap, so the next step would usually be to manually trace the lines in something like Inkscape

(At the moment there aren’t really any AI tools that generate clean, laser-friendly vectors. Even the newer ones in Illustrator tend to produce results that still need quite a bit of cleanup. I’m following developments though :blush: )

You can try auto-tracing this, but it probably won’t get you where you want to be. What we’re aiming for is for each visible line to be a single stroke, rather than a filled shape with outlines going all the way around. There are ways to “skeletonise” shapes afterwards, but they’re not especially reliable.

Even if you do trace it cleanly as single strokes, there’s still quite a lot going on for a single laser to render nicely. You might find it works better if you simplify it a bit (for example, fewer rectangles around the edge), or split it across multiple lasers using the canvas system.

Hope this helps,

Seb

Hii Seb!

Thanks for the input. This image is from a bitmap image; I’ve traced the image in Inkscape to get rid of messy conversion artifacts; the result, while closer to what I’m looking form, is still uneven and too busy. I’ve got one laser to work with, so across the canvas will not be an option for me. Before simplifying the image itself, I’m looking to exhaust my options as far as cleaning up the laser pathing. Is there anything available that can work by correcting and/or reducing points?

Sorry for the extra topic thread; I’ll keep using this thread for this topic.

Thanks!

David

It’s difficult to say without seeing where you’re at, but feel free to upload the SVG and I’ll take a look.

But if you made a single stroke trace of that entire image, it’s gonna be too much for one laser. Especially with all the tight corners!

Seb